Professional Documents
Culture Documents
Encyclopedia of Physical Science and Technology EN002C-64 May 19, 2001 20:39
Table of Contents
(Subject Area: Analytical Chemistry)
Pages in the
Article Authors
Encyclopedia
Microwave molecular
Robert L. Cook Pages 799-852
spectroscopy
Nuclear Magnetic
Bernard C. Gerstein Pages 701-720
Resonance (NMR)
Organic Chemistry, Raphael Ikan and Bernard
Pages 459-496
Compound Detection Crammer
Photoacoustic Konka Veeranjaneyulu and
Pages 1-13
Spectroscopy Roger M. Leblanc
Photoelectron G. Hohlneicher and A.
Pages 57-90
Spectroscopy Gildenpfennig
R. P. Van Duyne and C. L.
Raman Spectroscopy Pages 845-866
Haynes
Scanning Probe
C. Daniel Frisbie Pages 469-484
Microscopy
Sonoluminescence and
Kenneth S. Suslick Pages 363-376
Sonochemistry
Spectroscopy in
Michael B. Eyring Pages 637-643
Forensic Science
Thermal Analysis David Dollimore Pages 591-612
Analytical Chemistry
Ulrich J. Krull
Michael Thompson
University of Toronto
I. Classical Methods
II. Instrumental Methods
III. Computers in Analytical Chemistry
IV. Future Perspectives
543
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
elements in alloy steels to part-per-billion levels of drugs ganic species lies in their chemical and physical behavior.
in biological tissue, are handled by the analyst. The field For example, reagents are used to yield distinct chemical
is founded on the conversion of a measured physical prop- effects such as the production of colored solutions or pre-
erty of the species being examined to a usable signal. It cipitates, the generation of easily observed gases, and the
is generally divided into two categories, classical and in- dissolution of previously insoluble substances. In a typical
strumental, on the basis of its historical development. The analysis, the identification of several species is required,
overall strategy is to prepare a sample correctly, choose and in this circumstance, it is desirable to employ selective
a particular method of analysis, and report the results reactions for each component of the system.
in a meaningful format, which may include a statistical This type of chemistry is incorporated into a strategy
evaluation. that involves the separation of the original mixture into
several parts in order to avoid the buildup of a highly
complex array with a specific sample due to the addition
of a number of reagents. Each part is then subjected to
I. CLASSICAL METHODS an analysis of a small number of species. In summary,
the analysis involves a set of sequenced separations and
A. Semimicro Qualitative Analysis
identifications.
A complete system of qualitative analysis is usually re- The strategy for the separation of cations involves their
stricted to the detection of a complex array of inorganic division into a set of groups by treatment with a particular
cations and anions. In principle, schemes are developed to selective reagent. After solution of the unknown in water,
include less common species such as rhenium and tellurate a reagent is employed to cause all cations of the group to
anions, but more often than not attention is concentrated on precipitate, with all other ions remaining in solution. A
more common cations such as Na+ and Cu2+ and oxy an- typical overall scheme (simplified) is depicted in Fig. 1.
−
ions such as SO2−
4 and NO3 . The basis for identifying inor-
After separation, the precipitate is reexamined for specific
FIGURE 1 Simplified strategy for separation of cations into groups for qualitative analysis.
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
cations. As one example let us consider the silver group. and therefore the error, on reagent precipitation. The fa-
Treatment of the white precipitate of Hg2 Cl2 , AgCl, and cility with which a precipitate is removed from solution
PbCl2 with hot water results in solubilization of the lead is related to the particle size of the solid phase, which in
compound to yield a colorless solution. The presence of turn is governed by the conditions that exist at the time
Pb2+ is confirmed by the addition of K2 CrO4 , which gives of formation of the precipitate. Thought to be important
a yellow precipitate of PbCrO4 . Addition of NH4 OH to are the solubility of the precipitate, temperature, reactant
Hg2 Cl2 and AgCl results in either a gray mixture of mer- concentration, and mixing phenomena. These parameters
cury and HgNH2 Cl, confirming Hg2+ 2 , and/or colorless control the supersaturation that exists in solution at a par-
solutions of Ag(NH3 )+ 2 Cl −
. Acidification of the latter re- ticular time. Although the usual aim is to achieve an eas-
precipitates AgCl, identifying the presence of Ag+ . ily manipulated crystalline precipitate, colloidal particles
The detection of anions in the original sample does not (10−6 to 10−4 mm in size) are sometimes obtained. In
proceed in the same way in that the material is usually sub- these cases, the individual particles must be coagulated
jected to a series of preliminary tests. Moreover, the cation by an experimental procedure such as heating, stirring,
analysis described above can often be assessed to infer the and the addition of auxiliary electrolyte. The mechanism
presence of certain anions. A sample solution is separately of this process lies in the reduction of electrical repulsive
treated with AgNO3 and BaCl2 , and the various precipi- forces on the particles. Purer, more dense precipitates can
tates of silver and barium are used to verify the presence of often be produced by precipitation from homogeneous so-
anionic components. This procedure is then followed by lution, where the reagent is generated in solution. Finally,
the addition of concentrated cold H2 SO4 to the unknown after filtration a gravimetric precipitate is heated until its
solid, which results in the liberation of characteristic gases weight becomes constant. A wide range of temperatures
(e.g., I− gives the odor of H2 S and violet fumes of I2 ), and (110–1200◦ C) are used for this purpose, the most impor-
by a series of confirmatory tests for each anion. tant requirement being weight constancy and unequivocal
On a practical level the whole analysis is performed knowledge of the composition of the precipitate at a par-
at the semimicro level. This means that sample sizes are ticular temperature.
in the range of 10 mg and solutions are of the order of Some examples of inorganic and organic precipitation
1–2 ml in volume. Special equipment is required to han- agents for inorganic analytes are given in Table I.
dle low volumes of reagent solutions and to avoid serious There are also methods available for the precipitation
losses in separative and transfer procedures. The sepa- of organic compounds through the reaction of a partic-
ration of precipitates is performed by centrifugation and ular functional group. For example, compounds with
decantation, and heating is achieved in specially designed carbonyl functional groups can be precipitated with
test tube heating blocks. 2,4-dinitrophenylhydrazine according to the following
reaction:
B. Gravimetric Analysis RCHO + H2 NNHC6 H3 (NO2 )2 →
Here, quantitative analysis is based on the measurement of R—CH NNHC6 H3 (NO2 )2 ↓ + H2 O.
the weight of a substance of precisely known composition
that is chemically related to the analyte. Most often the In summary, the gravimetric method does not require
unknown is precipitated from solution by a reagent and, calibration, as is the case with many other analytical
after separation and drying, is weighed. Less frequently
the species being determined is volatilized, and the weight TABLE I Typical Inorganic and Organic Precipitat-
of the condensed gas or residual solid serves to complete ing Agents
the analysis. The precipitate must be insoluble in water Element precipitated
(or other solvent) in order to minimize obvious losses, be Agent (weighed form)
easily filtered and washed, and be stable after drying or ig-
NH3 Al (Al2 O3 ), Fe (Fe2 O3 )
nition procedures. With respect to the first requirement, an
H2 S Zn (ZnO), Ge (GeO2 )
important quantitative parameter is the solubility product
H2 SO4 Pb (PbSO4 ), Ba (BaSO4 )
of the compound produced from the analyte. For exam-
HCl Ag (AgCl)
ple, for the precipitation of SO2− 2+
4 with Ba , the solubility
(NH4 )2 CO3 Bi (Bi2 O3 )
product (K sp ) for BaSO4 is defined by
BaCl2 SO2−
4 (BaSO4 )
K sp = [Ba2+ ] SO2−4 = 1.3 × 10−10 MgCl2 , NH4 Cl PO3−
4 (Mg2 P2 O7 )
8-Hydroxyquinoline (HQ) Al (AlQ3 )
for a saturated solution. Clearly, this value can be used
Dimethylglyoxime (DMG) Ni (NiDMG2 )
to compute the loss of analyte (remaining in solution),
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
methods (see later). Accordingly, it is still frequently used ion, bases are standardized against potassium hydrogen
as a “standardizing” technique for instrumental methods. phthalate (KHC8 H4 O4 ). The end point in a strong acid–
Gravimetry, however, can be rather time consuming, es- strong base neutralization titration is usually found from
pecially if a large number of samples are involved. the in situ behavior of an added indicator, which is gener-
ally a weak organic acid or base that undergoes chemical
changes exhibiting different colors. For example, we can
C. Titrimetric (Volumetric) Analysis
write for the acid-type indicator HIn
In titrimetric analysis, which is often termed volumetric
H2 O + HIn = H3 O+ + In− ,
analysis, we obtain the volume of a standard reagent re-
quired to consume an analyte completely. On a practical (color 1 in (color 2 in
basis a standard solution of reagent, the concentration of acid solution) basic solution)
which is known accurately, is added by a buret until it is
Thus, it is very important to know the range of hydrogen
decided that the analyte is just used up. This condition is
ion concentration (i.e., pH) in which a change from color
usually called the equivalence point. Since it is difficult
1 to color 2 can be observed. Generally, this occurs within
to observe this point experimentally, it is usually approx-
approximately ±1 pH unit of the pK a of the indicator.
imated by the distinction of an end point, which is asso-
The titration of a strong acid solution by additions of
ciated with detectable physical changes at equivalence. It
a standard strong base solution from a buret would re-
is generally the case that standard solutions are calibrated
sult theoretically in the lower graph shown in Fig. 2 (the
against solutions of a primary standard that is a highly
titration curve). Accordingly, we would require an added
pure and stable reference substance.
indicator to respond to the abrupt approximately 3–10 pH
In the present description we introduce four types of vol-
change. In this case, phenolphthalein would be appropri-
umetric analysis. One involves the neutralization of acid or
ate, since its change from colorless to purple is easily
base, the second is concerned with precipitation, the third
detected by the eye. The titration curve for a weak acid
with complex formation, and the fourth with oxidation–
(with strong base) depends very much on the dissociation
reduction reactions. Before doing this we must consider
constant of the acid. Basically, additions of base set up a
several definitions. The equivalent weight of a substance is
buffer zone as in the set of curves shown in Fig. 2 (upper
generally the weight that combines with a standard amount
curve). The choice of an indicator for this type of titration
of reactant. The parameter is based entirely on the nature
is significantly more limited than for a strong acid, since
of a particular reaction and, therefore, can have no more
there is now no abrupt change in pH. Not surprisingly,
precise meaning at this stage. Titration refers to the pro-
the appropriate curves for polyprotic weak acids are even
cess of gradual addition of standard reagent, whereas the
more complex.
titer of a solution is the weight of a substance that is chem-
Neutralization methods are employed wherever inor-
ically equivalent to 1 ml of the solution. Finally, one often
ganic or organic substances possess acidic or basic groups.
sees the term normality, which expresses the number of
An important application is the conversion of elements
milliequivalents of solute contained in 1 ml of solution.
The end point in titrimetric methods is usually based
on a way of dynamically distinguishing analyte or reagent
concentration. The most widely used method is change in
color due to reagent, analyte, or indicator, but the moni-
toring of electrical potential and current, and turbidity, are
sometimes used.
1. Neutralization Titrations
In this technique the concentration of acid or base is de-
termined through the abrupt change of pH that occurs at
the titration end point. Here, the equivalent weight of the
acid or base is the weight that either contributes or reacts
with 1 mol of hydrogen ion in that specific reaction. The
standard reagents used in the titration are always strong
acids or bases, although the analyte may be a weak acid
or base. A-standard acid solution can be prepared by titra- FIGURE 2 Neutralization titration curves for 50 ml of 0.05M
tion of the acid, say HCl, against a solution containing an strong acid (lower curve) and weak acid (upper curve) with 0.1M
accurately known weight of Na2 CO3 . In a similar fash- NaOH solution.
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
in organic or biological systems to acid–base character. of Cl− and Br− is that of Mohr, who used chromate ion
For example, organic nitrogen is converted to NH+
4 in the (CrO2−4 ), the end point being specified by the appearence
Kjedahl method to be determined in a neutralization pro- of Ag2 CrO4 , which is brick red in color. The essence of
cedure as liberated NH3 . the principle here is that the solubility of silver chromate
is much greater than that of silver halide. Accordingly, the
conditions of the titration can be adjusted such that the
2. Precipitation Titrations formation of silver chromate occurs only in the vicinity of
In this method, the reagent solution, added from a buret, the end point. In the Volhard method a standard solution of
is allowed to cause precipitation with the analyte, and the thiocyanate ion (SCN− ) is titrated against Ag+ using Fe3+
end point is associated with an abrupt change in the con- as the indicator. At the initial excess of SCN− , a red color
centration of the reagent species. The equivalent weight of due to red Fe(SCN)2+ is produced. An important applica-
a participant in a precipitation titration is the weight that tion of this method is the “backtitration” procedure for the
reacts with or provides 1 g formula weight of the reacting Cl− ion. Here, an excess of standard silver nitrate solution
cation if it is univalent (0.5 for divalent, etc.). The tech- is added to the unknown Cl− solution and then titrated
nique can be best understood by the example provided by with SCN− with the Volhard indicator in place. Finally,
the long-standing silver ion procedure for the determina- adsorption indicators (Fajans method) are employed that
tion of halide anion, particularly Cl− (sometimes termed are based on the adsorption of a colored organic compound
argentometric titration). Typical titration curves for vari- to the precipitate, only in the vicinity of the end point.
ous halide anions with silver cation are shown in Fig. 3.
Notice that the concentration of the free reagent cation
3. Complexometric Titrations
present at a particular point in the titration is usually ex-
pressed as −log[Ag+ ], that is, pAg, in a fashion analogous Titrimetric methods based on the reaction of metal
to pH for hydronium ion. Before the equivalence point ions with a coordinating species (ligand) are more than
(with abrupt change of pAg), the concentration of Ag+ is 100 years old. If the ligand is attached to the metal by
determined by the solubility product of the silver halide more than one functional group, it is said to be polyden-
concerned. After this point we simply have the concentra- tate and the compound produced is called a chelate. The
tion of excess added reagent, which is not now reacting complexometric reagent that is easily the most widely used
with halide anion. To detect the end point an indicator is for the titration of unknown concentrations of metal ion
required that will respond to the particular range of abrupt is the hexadentate ligand ethylenediaminetetraacetic acid
pAg change shown in Fig. 3. A common method in the case (EDTA), first recognized by Schwarzenbach in 1945. The
structure of the molecule is
4. Oxidation–Reduction Titrations
Here, we are dealing with the reaction of an oxidizing derive
3+ + 5E
0 0
(reducing) reagent as titrant with a reducing (oxidizing) E Fe MnO−
species as unknown. The equivalent weight of a partici- E eq = 4
6
pant in this type of system is the weight that directly or
indirectly produces or consumes 1 mol of electrons. Thus, 0.0591 5 MnO− 2+
4 [Mn ]
− log .
the equivalent weight for the permanganate ion (MnO− 4)
6 5[Mn2+ ] MnO− + 8
4 [H ]
in its oxidation of oxalate anion in the reaction
Thus,
5C2 O2−
4 + 2MnO−
4 + 16H + 0
E Fe3+ + 5E
0
MnO− 0.0591 1
E eq = 4
− log + 8 .
→ 10CO2 + 2Mn2+ + 8H2 O 6 6 [H ]
is the molecular weight of MnO−
divided by 5.
4
In much the same manner as for the other types of
A great variety of both oxidizing and reducing agents titration described above, it is the electrode potential dur-
have been employed for this type of titration, and some ing the procedure that exhibits an abrupt change on ad-
common compounds are given in Table II. The equivalence dition of reagent. Accordingly, we require indicators that
point of oxidation–reduction titrations can be computed change color during this change (i.e., that show oxidation–
from a knowledge of solution concentrations and electri- reduction behavior themselves). Two examples are 1, 10-
cal potentials. For example, in the titration of ferrous ion phenanthroline–iron (II) complex, which changes from
against KMnO4 according to the following reaction pale blue to red at an electrical potential of +1.11 V (in
1M H2 SO4 ), and diphenylaminesulfonic acid, which
5Fe2+ + MnO− +
4 + 8H → 5Fe
3+
+ Mn2+ + 4H2 O, changes from colorless to violet at 0.85 V.
we can write Nernst electrode potentials for each
oxidation–reduction system:
II. INSTRUMENTAL METHODS
[Fe2+ ]
E = E Fe
0
3+ − 0.0591 log 3+ A. Absorption of Electromagnetic Radiation
[Fe ]
0.0591 [Mn2+ ] Matter interacts with incident electromagnetic radiation
E = E MnO
0
− − log , by the three distinct processes of transmission, scattering,
4 5 MnO−
4 [H ]
+ 8
or absorption. The nature of any interaction is a function of
At the equivalence point we know that these two poten- the properties of the radiation, such as energy, phase, po-
tials, now called E eq , are equal; therefore, on adding we larization, and the chemical properties of the matter under
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
FIGURE 4 Analytical potential for absorption spectroscopy across the electromagnetic spectrum.
investigation. Chemical evaluation is derived from obser- concentration of the analyte, and A represents a value for
vation of the extent of interaction by measurement of the absorption.
energy and intensity of transmitted, scattered, absorbed,
or latently released radiation.
Absorption is said to occur when radiation passes 1. Atomic Absorption
through matter and interacts with the ions, atoms, or The absorption of radiation by a sample of atomic parti-
molecules constituting the sample in such a way that they cles, created by vaporizing the sample, represents a rel-
gain energy and move from a quantized low-energy state atively simple spectral situation that has great practical
to a higher energy state. Analytically, this phenomenon is value for elemental identification and concentration de-
observed as a reduction of the intensity of radiation after termination. The absorption spectrum observed with poly-
passage through a sample of matter. Two distinct param- chromatic light contains only a few areas of reduced in-
eters can be investigated. The reduction of intensity is a tensity occurring at very well-defined frequencies due to
function of the probability of interaction of the radiation the small number of energy states available to the atoms.
with appropriate analyte species and indicates concentra- The natural width of such absorption lines can be less than
tion. Also, the energy of the absorbed radiation indicates 10−4 nm, but broadening often occurs due to collisional
the quantized energy levels in which energy is deposited and Doppler effects in the sample matrix. The quantized
(Fig. 4) and therefore assists in species identification. The energy transitions responsible for atomic absorption are
Beer–Lambert law generally applies to all absorption pro- of electronic origin, indicating that electromagnetic spec-
cesses at low concentrations when monochromatic radia- trum energies from X-rays to ultraviolet–visible rediation
tion is used. It is conventionally written are sufficient to observe this process.
P0 The technique known as atomic absorption spec-
log = εbc = A, troscopy is of particular analytical importance for the de-
P termination of metals due to its sensitivity and potential
where P0 is the initial power of an incident beam of radi- for selectivity by virtue of the narrow atomic absorption
ation, P is the final power (decreased due to absorption), lines. A schematic representation of the spectrophotome-
ε represents a value characteristic of the extent of absorp- ter is shown in Fig. 5. To take advantage of characteristic
tion expected for a certain species at a defined wavelength selectivity, a special radiation source must be provided
in a defined matrix, b is the sample path length, c is the to produce extremely monochromatic radiation with a
linewidth less than the absorption bandwidth of the atomic TABLE III Atomic Absorption Detection Limits for Common
sample. In most cases the same element being analyzed is Analytes
also used in the source lamp to provide appropriate emis- Limit of detection (ppm)
Analytical
sion line spectra by stimulation with a plasma discharge. wavelength Flame Nonflame
Common sources are hollow cathode lamps, electrode- Element (nm) absorptiona absorption
less discharge lamps, and gaseous discharge lamps. The
source is often modulated and coupled to a lock-in am- Aluminum 396.2 0.03 0.00001
plifier in order to determine the relative absorbance of Calcium 422.7 0.001 0.00005
the atomic sample as compared with sample emission at Cadmium 326.1 0.0005 0.000003
the same wavelength. Other methods of background cor- Iron 372.0 0.003 0.00002
rection involve monitoring of a nonabsorbed radiation by Lithium 670.8 0.0005 0.0003
the use of distinct spectral lines, continuous sources, or Magnesium 285.2 0.0001 0.000004
the Zeeman effect (separation of degenerate energy states Potassium 766.5 0.005 0.0009
of an atom by application of a powerful magnetic field; a Data for acetylene–air flame.
provides closely spaced energy levels which are sensi-
tive to polarized light). The sample itself is usually pro-
2. Molecular Absorption
duced by volatilization of solutions in a high-temperature
flame or solids and in a graphite furnance. Flame methods The absorption spectrum commonly produced by
generally provide relative errors of approximately 1–2%, molecules is significantly more complex than that
whereas the equivalent furnace methods provide only 5– produced by atoms due to the large number of energy
10%, but sensitivities up to 1000 times greater than those states available for energy deposition. The total energy of
observed for flames. Chemical interferences commonly a molecule that has no translational velocity is given by
originate from the atmosphere supporting the atomic pop-
E total = E electronic + E vibrational + E rotational ,
ulation and from other sample matrix components. The
formation of oxides in flames causes the reduction of the where the electronic levels due to electrons in bonding and
atomic population and can be controlled by the employ- nonbonding orbitals are related to numerous interatomic
ment of appropriate fuel–oxidant ratios and by sampling vibrations and molecular rotations about a center of grav-
appropriate portions of the flame. This problem is avoided ity. The large energy difference of the three energy types
in the furnace by the use of a continuous flowing inert implies that each can be individually studied with radiation
gas atmosphere. Poor volatilization and atomization often from distinctly different portions of the electromagnetic
occur when an analyte binds with anions in the sample spectrum.
matrix. This is overcome by the use of high concentra-
tions of cationic releasing agents, which preferentially Optical polarization. Since the early 1800s it has
couple with the interferent, or by the use of protective been recognized that certain molecules have the capacity
agents such as chelating ligands, which form stable but to rotate plane-polarized light, which has since led to the
volatile species with the analyte of interest. Spectral in- development of a number of techniques suitable for quali-
terferences can also occur, though direct overlap of inter- tative structural determination and quantitative concentra-
ferent and analyte atomic absorption bands is rare. These tion analysis. All these techniques are based on the fact that
interferences include molecular band absorption such as electromagnetic radiation has wavelike properties and can
that experienced from the formation of combustion prod- be represented as a combination of electric vectors. The
ucts, which can often be eliminated by the use of higher electric vector can interact with the electrons of matter
temperatures, and ionization of the atomic population due in an absorption–reemission process taking place over a
to excess temperatures. Since many of the chemical pro- time period of 10−14 to 10−15 sec. Even though this pro-
cesses occurring in certain localized areas of the atomic cess does not change the energy of the radiation, a slowing
atmosphere are approximately in equilibrium, such ion- does occur. If a monochromatic beam of radiation passes
ization can be suppressed by the addition of a radiation through an anisotropic solid sample, the electric vectors
buffer, which selectively ionizes in contrast to the ana- that encounter greater particle density will be slowed more
lyte of interest and therefore increases the probability of than the vectors passing through less dense areas. Plane-
ion–electron recombination to form the required atomic polarized light is represented as the resultant of two inter-
population. Table III presents some of the analytes com- fering electric vectors such that the resultant always lies in
monly investigated by the techniques of atomic absorption a single plane. In the case of plane-polarized monochro-
spectroscopy. matic light with two coherent perpendicular electric vector
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
components, the components can pass through the sample thickness of ultrathin films. Circularly or elliptically polar-
and be unequally retarded so that the recombination after ized light is reflected from a surface on which the sample
exiting the sample results in constructive interference with is coated. The reflection through the sample alters the po-
a phase difference from that originally observed. This re- larization, and the degree of alteration provides thickness
sults in a rotation of the plane of polarization and propaga- data to 0.1-nm precision. These instruments have wide ap-
tion of circularly or elliptically polarized light dependent, plicability in the semiconductor industry and as tools for
respectively, on phase separations of multiplies of 0, 14 , 12 , the measurement of organic films of monolayer dimen-
or combinations of these numbers. sions.
Polarimetry is a technique suitable for structural de-
termination and quantitative analysis and simply studies Ultraviolet–visible. This form of electromagnetic
the direction and degree of the rotation of plane-polarized radiation is of sufficient energy to cause electronic ab-
monochromatic radiation of transparent substances. In- sorption transitions from the ground state to an excited
struments generally make use of vapor lamps, which iso- state within molecules. In contrast to sharp atomic elec-
late the sodium D line or mercury line at 546 nm. The light tronic absorptions, molecular spectra show broad bands
is then processed by a Nicol polarizer prism assembly, covering tens of nanometers due to the availability of a
passed through a half-shadow Lippich prism (to assist de- multitude of distinct energy states caused by the overlap
tection of the optical minimum intensity), passed through of the vibrational and rotational energy sublevels with the
a thermostated transparent sample holder, passed through primary electronic level. Only high-resolution spectrom-
a second Nicol prism known as the analyzer, and directed eters can clearly resolve these energy levels.
to an eyepiece or photoelectric detector. Rotation of the The design of a typical double-beam instrument that al-
analyzer allows matching of the split-field bright image. lows for relative background correction is schematically
The instrument is calibrated by analyzing the rotation of shown in Fig. 6 and consists of a radiant source, such as
a standard solution of known concentration and rotatory a hydrogen–deuterium lamp for ultraviolet and a tung-
power and provides precision to 0.001◦ when coupled to sten lamp for visible radiations; a filter, prism, or grating
photoelectric detectors. monochromator for wavelength selection; sample and ref-
Other techniques that find widespread application are erence cells; and a photocathode, usually associated with
optical rotatory dispersion and circular dichroism. These a photomultiplier for detection.
are closely related methods that rely on the wavelength The most important types of electronic transition in-
dependence of refractive index and molar absorptivity, re- volve the following:
spectively, toward the direction (d or l ) of circularly po-
larized radiation. Optical rotatory dispersion experiments 1. n, σ , and π electrons, covalent compounds
provide a measurement of molecular rotations as a func- 2. Charge-transfer electrons
tion of wavelength. Circular dichroism results are pre- 3. Electrons, ligand field strength
sented in a plot of wavelength vs a value [θ ], and [θ ] is 4. f electrons, lanthanide and actinide ions
a corrected value of θ, the molecular ellipticity measured
as the singular difference between the emergent ellipti- Of particular importance are those transitions involving
cal beam and the original incident plane of polarization. bonding electrons, since particular absorption energies are
These techniques have been important tools in the elucida- characteristic of the bonding of certain functional groups,
tion of structural information about complex organic and known as chromophores, providing chemical identifica-
biological systems and metal–ligand complexes. tion and concentration information. A summary of approx-
The use of elliptically polarized light has also found imate electronic energy transitions for bonding and non-
application in the determination of refractive index and bonding electrons is given in Fig. 7. This can be correlated
means of a metallic waveguide transmits the electromag- TABLE VI Common Nuclear Magnetic Resonance Nuclei of
netic radiation to a quartz tube sample holder positioned Quantum Spin 12
between the poles of a permanent magnet. The experi- NMR frequency Natural Relative sensitivity
ment consists of varying the magnetic field strength by Isotope (MHz) at 20 kG abundance (%) per nucleus
means of secondary coils to induce resonance. The result- 1H 85.2 99.98 1.000
ing absorption signal is usually processed and outputed 2H 90.8 0.02 1.21
as a first-derivative curve to increase sensitivity and res- 13 C 21.4 1.11 1.59 × 10−2
olution. Though the technique does not have wide ap- 15 N 8.6 0.37 1.04 × 10−3
plicability, it continues to play an important role in the 19 F 80.1 100 0.834
probing of biological systems for order, viscosity, reactiv- 31 P 34.5 100 6.64 × 10−2
ity with spin-label reagents (since absorption is influenced 195 Pt 18.3 33.7 9.94 × 10−3
by chemical environment), and investigation of chemical
reactions, which include free-radical intermediates.
due to electrons that circulate in the molecule contain- higher, the resolution of the instrument improves, and
ing the absorbing nucleus. Circulation of charge creates a 600-MHz instruments for proton studies are now avail-
magnetic field, which can reduce or enhance the applied able. A pair of secondary field coils are located paral-
field in a local area so that lel to the high-power magnet faces for the adjustment of
field strength over small ranges. A radiofrequency source
H0 = HA (1 − σ ),
consisting of oscillator coils is mounted perpendicular to
where H0 is the resultant field and is equal to the origi- the magnetic field direction and provides plane-polarized
nal applied field HA corrected for the shielding parameter radiation. The signal produced by nuclei in resonance
σ . A compound is chosen as a standard for calibration is detected by another coil surrounding the sample and
if its shielding value is larger than the values commonly mounted perpendicularly to the radiofrequency source
associated with the nuclei of interest (e.g., tetramethylsi- coil. Samples are usually in liquid form and are placed
lane for proton magnetic resonance studies). Correlation in a narrow glass tube, which is rapidly spun to elimi-
of the chemical shift with structure can provide useful nate the effects of field inhomogeneities. Analysis of solid
group identification. Chemical shift values for protons are samples is possible when a special arrangement is avail-
the most common and are often referred to on a δ or τ scale, able to orient and rapidly spin the sample (>2 kHz) at a
“magic angle” in relation to the magnetic field. This an-
Href − Hsample × 106 gle of 54.7◦ is dictated by geometrical constraints. The
δ∼
=
Href use of Fourier transform methods (see Section III.A.3)
τ = 10 − δ, has provided a means of amplifying the nuclear mag-
netic resonance signal so that many insensitive nuclei are
where Href and Hsample are the field strengths required to now routinely investigated, as shown in Table VI. Samples
produce tetramethylsilane and sample resonance, respec- are irradiated with a broad radiofrequency spectrum for a
tively. A summary of some common proton chemical short period of time. After this pulse of energy is applied,
shift values is given in Table VII. Additional information the excited nuclei relax to the lower energy states, pro-
can be garnered from the fine structure of absorption viding a time-based free induction decay spectrum. This
bands, which is known as spin–spin splitting. This occurs spectrum represents the overlap of the different resonant
when the field about one nucleus is affected by the fields frequencies, producing a characteristic envelope of time-
from neighboring nuclei attached to an adjacent atom. dependent oscillations. The time-domain spectrum can be
The degree of splitting reported as frequency differences collected in seconds, allowing experiment replication to
and the relative areas under each separate absorption occur hundreds of times in a practical time period. These
signal can, therefore, provide quantitative information spectra can be collected and averaged by computers to
about the chemical environment. provide a tremendous signal-to-noise enhancement and
Instrumentation incorporates a high-strength, high- are finally displayed as conventional frequency-domain
quality magnet, which may be permanent, electrically in- spectra. Modern techniques now allow two-dimensional
duced, or superconducting. As the field strength becomes analysis where excitation scans across different frequency
ranges are concurrently analyzed to provide information
about coupling between nuclei.
TABLE VII Correlation of Common Proton Chemical Shifts
Properties of interest for chemical analysis are the so- This technique allows a multitude of species to be de-
called isomer shifts, which represent the displacement of termined concurrently since characteristic lines originate
resonance from zero Doppler velocity, and two types of from all suitable species simultaneously. A schematic rep-
line splitting. Splitting can be induced by interaction of resentation of a simple instrument would be equivalent
the electric field gradient about the nucleus with the elec- to that shown in Fig. 5 if the hollow cathode lamp and
tric moment of the excited nucleus and also by interaction modulation system were removed. A more energetic atom
of the nuclide magnetic dipole moment with internal or source such as an inductively coupled plasma (a zone of
external magnetic fields to produce magnetic hyperfine highly energetic ionized inert gas such as argon, which
structure. Instrumentation is based on the γ -ray source, a provides a high constant operating temperature, a long
Doppler motion device, and an energy-proportional γ -ray sample residence time, and a nonoxidating amosphere) is
detector coupled to a multichannel analyzer. Chemical usually employed. Detection limits are generally in the
information extracted from Mössbauer spectra indicate range observed for atomic absorption methods, though
oxidation states, sample identity with respect to mineral the two techniques are complementary, each surpassing
structure, and quantitative sample mineral compositions. the other in the analysis of certain species.
The technique is employed most often for iron and cobalt,
though over 50 nuclides have demonstrated the Mössbauer Atomic fluorescence spectroscopy. Fluorescence
effect. refers to a process whereby absorption and reemission
of radiation are separated temporally. A pulsed source
of high intensity such as a laser, electrodeless discharge
B. Emission of Electromagnetic Radiation
lamp, gaseous discharge lamp, or specially adapted hol-
Matter can exist in a physically excited energy level and low cathode lamp at the required resonant frequency is
can relax to a lower ground state energy by releasing the used to irradiate an atomic population created usually by
difference in energy. One mechanism of such relaxation a nonflame method. Emission can occur in any direction
can occur by emission of electromagnetic radiation. Con- and is commonly observed by standard atomic absorption
tinuous broadband radiation is observed from a thermally instrumentation set at an angle of 90◦ to the high-intensity
excited continuous solid and results from the complicated source beam. This arrangement has been shown to im-
oscillations originating from the molecules and atoms of prove detection limits over those for atomic absorption or
the sample. If these molecules or atoms emit indepen- emission for up to 10 elements.
dently, then a discontinuous spectrum composed of dis-
tinct bands or lines can be observed. The structure of this X-ray fluorescence spectroscopy. This technique
discontinuous emission is related directly to the quantized is one of the most widely used for qualitative and quan-
electronic energy levels of molecules and atoms as dis- titative elemental determination for elements of atomic
cussed previously, and energy is released in a reversal of number greater than 8. A beam of X-rays is directed from
the absorption process. Though this can be strictly true a source such as a Coolidge tube or radioactive substance
for atomic samples, often radiationless decay either elim- to a sample. Figure 4 indicates that such radiation is suf-
inates or precedes radiative emission from molecular sam- ficient for ionization of the core electronic level in atoms
ples and provides interesting chemical information. of the sample. The process of absorption of X-rays can be
used to provide structural information about atomic po-
sitions in three-dimensional space by a technique known
1. Atomic Emission
as Extended X-ray Absorption Fine Structure (EXAFS).
Emission spectroscopy. Atomic or ionic popula- Relaxation of such excited elements occurs by the shifting
tions can be prepared by the use of high-energy sources of an electron from a higher energy level to fill the elec-
such as flames, electrical discharges, and plasmas. If the tronic vacancy. The difference in energy between the two
energy in these sources is sufficient, electronic excitation electronic levels is released as quantized fluorescence in
occurs by collision processes, and relaxation can take the the X-ray region of the electromagnetic spectrum. A large
form of discrete narrow line emission. Since the wave- number of different X-ray energies can be emitted from a
lengths of such emissions reflect the energy differences single sample simultaneously, necessitating the employ-
in the quantized electronic energy level distributions, the ment of wavelength- and/or energy-dispersive devices in
emission wavelengths are characteristic of the excited el- many cases. Wavelength dispersion is accomplished in a
ement and can be used for identification purposes. The manner analogous to that for grating monochromators,
intensity of emitted lines can be used for quantitative but in this case a collimated beam of fluorescent X-rays
analysis by comparison with calibration standard signals is allowed to impinge on a solid crystal of well-defined
recorded on film or collected by photoelectric detectors. lattice spacing d (e.g., lithium fluoride, sodium chloride,
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
FIGURE 10 Schematic representation of an X-ray fluorescence spectrometer employing both wavelength dispersion
and energy dispersion.
topaz) mounted on a goniometer to satisfy Bragg’s nant effect between an incoming photon and the electrons
law, of a chromophore, elevating the valence electronic con-
figuration to a higher energy state. Fluorescence is said to
nλ = 2d sin θ,
occur when the system relaxes to a lower energy state by
where n is an integer, λ is the wavelength, and θ is the the release of radiation within 10−8 sec after absorption
angle at which constructive interference occurs. Energy- in a process that ceases in less than 10−6 sec. Resonance
dispersive systems (see Section II.B.3.b) employing pulse fluorescence occurs when the absorbed radiation is reemit-
height analysis can be used in conjunction with wave- ted at the same wavelength. Polyatomic molecules often
length dispersion to distinguish between two or more en- reemit at wavelengths longer than those absorbed, produc-
ergies of photons appearing at one angle due to differ- ing a spectral feature known as a Stokes shift. This is due
ent values of n, or a multichannel analyzer for discrete to nonradiative energy losses by molecular vibration in the
energy measurement can be independently employed. A electronically excited state. The closely related processes
schematic representation of a combined wavelength- and of fluorescence and phosphorescence are summarized in
energy-dispersive system is shown in Fig. 10. Fig. 11. Phosphorescence occurs from a forbidden excited
triplet state after inter-system crossing and may take a pe-
riod of seconds. This phenomenon is often of very weak
2. Molecular Emission
intensity and is not as common as fluorescence, thereby
Fluorescence spectroscopy. Molecular absorp- limiting its analytical potential. Competition between non-
tion of ultraviolet or visible radiation results from a reso- radiative and fluorescent energy loss occurs, reducing the
Stationary phases can be solid surfaces or surfaces rate at which equilibrium is attained include the volume
coated with liquids, while mobile phases can be liquids or and average thickness of the liquid stationary phase and
gases. Standard forms that stationary phases take are often diffusion coefficients and relative equilibrium concentra-
very fine granular powders spread evenly on planar glass tion ratios of the mobile and stationary phases.
or polymer supports (e.g., thin-layer chromatography) or A feature of fundamental interest for any chromato-
packed in columns. Also common are porous polymeric graphic work is the resolution or separation capability of
solids usually containing a microcoating of water (liquid the technique. One equation useful for quantitative mea-
stationary phase) or capillary tubes whose inside walls are surement of resolution R can be written
coated with a liquid stationary phase.
The theory of chromatography has been reasonably well R = 2 Z /(W1 + W2 ),
established with a “kinetic” or “rate” theory that describes
where W1 and W2 represent the base width of two different
the broadening of the bands or zones of separated com-
peaks in a separation and Z represents the difference in
ponents on a stationary-phase bed, describes their time
elution of the maxima of the two peaks.
of appearance at any particular point, and provides de-
tails of the separation power or resolution of the particular
system employed. Numerous equations describing zone 2. Gas Chromatography
broadening have been proposed, the simplest being the
general form of the van Deemter equation as derived for This technique is one of the cornerstones of chromato-
gas–liquid chromatography, graphic analysis, being suitable for the separation of very
complex organic and inorganic gas samples. The mobile
H = A + B/V + Cv, phase is a carrier gas chosen to maximize separation ef-
where A, B, C represent three kinetically controlled pro- ficiency as per the van Deemter equation, while concur-
cesses known as eddy diffusion, longitudinal diffusion, rently maximizing the sensitivity of the detector. The sta-
and nonequilibrium mass transfer, respectively. The term tionary phase is either a solid or a viscous organic liquid
v represents the mobile-phase flow rate, and H is defined coated on a solid support (hundreds available) or an open
as the height equivalent of a theoretical plate. This ter- tubular capillary column (for very high efficiency separa-
minology derives from the plate theory of distillations, tions).
where a distillation tower can be divided into a number of A schematic representation of a gas chromatograph is
equilibria steps or plates. In the chromatographic sense, shown in Fig. 13, and it indicates that the sample is first
the height equivalent of such a step represents simply the vaporized by injection into a heated port, then passed
length L of the separation bed divided by the number of through a heated separation column, and finally detected.
steps in the separation bed N (determined from efficiency The small sample size that can be passed by capillary
of separation): columns usually necessitates supplementation of special-
ized splitters at the injection port. A splitter accurately and
H = L/N , reproducibly reduces the volume of sample that enters the
The three kinetically controlled factors responsible for column from standard sample injections done by syringe.
zone broadening can be readily analyzed to improve sep- Numerous detection devices exist, and the most common
aration efficiency, which is equivalent to minimizing the are summarized in Table VIII. Separation times of vari-
value of H . Eddy diffusion refers to the torturous path ous species eluting from the column are often controlled
that sample species must take through a packed particle by varying column temperature by a technique known as
stationary-phase bed. The distance traveled by different temperature programming.
molecules as they migrate across the bed must vary since
different paths are highly probable. This results in zone
broadening and is critically dependent on average particle
diameter and size distribution. Longitudinal broadening
refers to the natural diffusion of molecules from a concen-
trated band toward areas of lower concentration on either
side. The most important parameters are those that con-
trol molecular diffusion rates in the stationary phase and
in the mobile phase. Nonequilibrium mass transfer recog-
nizes the fact the mobile phase may move too quickly to
allow concentration equilibrium to be established between FIGURE 13 Schematic representation of a single-column gas
the mobile and stationary phases. Factors that control the chromatograph.
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
TABLE VIII Summary of Common Gas Chromatographic TABLE IX Summary of Common High-Performance Liquid
Detector Characteristics Chromatographic Detector Characteristics
chromatographic detectors for sample elution analysis and the oxidized and reduced forms of any species (a ox , ared )
subsequent species collection. by the general form of the Nernst equation,
RT ared
D. Analytical Electrochemistry E = E◦ − ln ,
nF aox
Electroanalytical chemistry includes a broad range of tech-
niques that have as their focus the fact that the analyte par- where E ◦ is the standard electrode potential, R is the molar
ticipates in a galvanic or electrolytic electrochemical cell. gas constant, T is the absolute temperature, n is the charge
All techniques can be classified into one of three major associated with the redox reaction, and F is the faraday.
areas: those that measure electrical properties of the cell,
those that measure cell electrical properties as a function Indicator and reference electrodes. Complete an-
of a chemical reaction in the electrolyte, and those that alytical electrochemical cells are composed of combina-
physically collect the analyte at an electrode for further tions of an indicator and a reference electrode. Each of
analysis. these electrodes contributes half-reaction chemistry; in
combination they provide a complete redox reaction. The
indicator electrode is the analyte activity sensing element
1. Conductimetry
and develops a potential E Ind relative to the reference elec-
When an electrical potential is applied between two elec- trode, which represents a constant known potential E Ref
trodes immersed in an electrolyte solution, ions are accel- independent of the composition of the sample solution.
erated and migrate to electrodes of opposite charge. The The measured cell potential E cell can be written
rate of ion migration is a function of the applied potential
acting as a driving force countered by frictional forces and E cell = E Ind + E Ref + E Junc ,
electrophoretic and relaxation effects in solution. Conduc-
tance is represented as reciprocal resistance and can be where E Junc represents the liquid junction potential origi-
corrected for the cross-sectional area and length of the nating at the interfaces of the electrodes in the sample so-
conduction solution (specific conductance), or it can be lution due to the unequal distribution of ions as a function
stated as the specific conductance of 1 equivalent of so- of their different rates of migration across the interface.
lute (equivalent conductance). A linear relationship exists Ideally, the reference electrode potential is constant, and
between equivalent conductance and the square root of the junction potential is constant or negligible. A number
the concentration of a strong electrolyte. The measure- of reference electrodes are available, the most common
ment of solution conductance is highly sensitive, though being silver/silver chloride, calomel, or hydrogen gas sys-
not selective for any particular species. The device finds tems.
application as a detector for ion chromatographic instru- Indicator electrodes are generally employed for titration
ments, due to its inherent sensitivity. Conductance can be analysis where the cell potential varies as ions are con-
employed as a parameter to follow certain titrations such sumed or produced in a chemical reaction. At least four
as those involving neutralization, precipitation, and com- important types of indicator electrode can be identified.
plexation reactions that consume or generate ionic species. The simplest systems involve a strip of material suitable
The conductance experiment uses an ac source to avoid for the provision or acceptance of electrons, such as gold,
faradaic currents, with oscillation frequencies from 60 to platinum, or carbon, immersed in a solution containing
1000 Hz. A Wheatstone bridge assembly can be used as both the oxidized and reduced forms of some species so
the detection element if the electrochemical cell acts as that a redox potential can be measured. Another situation
one arm of the bridge, though other electronic assemblies involves the use of a metal that is reversible with respect to
are suitable for direct instrumental measurement of con- analyte ions of the metal so that the metal electrode partic-
ductance. The electrochemical cell consists of two large ipates in mass and charge balance rather than only charge
area platinized platinum electrodes placed in fixed orien- balance. A third class of electrodes uses two interfaces, as
tation with defined separation. in the case of the silver/silver chloride reference system,
where the metal is coated with a sparingly soluble salt and
is suitable for titration analysis as well as determination
2. Potentiometry
of the activity of the metal or the anion in the salt coating.
The measurement of the electromotive force between two Another class of electrodes representing a specialized se-
electrodes can provide information about the activity or ries of membrane-based devices designed to monitor the
concentration of an ion in the analyte solution. The poten- activities of certain ions selectively is described in the fol-
tial of any electrode (E) can be related to the activities of lowing two sections.
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
in porous lipophilic
0.059 n/z
E cell = const − log a1 + K 1,2 a2 , diaphragm
n Neutral carrier Neutral chelating agent in K+ , Na+ , Rb+
where the cell potential E cell is related to a constant in- liquid trapped in porous
cluding internal and external reference electrode poten- diaphragm or supported
in polymer matrix
tials, the activity a1 of the analyte, and the activity a2 of
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
and wax-impregnated graphite electrodes. Conventional TABLE XI Common Techniques for Polarographic Analysis
electrochemical cells are designed on the basis of three- Technique Detection limit (M) Resolution (mV)
electrode (working, reference, counter) potentiostatic con-
trol, which allows work in solvents of high resistance. Classical polarography 10−5 200
Polarography applies a continuous or pulsed or oscillat- Rapid-scan polarography 10−7 50
ing potential ramp to the working electrode and observes Cyclic voltammetry 10−7 50
the generation of a redox current as a function of applied Differential pulsed 10−8 50
potential. Three mass transfer processes can control the polarography
rate of analyte deposition at the surface of the working Square wave voltammetry 10−8 50
electrode: convection in the sample solution, migration of Stripping voltammetry 10−10 100
charged ions in the applied electrostatic field, and diffusion
due to a concentration gradient. Experiments are usually
conducted without stirring and in 100-fold concentration lyte diffusion coefficient in square centimeters per second,
excess of an electrochemically inactive electrolyte to elim- C is the analyte concentration in millimoles per liter, m is
inate the first two processes. When the working electrode the rate of mercury mass flow in milligrams per second,
potential reaches a threshold value, dependent on the stan- and t is the drop time in seconds. This equation demon-
dard potential for the half-reaction of interest, oxidation strates that the diffusion-limited current is directly propor-
or reduction of the analyte at the electrode surface be- tional to analyte concentration.
gins. Concentration diffusion begins from bulk solution Numerous procedures for applying electric potential
and eventually reaches a steady state, indicated experi- have been developed to increase the sensitivity of polaro-
mentally as increased but constant redox current (Fig. 17). graphic methods. The most sensitive of these procedures
The potential at the midpoint of the polarographic step is makes use of application of short (millisecond) voltage
known as the half-wave potential E 1/2 and is related to the pulses and sample the current which flows as a func-
applied potential E appl for the half-reaction, the measured tion of the pulse regime. These procedures concentrate
current i, and the diffusion-controlled current id as on maximizing the desired faradaic current while mini-
RT i mizing background signals originating from nonfaradaic
E appl = E 1/2 − log ,
nF id − i processes. A summary of these techniques is presented in
The current, which is diffusion limited, is a function of Table XI to indicate electric potential application methods.
electrode geometry as well as other parameters, as shown The speed of techniques such as square wave voltammetry
in the simple form of the Ilkovic equation for a classical makes such methods preferred for many analytical situa-
dc polarographic experiment using a dropping mercury tions. Polarographic techniques are often used in analyses
electrode, of trace concentrations of toxic metals and are applicable
to almost every element in the periodic table. Analyses of
i d = 607nCD 1/2 m 2/3 t 1/6 , concentration and reaction mechanisms of organic species
where i d is the average diffusion current in microamperes containing certain functional groups are also important.
for each drop, n is the redox electron transfer, D is the ana-
4. Amperometric Titrations
The methods of polarography can be used to follow the
progress of a titration that produces or consumes some
electroactive species. The voltage applied to the work-
ing electrode is constant and sufficient to provide the
diffusion-limited current for the observable analyte. The
current is then recorded as a function of the volume of
titrant added to the sample solution. Since the working
electrode is a microelectrode with an area of only a few
square millimeters, and the current that flows is usually
of the order of microamperes or less, only an insignificant
amount of analyte ion is removed from the sample solu-
tion. This implies that the redox process involved at the
working electrode has a negligible effect on the chemical
FIGURE 17 Electrochemical profile observed in a classic polaro- titration reaction and does not introduce significant error
graphic experiment. in most instances.
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
CH+ +
3 + CH4 → C2 H5 + H2 .
Typical electrochemical cell arrangements employ a three-
electrode potentiostat with a platinum working electrode. These particles can react with the sample molecule (MH)
It is possible in many cases to arrange the experiment in a number of ways:
so that the electrolysis deposits the analyte quantitatively
on the working electrode. The mass change of the elec-
CH+ +
5 + MH → CH4 + MH2
trode can be related to the number of coulombs of the
charge associated with the deposition to determine equiv- CH+
3 + MH → CH4 + M
+
Other sources are field desorption, spark source, thermal are brought to a focus close to where the ion detector is
ionization, fast atom bombardment, and secondary ion located. Thus, the analyzer separates the ions into streams
mass spectrometry (see SIMS, Section G.4, Surface Anal- of different M/e,
ysis).
M/e = H 2r 2 /2V,
where M is the mass of the ion, e is the charge, H is
2. The Spectrometer
the magnetic field, r is the radius of curvature of the an-
The purpose of the spectrometer is to separate the ions alyzer tube, and V is the accelerating potential employed
emanating from the ion source as efficiently as possible. in the ion source. Usually, V is swept to obtain a mass
Quantitatively, this is expressed as the resolving power spectrum. In the time-of-flight mass spectrometer, ions
of the instrument, which is defined as the ratio M/ M, of different mass, which are produced by pulsed electron
where M and M + M are the mass numbers of two beam impact in the ion source, are accelerated to the same
neighboring peaks of equal intensity in the spectrum. kinetic energy. The ions are then allowed to drift in space
The key factor here is the ability to distinguish M and down a tube of particular length before they are detected
M + M; usually, this is said to be achieved when the (Fig. 19b). Because they have different velocity, the transit
“valley” between the two peaks is no more than 10% of time t varies as
the intensity of M or M + M. Resolution is controlled
M 1 1/2
by a number of instrumental factors, including the method t=L µsec.
of ion separation. e 2V
One widely used system of separation is the magnetic This instrument is useful for the study of fast reactions. An
deflection apparatus. Here, a controllable magnetic field additional spectrometer known as the quadrupole device
causes ions to deflect along curved paths according to their is based on the passage of ions through a region between
mass-to-charge ratio. The instrument is designed such that four, short parallel metal rods of alternating electrical volt-
only those ions that follow the path coinciding with the age and provides low resolving power. This is widely used
arc of the analyzer tube (Fig. 19a) in the magnetic field because of its simplicity, although its physics is very com-
plex.
All the spectrometers described to this point are called
single-focusing systems. These are characterized by low
resolving power, which is due to the variety of kinetic en-
ergies produced in the ion source. In a double-focusing
spectrometer (Fig. 19c), an electrostatic field for ion de-
flection is introduced between the ion source and the mag-
netic deflection device. The electrostatic analyzer ions are
effectively focused into highly defined kinetic energies be-
fore they pass into the magnetic analyzer. The resolving
power of such instruments is on the order of 50,000.
Another relatively new device known as the ion cy-
clotron system, which is not a double-focusing system, can
provide resolution to well over 100,000 and can operate in
a Fourier transform mode, offering speed and sensitivity.
3. Other Components
The sample inlet system for a typical mass spectrometer
is versatile enough to handle gas, liquids, and solids. The
device is usually held at ∼200◦ C and 0.02 torr pressure.
Accordingly, any solids must have a sufficient vapor pres-
sure under these conditions to allow transport to the ion
source as a gas before a spectrum can be recorded. Mod-
ern instruments usually incorporate an electron multiplier
or channel electron multiplier array as a detecting system.
FIGURE 19 Magnetic deflection (a), time-of-flight (b), and Both these devices work on the principle of electrons re-
double-focusing (c) mass spectrometers. leased from a material on ion impact. The electrons are
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
then amplified in number before measurement of the cur- for bond breakage, which allow us to predict the original
rent. It goes without saying that nearly the entire mass structure of a particular species. For example, ring com-
spectrometer must be kept under high vacuum with diffu- pounds usually contain mass numbers characteristic of the
sion pumps, backed by rotary oil pumps. If the spectrom- ring (due to its inherent stability). As can be appreciated,
eter is connected to a gas chromatograph, an interface is the whole philosophy behind working out the structure of
inserted between the inlet system and the chromatograph a complete unknown from its mass spectrum is akin to the
to allow sample transport but to remove carrier gas. Effu- development of a solved jigsaw puzzle. This procedure is
sion, membrane, and jet separators are used for this pur- quite different from the use of the computer to match the
pose. Finally, most modern high-resolution instruments mass spectrum of an “unknown” with its memory bank
utilize dedicated computers not only for data handling, but of data, as mentioned above. In this case we are using the
also for total control of the instrument. Many facilities are “fingerprint” approach to identification.
incorporated in these configurations—for example, com-
parison “libraries” of recorded mass spectra of thousands
F. Thermal Methods
of known compounds.
Thermal methods of analysis involve the measurement of
a physical property of a sample as a function of controlled
4. Analytical Use temperature. Equipment is available for detecting transi-
The mass spectrometer is clearly of tremendous impor- tion temperatures, energies of transitions, weight loss, size
tance in the identification of chemical compounds. This changes, elasticity, and changes in solution temperature
is because the ionization of compounds produces unique plotted as a function of added volume of a reagent.
fragmentation patterns. Accordingly, the correlation of
mass spectra with molecular structure provides a beau-
1. Thermogravimetry
tiful foundation for chemical analysis. In identification
procedures at a reasonably low level, we can use the mass In thermogravimetry (TGA), weight change associated
spectrum in two distinct ways. First, the highest mass peak with a transition in the sample as a function of temper-
in the spectrum (Fig. 20) usually is associated with the ature is measured. For example, if water molecules of hy-
molecular weight of the compound, a particularly valu- dration can be removed thermally, they can be detected
able piece of initial information. Furthermore, at masses 1 by apparatus for TGA. Practically, the sample is placed
and 2 larger than this value, we find the effect of naturally in a crucible, which is part of an automatic recording an-
occurring heavier isotopes. This aids us in deciding on a alytical balance. The sample configuration is then intro-
formula for a particular molecular weight when we might duced into a furnace in which the temperature is mea-
have several choices. Second, fragmentation produces a sured accurately (usually room temperature to ∼1200◦ C).
set of characteristic mass peaks for each compound. De- The temperature of the furnace is raised gradually, often
struction of a particular ion occurs because there is enough in the heating range 5–10◦ C/min, and any weight losses
residual energy to cause bond breakage, as well as removal (increases) are recorded. From plots of loss (increase)
of electrons. Fortunately, a number of rules seem to apply in weight vs temperature, information about the gross
structure of the sample components and reaction mech-
anisms can be obtained. The technique is often employed
in conjunction with other thermal methods such as differ-
ential thermal analysis (DTA) and differential scanning
calorimetry (DSC).
FIGURE 22 Thermogravimetric and differential thermal analysis FIGURE 23 General schematic representing the experiment of
profiles for Ca(CH3 COO)2 ·H2 O. electron spectroscopy.
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
1. Vacuum Ultraviolet
Photoelectron Spectroscopy
In vacuum ultraviolet photoelectron spectroscopy (UPS),
the sample atom or molecule is exposed to radiation in the
vacuum ultraviolet region of the electromagnetic spec-
trum. A readily available source of radiation is the he-
lium discharge lamp, which produces a sharp HeI line at
21.2 eV. Since the energy required for photoionization of
sets of valence electrons is in the vicinity of 6 eV to this
energy, we obtain a polyenergetic emission of electrons
described by the Einstein relation
E n = hν − In ,
where In is the ionization energy of the nth species of
electron and E n is the kinetic energy of the bunch of elec-
trons ejected by a photon of energy hν. Usually, we form
an experimental plot of numbers of electrons vs ioniza-
tion potential (the photoelectron spectrum), since hν is a
known quantity and we wish to know In .
A great deal of work has been carried out by UPS
on gas-phase molecules. In particular, electronic energy FIGURE 24 Typical vacuum ultraviolet photoelectron (a), wide-
scan X-ray photoelectron (b), and narrow-scan X-ray photoelec-
level and vibrational fine structure information can be
tron (c) spectra.
generated. A typical ultraviolet photoelectron spectrum
is shown in Figs. 24a. Ionization of a nonbonding electron
results in little change in the molecular internuclear dis-
tance and sharp peaks are obtained, whereas the removal
2. X-Ray Photoelectron Spectroscopy or Electron
of a bonding electron causes changes in bond length and
Spectroscopy for Chemical Analysis
a distribution of vibrational transitions resulting in broad
peaks. Such vibrational structure has been used to study The development of this technique resulted in the award
vibrational frequencies in ions, to distinguish the nature of of the Nobel Prize in physics to Kai Siegbahn of Sweden.
electron bonding, and to examine the consequences of the Here, the sample is exposed to a beam of X-rays (although
Jahn–Teller theorem. Also found in spectra are the conse- synchrotron radiation has been used), causing ejection of
quences of spin–orbit coupling, which occurs as a result of both valence and more deeply held (core-level) electrons.
the production of a molecular ion in an orbitally degenerate Experimentally, by far the most work has been accom-
state. The hydrogen halide molecules show the expected plished with X-ray sources MgKα at 1253.6 eV and AlKα
progressive increase in splitting of the orbitals of halogen at 1486.6 eV, and most studies have been related to solid
lone-pair character (HF, 0.033 eV; HCl, 0.073 eV; HBr, samples and core-level spectra. A very important feature
0.32 eV; HI, 0.66 eV). In terms of chemical applications, of X-ray photoelectron spectroscopy (XPS) is that it is a
UPS has been employed for “fingerprint” identification surface-sensitive technique. Electrons produced by X-ray
in analytical chemistry, in molecular conformation stud- beam ionization are greatly attenuated by collision with
ies, for examination of the orbital structure of transient sample atoms or molecules according to the expression
species, and in correlation experiments with theoretical
calculations of orbital energies. I (x) = I 0 (x)e(−x/λ) ,
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
sample. Here, lateral analysis can be performed by move- 5. Has high sensitivity
ment of the impacting electron beam, and depth studies 6. Has high spatial resolution
can be achieved by removing layers of sample by impinge- 7. Is applicable to a wide range of samples
ment of a sputtering ion beam. 8. Does not discriminate against any component
Not surprisingly, AES has found tremendous use in the 9. Has no influence on surface composition and structure
analysis of surfaces of samples in metallurgy and mate-
rials science. It has been used extensively in alloy analy- As expected, no single technique possesses all these
sis, metal oxidation, segregation, adsorption phenomena, requirements. The battery of methods that are available
catalysis, electrodeposition, corrosion, films and coatings, is outlined in Table XII. From these data it is clear that
tribology, adhesion, and the semiconductor industry. the overall strategy is the study of information carried by
emitted photons, ions, or electrons after perturbation of a
4. Surface Analysis
We now describe briefly the principles of a number
In view of the comments regarding the use of XPS and of important methods. In secondary-ion mass spectrome-
AES in surface analysis, it is appropriate to summarize try (SIMS), solids are bombarded by 1- to 30-keV ions,
this area concisely. An ideal method for surface analysis resulting in the ejection of substrate species as posi-
should possess the following features: tively and negatively charged atomic and molecular par-
ticles (and neutrals). The charged species are subjected
1. Is capable of monolayer examination to mass spectral analysis. The method is used in both
2. Detects elements dynamic and static modes; in the latter the target is
3. Identifies molecular species bombarded “gently,” resulting in a low sputtering rate
4. Elucidates surface topography and a relatively long average lifetime of the monolayer.
Photons Laser optical-emission spectroscopy X-ray photoelectron spectroscopy (XPS) Photodesorption (PD)
(LOES)
Light (Raman) scattering spectroscopy Ultraviolet photoelectron spectroscopy
(LS) (UPS)
Fourier transform infrared spectroscopy
(FTIR)
Ellipsometry (E)
Evanescent wave spectrofluorimetry
(EWS)
Electrons Electron microprobe (EMP) Auger electron spectroscopy (AES) Electron-stimulated desorption
Scanning electron microscopy X-ray Scanning electron microscopy (SEM) (ESD)
detection (XSEM) Low-energy electron diffraction (LEED)
Electron-impact energy loss spectroscopy
(EELS)
Ions Ion-induced X-ray spectroscopy (IIX) Ion-neutralization spectroscopy (INS) Secondary-ion mass spectrometry
(SIMS)
Proton-induced X-ray spectroscopy Ion-induced Auger electron spectroscopy Ion-scattering spectroscopy (ISS)
(PIX) (IAES)
Surface composition by analysis of Rutherford backscatter
neutral species and ion-impact spectroscopy (RBS)
radiation (SCANIIR)
Glow-discharge optical spectroscopy
(GDOS)
Electric field — Field electron microscopy (FEM) Atom probe field-ion microscopy
(APFIM)
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
In ion-scattering spectroscopy (ISS), low-energy ions manipulators. Finally, in Table XIII we carry out a direct
(0.5–2 keV) bombard the surface to provide energy spectra comparison of the methods outlined in this article with
characteristic of the masses of the scattering centers. Low- respect to such analytical criteria as sensitivity, limit of
energy electron diffraction (LEED) is concerned with the detection, and lateral resolution. An examination of the
impingement of electrons of energy 5–500 eV on a sur- data reveals that each method has its own strengths and
face to provide a study of the surface diffraction process. weaknesses and that surface methods in general are truly
In electron-impact energy loss spectroscopy (EELS), a complementary.
monoenergetic beam of electrons is imposed on a surface,
and the electrons obtained from the surface are exam-
H. Chemical Sensors
ined for discrete energy losses associated with the vibra-
tional frequencies of adsorbed molecules. Direct observa- Since 1975 there has been a movement toward the in situ
tion of individual molecules and atoms has been achieved analysis of specific species in complex mixtures without
by Scanning Tunnelling Microscopy (STM) and Atomic initial application of separation science to simplify the
Force Microscopy (AFM), both of which bring a fine wire problem. The possible applications of such chemically
with a tip of atomic dimensions to the surface and observe selective devices, particularly those designed to quanti-
interactions of the wire with the surface while the wire tatively monitor organic compounds, cover a wide range
is displaced laterally by piezoelectric crystal mechanical of situations, including on-line monitoring and control of
aFor definitions of acronyms, see Table XII. [From Thompson, M., Baker, M. D., Christie, A., and Tyson, J. F. (1985).“Auger Electron Spectroscopy,”
Wiley-Interscience, New York.]
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
industrial chemical processes, long-term remote environ- selective membrane. The variation of potential of one sur-
mental monitoring, automation of clinical analyses, in vivo face of this membrane can control any electronic current
drug monitoring, and feedback control of artificial organs. through the transistor known as the drain current. A simi-
All chemical sensors employ a basic union between a lar arrangement employing an enzyme–substrate receptor
selective receiving site for ionic or molecular binding and system at the gate also provides for ionic charge control
a transducer that is capable of translating a chemical bind- of transistor electronic current conduction, resulting in a
ing into a useful analytical signal. Transducers can be di- chemically sensitive field effect transistor (CHEMFET).
vided into four categories: semiconductors, conventional Multiple enzyme systems, where one enzyme produces
ion-selective electrodes, optical devices, and piezoelectric an electrochemically inactive product that is consumed as
devices. Other than the ion-selective electrodes described a substrate by another enzyme to form an active prod-
previously, the semiconductor-based gas sensors are the uct, have been successfully used to extend enzyme se-
best known chemical sensors. These devices usually op- lectivity. The selectivity of immunochemical systems has
erate at elevated temperatures, allowing certain gases to been employed by implementation of enzyme-linked as-
interact chemically with the semiconductor surface to alter says. Direct coupling of redox relay centers of enzymes
its electrical characteristics. Such devices may be selective to conductive electrodes has been achieved by a technique
to certain gases such as CO2 and O2 or to certain classes known as molecular wiring and avoids the indirect anal-
of molecules such as hydrocarbons and may take the form ysis of products of enzyme–substrate reactions. This fast
of chemiresistors or thin-film metal oxides. and sensitive technique measures current flow and is com-
Better selectivity can be attained by the use of special- mercially available.
ized chemical receptors. The ion-exchange chemistry of
ion-selective membranes is one manner in which selec-
2. Optical Devices
tivity is enhanced for simple inorganic ions and gases.
Nature provides the materials for the complicated task of In analogy to electrodes, chemically selective optical de-
selectively complexing biochemicals through molecular vices have been termed optrodes. These systems employ
recognition processes. The selectivity of these receptors is absorption and luminescence strategies and make use of
based on the tertiary structure of proteins and polypeptides waveguide fiber-optic and laser technology for miniatur-
and on nucleic acid complementary pairing. Enzyme– ization. Light in the ultraviolet or visible region of the
substrate, antibody–antigen, lectin–saccharide, and hor- electromagnetic spectrum can be passed through an opti-
mone receptor–hormone selective reactions have all been cal fiber to a remote reaction vessel. Detection may de-
employed in conjunction with “building block” construc- pend on observation of light passing through the reaction
tion and a variety of transducers to produce sensitive and zone, or being transmitted toward a detector by another
selective devices. Such selective chemistry has also found fiber, or by the evanescent wave phenomenon encoun-
wide application in techniques such as immunoassay. The tered in techniques such as attenuated total reflection or
development of catalytic antibodies (synzymes) and ge- total internal reflection fluorescence spectroscopy. In con-
netic engineering to express proteins of DNA of inter- trast to electrochemical devices, these systems offer ad-
est for chemical selectivity and the introduction of artifi- vantages such as insensitivity to electrical interference,
cial receptor sites and selective surfaces have substantially elimination of reference electrodes, possibility for change
broadened the commercial potential for long-term devel- of reagent phase, distributed multisensor operation with
opment of a wide range of chemical sensors. the use of one optical detector (spectrophotometer), mul-
tiwavelength analysis, time-resolved analysis, and a com-
plementary range of analytes that may not be electrochem-
1. Electrochemical Devices
ically active. One example is a pH sensor, based on optical
These devices are based on the measurement of either elec- absorption changes of an indicator dye trapped in a gel ma-
trochemical potential or faradaic current associated with trix in a cell at the tip of a fiber-optic system. Ratio meth-
redox reactions at an electrode. They are particularly suit- ods of quantitative analysis and multiwavelength analy-
able for enzyme–substrate receptor systems by virtue of sis for selective observation of two or more species are
the ionic products often produced in such reactions. The preferred for elimination of background drift and interfer-
sensing membranes of the ion-selective electrodes previ- ence problems. A form of evanescent wave spectroscopy
ously described have been combined with semiconductor in which the electromagnetic field is coupled into the con-
devices for miniaturization, low-impedance output, signal duction band of a thin metal film is currently being com-
amplification, and capability of on-chip processing. The mercialized. This technique, known as Surface Plasmon
ion-sensitive field effect transistor (ISFET) is based on re- Resonance Spectroscopy (SPR), is very sensitive to any
placement of the conventional transistor gate with the ion- organic reagents located on the exterior of the metal film
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
and can be used to picomolar detection limits for many aging, statistical analysis, integration, and data storage.
immunochemical analyses. Often these functions can be handled by a micro- or mini-
computer with a minimum of 8-bit word lengths, though
greater word length and greater capability of microproces-
3. Piezoelectric Devices
sor addressing allow for increased processing speeds and
It is well known that certain crystalline substances such greater on-line data storage capability. Data acquisition is
as quartz generate an electrical voltage when subjected to usually performed by specialized analog-to-digital circuit
physical compression. Similarly, it is possible to induce packages, which may receive polling requests and timing
a constant frequency of physical oscillation of the crystal signals from the microprocessor. Two common devices
planes when a regular alternating voltage is applied to such for transducing analog instrument measurements into dig-
a crystal. Piezoelectric crystals coated with thin films of ital form are known as counter converters and successive
selective adsorbent have found application as conventional approximation counters. Some instruments acquire com-
gas detectors when employed as physical oscillators. The plicated and extensive data sets at rapid rates. Such data
mass changes caused by the adsorption of gases to the may be difficult to interpret manually, but with the aid of a
crystal cause the oscillation frequency to change, f, in reference library, aspects such as compound identification
proportion to adsorbed mass as can readily be established by computer. Such reference li-
braries are used extensively for infrared and mass spectral
f = −2 f q2 (Pq Vq )m f , analysis.
where f q is the quartz crystal frequency, Pq is the quartz
density, Vq is the velocity of the wave in the quartz, and 2. Microprocessor Control
m f is the mass per unit area for the deposited matter on
Due to its great speed, a computer can often control an in-
the quartz surface. It has been demonstrated that mass
strument more efficiently than is possible by manual oper-
responses and also microviscosity responses associated
ation. Such active instrument control must occur in at least
with surface reactions can be obtained from enzyme–
two steps. The microprocessor must first ascertain the state
substrate and immunochemical complexation on bulk
of some variable, and then action must be taken to activate
acoustic wave piezoelectric quartz crystals. Furthermore,
an instrument control to adjust the value of the variable.
both bulk acoustic and surface acoustic wave devices can
Many instruments employing microprocessor control con-
be used directly in aqueous media in certain configurations
tain more than one processor and can perform a series
for quantitative analysis.
of complicated feedback functions. These processors are
considered to be “dedicated” to a particular series of tasks
and receive their operating commands from a hard-wired
III. COMPUTERS IN
instruction set as well as the instrument operator. The sim-
ANALYTICAL CHEMISTRY
plest microprocessors operate with instruction sets based
on 8-bit word lengths. Memory increments of 256 words
A. Instrument Control and Data Handling
are available for such systems, and often only one or two
Since 1975 there has been a tremendous revolution in ca- of these memory increments are required for instrument
pability and cost of computers and microprocessors, re- control. Control usually takes the form of activating a stan-
sulting in the incorporation of such technology into al- dard electrical relay device (which may drive a motor or
most every modern analytical instrument available today. switch) by toggling an active line connected to the relay
Computers are fundamental to two areas of analytical and controlled by the microprocessor between electrical
chemistry: (1) the applications of mathematics to analysis ground and a standard 5-V high level. More complicated
(chemometrics), encompassing such topics as factor anal- control systems involve more control relays and necessi-
ysis, multiple regression, pattern recognition, optimiza- tate the availability of more control lines and microproces-
tion, and statistics, and (2) interactions with analytical sors. The time-sharing capabilities of a minicomputer are
instruments. The latter can actually be considered to be often used in controlling a number of microcomputers or
either passive or active interaction, as determined by com- microprocessors in complex instruments. Significant ad-
puter control of instrument operation. vances in speed and therefore “real time” complexity of
instrument operation have been achieved as new 16- and
32-bit microprocessors have entered the market. The ad-
1. Data Acquisition and Processing
vantages of speed and increased memory have aided in the
This area usually implies a passive function involving the development of expert systems and parallel processing. A
actual collection of data, simple calculations, data aver- revolution in instrument control and data processing is
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
currently developing with the advent of parallel process- processing, resultant conversion, and information organi-
ing and neural networks for interactive “learning.” zation. Software packages for these purposes are commer-
cially available.
3. Fourier Transform Analysis
1. Common Chemometric Methods
One application of high-speed computers to data anal-
ysis is often found in spectrophotometric applications, The three most commonly used chemometric methods are
such as infrared and nuclear magnetic resonance tech- discussed in the following subsections.
niques. Samples can be irradiated with broad ranges of
frequencies from the appropriate regions of the electro- Multiple regression analysis. This is suitable for
magnetic spectrum and will absorb certain discrete fre- data modeling and expresses data as a simple equation.
quencies dependent on sample chemistry. Each indepen- The process begins with experimentation to produce a vec-
dent frequency that can be observed (resolved) in the range tor of measured data known as the “dependent” variables.
of energies employed can be represented as a sinusoidal Then a limited number of “factors” are considered to be
oscillation. The simultaneous superpositioning of all the significant for the determination of data values, and these
available frequencies produces both constructive and de- “independent” variables are used to prepare a model for
structive interference, resulting in a well-defined complex the data. Finally, coefficients, as shown below, are calcu-
waveform pattern. Interaction of the sample with discrete lated by least-squares analysis to represent the significance
frequencies will alter the waveform pattern, which will or weighting of the independent variables. The result is a
then contain the analytical interaction information in the calculation of “regression coefficients” to prepare a math-
form of a time “domain.” This can be converted to a con- ematical model that is suitable for preditions,
ventional frequency-domain spectrum by the fast Fourier d = c 1 i 1 + c2 i 2 + · · · + c n i n ,
transform algorithm, so that individual frequencies that
make up the superimposed waveform can be individually where d represents the dependent variable, c represents
identified and plotted in conventional formats. Data must the regression coefficient, and i represents the independent
be sampled and digitized at a rate at least twice the value of variable.
the ratio of the range of frequencies encountered divided
by the frequency resolution desired. The major advantage Factor analysis. This method is used to interpret
of this technique is that all frequencies are simultaneously underlying factors responsible for data and is one of the
measured, and a complete conventional spectrum can be most versatile chemometric methods. Factor analysis pro-
constructed in seconds for any one measurement. Since vides a purely mathematical model prepared from abstract
these spectra are digitized and contain frequency refer- values, which are related to a data matrix as follows,
ence information, it is possible to sum sequential spectra D = RC,
to improve signal-to-noise ratio. Signals increase linearly
with spectral addition, while noise increases as the square where D represents the data matrix and R and C represent
root of the number of spectra that are combined. factors for each row and column. The factors are math-
ematically transformed so that their significance can be
interpreted with respect to the data. This results in the es-
B. Chemometrics tablishment of the number of significant factors and assists
The term chemometrics describes the interface between in the correlation of data and the application of physical
analytical chemistry and applied mathematics, where significance to the factors.
mathematical and statistical methods are employed to
maximize information quality in a chemical experiment. Pattern recognition. This procedure allows the
Most chemometric methods involve matrix algebra, which classification of a species to be made on the basis of a
is efficiently handled by computer, and numerous pro- series of measurements that establish a pattern. Proce-
grams are presently available. A number of reviews have durally, a matrix describing the patterns of a number of
been written on this broad subject area, which includes species is constructed. Then a decision vector is designed
such topics as statistics, modeling and parameter esti- by the use of standards to divide the patterns into discrete
mation, resolution, calibration, signal processing, image classifications, resulting in a mathematical form,
analysis, factor analysis, pattern recognition, optimization p = V1 d1 + V2 d2 + · · · + Vn dn ,
strategies, and artificial intelligence. Appropriate topics
can be chosen to optimize an analysis at each level of where p represents a set of patterns, V represents com-
experimentation, including sampling, measurement, data ponents of the decision vector, and d represents the data
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
for the species of interest. The pattern p is assigned and method originally described must employ steps of fixed
values of d are determined so that a decision vector V can size, which can result in excessive experimentation when
be calculated. When the calculated vector V is mathemat- step size is small or in poor precision for large steps. A
ically combined with a new set of experimental data, a more efficient solution employs variable step size through-
pattern p is calculated for the experimental species and out the entire procedure, allowing expansion (accelera-
can be fitted to previous classifications. tion) of the simplex in favorable directions and contrac-
tion in zones that produce poor results. The distance to be
moved is controlled by constant arbitrarily chosen multi-
2. Optimization by the Simplex Method
plication factors, which are multiplied with the distance of
Numerous mathematical techniques exist for solving a se- movement obtained on reflection. Eventually, the simplex
ries of simultaneous equations given defined boundary contracts as movement to the optimum occurs. The process
constraints in order to maximize or minimize a particu- halts when the distance of movement has dropped below
lar parameter. The general acceptance and implementa- some predetermined value, which is generally governed
tion of techniques such as linear programming attest to by experimental uncertainty or time limitations. Certain
the power of optimization strategies. The simplex method difficulties exist in the application of simplex methods
is an “evolutionary operations” method that has been used when considering error sources:
systematically in many problems. A simplex is a geomet-
ric figure whose vertices are defined by the number of ex- 1. The method cannot be used if discontinuous variables
perimental parameters plus 1. Each point of the simplex are chosen.
represents the actual measured analytical response at a set 2. Movement to a local optimum may occur if numerous
of chosen experimental parameters. Represented in some optima exist.
n-dimensional space, one vertex of the simplex always 3. Parameters must be judiciously selected to ensure that
represents the case of worst response in the experimental. nontrivial analyses occur.
A mirror reflection through a symmetry plane away from 4. As many significant parameters as possible should be
the point of worst response (assuming the response will be included in the simplex so that no important factors
greater at a point opposite to the worst case) generates an- are overlooked. This subsequently increases the
other simplex. An experiment is then performed using the experimental work for each step in the simplex
new parameters to determine which vertex represents the generation.
new worst case response. A reiteration process following
four well-defined rules allows movement along the “re-
sponse surface,” resulting in eventual convergence to the 3. Selectivity vs Specificity
optimal experimental conditions.
The basic rules are as follows: A definition of terminology has been attempted, where the
upper limit of the concept of selectivity implies specificity.
1. Rejection of the point with the worst result is followed A fully selective system can measure one component in the
by replacement with its mirror image across a line or presence of many others, while fully specific implies that
plane generated by the other remaining points. in all situations only one component is measured and other
2. If the new point has the worst response, the previous components in the experiment do not produce any signal.
simplex is regenerated and the process applied in rule A nonselective system produces an analytical signal due to
1 is repeated for the second worst case point. all components in the experiment. For any of these cases,
3. If one point is common to three successive simplexes, the measured signal x is a function of concentration c of
it represents the optimum, provided that the point the available component and is related to the latter by a
represented the best response in each case. If this is normalization parameter γ , where
false, the entire process must be repeated using new
x = γ c.
initial starting points.
4. Boundary conditions are defined so that if a point falls The element γ is determined by the sensitivity parameter
outside accepted bounds, it is assigned an artificially dx/dc, and the sensitivity of the method is numerically
low value, which forces the simplex to move into the determined by the value of γ . This example can readily
useful calculation area. be expanded to consider a multicomponent case, where
x, γ , and c become matrix representations and a partial
A variation of the latter optimization procedure known sensitivity ∂ x/∂c is employed. A mathematical rearrange-
as “modified simplex optimization” has evolved to elimi- ment of the γi j matrix to place the largest γ value in each
nate the limitations imposed by the simplex method. The row on the main diagonal results in a useful “calibration
P1: GJB Revised Pages
Encyclopedia of Physical Science and Technology En001f25 May 7, 2001 13:58
Sorum, C. H., and Lagowski, J. J. (1991). “Introduction to Semimicro Svehla, G., and Suehla G. (1996). “Vogel’s Qualitative Inorganic
Qualitative Analysis,” 7th ed., Prentice-Hall, Englewood Cliffs, Analysis,” 7th ed., Addison-Wesley, Reading, MA.
NJ. Wilson, I. D., Adlard, T. R., Cooke, M., and Poole, C. F., eds.
Strobel, H. A., and Heineman, W. R. (1989). “Chemical Instrumentation: (2000). “Encyclopedia of Separation Science,” Academic Press, San
A Systematic Approach,” 3rd ed., Wiley, New York. Diego.
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology en001d42 May 26, 2001 14:55
Atomic Spectrometry
Vahid Majidi
Chemistry Division, Los Alamos National Laboratory
765
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
level (10−9 g) can be determined in samples of a few is initially volatilized and ultimately decomposed into its
micrograms (10−6 g) by electrothermal vaporization constituent atoms (sodium and chlorine) by the heat of
(ETV) atomic absorption. Solution concentrations at the the flame. The thermal energy from the flame excites the
nanogram (10−9 g) per milliliter level can be routinely gas-phase sodium atoms from the ground state (0) to an
measured by most of these techniques. Conversely, us- excited state (1). The efficiency of this excitation depends
ing plasma mass spectrometry, absolute mass detection on the gas-phase temperature (the wavy line in Fig. 1A
limits of a few femtograms (10−15 g by ETV–MS) or represents the thermal excitation processes). The sodium
measuring solution concentrations of parts per quadrillion atom in the excited state then spontaneously decays back
(10−15 g/ml) are not uncommon. All the techniques use the down to the ground state, either by collisional transfer of
phenomenon of electronic excitation. This involves transi- energy to other species or by atomic emission (loss of en-
tions among the outer electrons in the atomic orbitals, with ergy by emission of a photon). This is illustrated by the
accompanying ionization, emission, or absorption of ra- straight line from the excited state down to the ground
diation. In photon-based detection, the intensity of these state in Fig. 1A. The wavelength of this emitted photon
emissions and absorptions can be related directly to the depends on the energy difference between the excited state
concentration of an element in a sample. For ion-based and the ground state.
detection, the obtained electric current is directly propor- The amount of light that is emitted from the analyte
tional to the analyte concentration. is proportional to the number of atoms in the flame or
plasma. Hence, atomic emission spectrometry can be both
a qualitative analytical technique, in which the identity of a
I. ATOMIC SPECTROMETRIC metal is revealed by the observed color (wavelength), and
TECHNIQUES a quantitative analytical technique, in which the intensity
of the light emitted from the analyte is a function of the
There are three traditional analytical atomic spectrometric number of atoms.
techniques: atomic emission spectrometry (AES), atomic The experimental arrangement involved in an AES mea-
absorption spectrometry (AAS), and atomic fluorescence surement is shown in Fig. 1E. The hot analyte environ-
spectrometry (AFS). In all three cases, the analytical sig- ment, which is able to break down and excite atoms, is
nal is generated by photons. More recently, however, the called an atom cell. The atom cell can be a flame, plasma,
detection of resulting ions from an electronic excitation a heated graphite tube, or any other environment where
has led to superbly sensitive detection of elements. It is the analyte is observed in a spatially confined arrange-
possible to distinguish among these techniques by con- ment. In Fig. 1, the detector box is used to represent a
sidering the excitation mechanism and the way the an- detection system, which is able to identify the wavelength
alytical signal is detected. The general schematic rep- and measure the intensity of the emitted radiation. The ex-
resentation for various types of atomic spectrometry is perimental arrangement is the simplest of the three optical
shown in Fig. 1, where two-level electronic systems are atomic spectrometric techniques.
depicted. The ground state is labeled 0, and the excited
state is labeled 1. In general, the transition originat-
B. Atomic Absorption Spectrometry (AAS)
ing from or terminating to the ground state (or another
of the lowest lying excited states) is most commonly Figure 1B illustrates the principle of atomic absorption
used for atomic spectrometric measurements (Ingle and spectrometry. The excitation from the ground state to the
Crouch, 1981). upper state is by the absorption of light energy. An atom
cell (a flame or a heated graphite tube) is used to decom-
pose compounds, but the energy for excitation is drawn
A. Atomic Emission Spectrometry (AES)
primarily from a light source such as a white light source
The most basic measurements in analytical atomic spec- (which emits all wavelengths) or a hollow cathode lamp
trometry can be traced back to Thomas Melville, who in (which emits narrow spectral lines). In AAS, the detection
1752 reported his observations on spectra of mixtures of system looks at the light source directly (Fig. 1F) and sees
alcohol with sea salts (Laitinen and Ewing, 1981). The an intensity of the light source (I0 ) before any atoms are
simplest example of atomic emission is the experiment of present within the atom cell. When atoms are introduced
putting table salt (sodium chloride) into a flame, which into the atom cell, they absorb some of the radiational
generates a yellow color. Although the example is a sim- energy (solid arrow in Fig. 1B) and are excited from the
ple illustration, the actual events that ultimately lead to ground state to the upper state. The detection system sees
the yellow plume are due to a complex series of chemical this absorption as a reduction in the intensity of the light
and physical processes outlined in Fig. 2. Sodium chloride source from I0 to I (Fig. 1F). The ratio I /I0 is called the
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
FIGURE 1 A simple two-level energy diagram representation for excitation and relaxation mechanisms in the atomic
emission process (A), atomic absorption process (B), atomic fluorescence process (C), and ionization process (D).
The resulting analytical signals from the above processes are measured by experimental arrangements depicted in
E, F, G, and H, respectively.
transmittance and is related to the concentration of the to yield excited atoms. The direction of observation of the
absorbing analyte atoms. In short, the greater the amount fluorescence is at right angles to the angle of incidence
of light that is absorbed, the more atoms are present in the of the excitation source on the atom cell to avoid looking
atom cell. directly at the light source. Once again, a higher population
of atoms in the atom cell produces a greater fluorescence
C. Atomic Fluorescence Spectrometry (AFS) signal.
FIGURE 2 Sequence of events that occur in flame during the conversion of crystalline salts into gas-phase atomic
and ionic species for spectrochemical analysis.
elements and as such an undiscriminating detector (e.g., spectrochemical analysis, one must be able to measure
a charge collector) may be used to selectively identify the photon intensity (to extract quantitative information)
a particular element (the selection is done through the and characteristic wavelength (to extract qualitative infor-
unique excitation wavelength). Conversely, in elemental mation). The optical information encoded in any of the
mass spectrometry, thermal sources (e.g., inductively cou- above spectrometric processes must be decoded in terms
pled plasma) are commonly used to ionize most elements of intensity and wavelength.
present within the plasma, to varying degrees. Therefore,
a mass analyzer is needed prior to the detection to provide
quantitative information for a given element without inter- A. Wavelength Selection
ference from all other ions present. Regardless of the mode The light emanating from an atom cell or the light used in
of ionization or detection, the ion current generated from atomic absorption experiments is generally composed of
an atom cell is proportional to the total amount of analyte several different wavelengths. The simplest approach for
initially present. The process of ionization and detection wavelength discrimination is to use an optical filter. But, to
is shown schematically in Figs. 1D and H. distinctly identify elements by their characteristic spectra
and to quantify the amount of analyte present by its signal
intensity, a wavelength-selection device is used. These de-
II. OPTICAL DETECTION vices are known as monochromators or polychromators,
depending on their ability to display a single wavelength
In atomic absorption, emission, and fluorescence spec- or multiple wavelengths at one time. Schematic represen-
trometry, the detection and characterization of photons tations of several different types of monochromators and
are the fundamental measurements that ultimately lead to polychromators are shown in Fig. 3.
the determination of elemental identity or the quantitative Typically, the light from an experiment is focused on
evaluation of a given analyte. Therefore, for a complete the entrance slit of the monochromator. The light on the
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
FIGURE 3 Schematic representation of the most frequently used grating-based monochromators (A, B, C) and
polychromators (D, E, F). A and D are known as Czerny–Turner design. Fastie-Ebert (B) and Littrow (C) have similar
designs in that both use a single reflective mirror. A Rowland circle polychromator is depicted in E, and F represents
an Echelle spectrograph.
slit is then collected by a collimating mirror and converted is the polychromator of choice due to its superb spectral
to parallel light rays that are directed toward a wavelength- resolution.
dispersion element (diffraction grating or prism). The
original light is now separated into discrete wavelengths,
B. Photon Detection
which are imaged by the mirror on the exit plane of the
monochromator. A slit is used to allow for propagation of a Historically, photographic films and plates have been used
unique wavelength out of the monochromator and into the at the exit plane of the monochromators to detect the spec-
detector. In most instances, a monochromator can be con- tral patterns from analytes. The main advantage of this ap-
verted to a polychromator by simply removing the exit slit proach is its inherent multi-wavelength detection ability.
and replacing the traditional detector by an area detector Because photographic films and plates are area detectors
(see next section). On the other hand, any polychromator with fantastic spatial resolution, the entire spectrum from
can be used as a monochromator if only a single output a sample may be recorded on a 30-cm plate, which ulti-
channel is used during an experiment. Although many mately results in simultaneous multi-elemental analysis.
wavelength-selection devices are available commercially, Unfortunately, the dynamic range of photographic emul-
the designs shown in Fig. 3 are among the most popular. sion is limited and the response function is typically non-
The Czerny–Turner monochromator is extremely popular linear. Furthermore, the overall process is extremely time
because of its compact design, while the Echelle system consuming and chemically intensive (need for a darkroom,
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
developing time, and chemicals for the development pro- III. ATOMIC EMISSION SPECTROMETRY
cess). Today, photomultipliers (PMT), photodiodes, and
array detectors completely dominate the photon detection A. Theoretical Background
arena. Schematic representations for a photomultiplier and
a charge injection device (a two-dimensional array detec- In atomic emission spectrometry, the intensity of emis-
tor) are shown in Fig. 4. Photomultipliers and photodiodes sion from the analyte atoms depends on the number of
are used as single-channel detectors with both monochro- atoms that are in the excited state. If the temperature of
mators and polychromators. Array detectors are used for the atom cell is increased, then more atoms undergo en-
simultaneous detection over a given spectral range. ergy exchange collisions with the surrounding hot gases.
In photomultipliers, a photon strikes the photoactive More atoms then possess enough energy to be excited to
material on a photocathode, resulting in ejection of an an upper state and drop back down to the ground state by
electron from the surface. The initial electron is acceler- loss of a photon. The expression that relates the tempera-
ated toward the first dynode, which upon collision releases ture of the atom cell to the number of atoms in the excited
several secondary electrons. This process is repeated by state is the Boltzmann expression. If n 0 is the number of
using several additional dynodes to yield an overall signal atoms in any given state with energy E 0 , then the number
amplification of 106 . Solid-state detectors (photodiodes n ∗ in an excited state with energy E q , is given by
and charge injection devices), however, do not intrinsi- n ∗ n 0 = gq g0 e−(Eq −E0 )/kT (1)
cally amplify the signal. In these devices, the initial pho-
ton results in a single charge-hole separation or single where k is the Boltzmann constant, and T is the absolute
charge storage. In all cases, the signal from the detec- temperature. The constants gq and g0 are the statistical
tor is electronically conditioned (e.g., amplified, filtered, weights (degeneracy), which account for the fact that there
time-gated, etc.) and is converted into an analog or digital are often several states with the same energy E q and E 0 ,
format to be used with a display device (e.g., computer, respectively.
analog gauge, digital display, etc.). The Boltzmann expression predicts the number of
atoms in the excited state, relative to the number of atoms
in the ground state. For example, considering the atomic
emission from Cs (852 nm; E q − E 0 = 1.46 eV) and Zn
(213.8 nm; E q − E 0 = 5.8 eV) at 4000 K, the n ∗ /n 0 ra-
tios for these elements are 2.98 × 10−2 (gq /g0 = 2) and
1.48 × 10−6 (gq /g0 = 3), respectively. Another way to use
the above equation is to generate a plot for the Boltzmann
factor (e−(Eq −E0 )/kT ) as a function of the excitation tem-
perature, as shown in Fig. 5. For a given transition (E),
the Boltzmann factor becomes dependent only on the tem-
perature; as a result, higher temperatures within the atom
cell will generate a larger population of the excited species,
which ultimately leads to higher emission intensities.
Flames have temperatures in the 2000–3000 K range,
while atmospheric pressure plasmas are in the 4000–
10,000 K range. In the air–acetylene flame, which has
a temperature of about 2540 K, it is possible to see rela-
tively large amounts of light emitted by sodium, cesium,
calcium, and other metals, but it is very difficult to see zinc
emission in the same flame. If a nitrous oxide–acetylene
flame is used, with its typical temperature of 3150 K,
larger signals from zinc and many other metals can be
seen. In general, flames are not a convenient source for
atomic emission measurements in light of the fact that
in plasmas, most elements can be determined with high
sensitivity by AES.
FIGURE 4 Examples of single-channel (photomultiplier) and two- The primary variable that determines the temperature
dimensional array (charge injection device) detectors for spectro- necessary to excite strong atomic emission signals from a
scopic applications. particular element is the energy difference (E q − E 0 ). The
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
FIGURE 5 The Boltzmann factor as a function of the excitation temperature. Each curve represents a particular E
in the ultraviolet or visible region of the electromagnetic radiation.
larger this difference, the more energy required to cause During the past two decades, plasma spectroscopy has
the element to be excited to the higher state. In general, dominated the analytical market for atomic emission spec-
elements with analytical lines (usually resonance lines) in trometry. Consequently, we will not discuss the use of
the blue or ultraviolet do not give a strong atomic emis- flames for atomic emission, as the industry, in large, has
sion signal in a flame. These trends can be seen in Fig. 5, moved toward plasmas for this application. On some oc-
where the analytes with short analytical wavelengths have casions, flame emission is used for analysis of sodium
a correspondingly low Boltzmann factor. and potassium. Nonetheless, these elements can be easily
In addition to the Boltzmann equation, the following determined by inductively coupled plasma spectrometry.
expression is used to account for other variables that affect
the atomic emission signal at low concentrations:
B. Inductively Coupled Plasma Spectrometry
= Ahν0 n(gq /g0 )a L(/4π )e−(Eq −E0 ) (2)
1. Instrumentation
where is the radiant power (in watts) emanating from the
atom cell, and n is the number of atoms in the atom cell A diagram of the inductively coupled plasma (ICP) device
per cm3 . A is the Einstein transition probability, which is shown in Fig. 6. The effective temperature of plasma is
indicates the probability that an excited atom, per unit two to three times higher than those observed in flames
time, will fall into a lower level by spontaneous emission (in the 4000–10,000 K region). Passing argon gas through
of a photon. This is a rate constant for the first-order decay a set of concentric quartz tubes within a strong radio fre-
of the atom from the excited state to the lower state. In the quency (rf) field generates the plasma. Radio-frequency
above equation, a is the area of the atom cell observed by energy at about 27 MHz and 2 kW of power is coupled to
the detector, L is the thickness of the atom cell, and is the the water-cooled induction coils at the top of the quartz
solid angle (in steradians) over which the emission is being plasma tube. This energy is sufficient to accelerate ion-
observed. There are other losses that are often difficult to ized particles into many collisions which causes further
quantify, such as light losses in the optical components. In ionization and emission of significant background radia-
principle, if there are n ∗ atoms per cm3 in the excited state, tion from the argon gas. The plasma is ignited by seeding
then n ∗ A photons with energy hν0 will be emitted per unit with electrons generated by an electrical discharge. Af-
time and per cm3 . In the case of sodium, which has two ter the onset, the plasma becomes a self-propagating and
excited states of similar energy (589.0 and 589.6 nm) and -sustaining process. The tangentially introduced argon gas
similar A values, the relative emission intensities are the (Fig. 6) spins the plasma into a torroidal shape at the point
same as the ratio of their statistical weights (1 : 2). where the sample is introduced. The sample is injected
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
3. Sensitivity
Atomic emission signals in the ICP are much larger
than those in the flame for nearly all elements and can
be obtained for a wider range of elements. The high-
temperature, inert-argon environment of the ICP leads to
a more complete atomization and efficient excitation of
analytes, resulting in larger signals.
The copious energy in the ICP causes some atoms to
become completely ionized. For analytes with extensive
ionization, the atomic emission signal is minute, but the
ionic emission signal is quite substantial. The concept of
an ionic emission signal is exactly the same as that of
atomic emission. An ion has a different set of energy lev-
els from the equivalent neutral atom (i.e., ionic emission
signals occur at different wavelengths from the atomic
emission signals). This duality of ICP spectrometry has
led to the use of the name optical emission spectrometry
(OES) for the ICP emission technique rather than atomic
emission spectrometry. Signals from ion transitions are
used just as routinely as atom lines for the determination
of the concentration of an analyte. Furthermore, the ions
FIGURE 8 Atomic emission profile and potential spectral inter-
from an ICP may be extracted into a mass analyzer for ferences. (A) A typical atomic emission line profile intensity as
elemental mass spectrometry (discussed later). a function of wavelength (note the narrow width of the emission
line). (B) Two independent atomic emission profiles emitted from
two different elements that can be resolved with the use of a high-
4. Chemical Interference Effects resolution monochromator. (C) A typical molecular emission band
as a function of wavelength (note the emission coverage over
One of the most important advantages of the ICP over other a wide wavelength range as compared to atomic emission). (D)
atomic emission sources is the near absence of chemical Atomic emission superimposed on a concurrent molecular emis-
interferences. The argon plasma is inert and the tempera- sion (points 1 and 2 represent the background correction regions
ture is so high that nearly all compounds are broken down for the atomic emission signal).
efficiently. Virtually all the analyte is released for excita-
tion and subsequent atomic emission.
width of the spectral line at half-height (the half-width)
is about 5 pm, but it can vary by a factor of ten or more
5. Spectral Interferences
depending upon many fundamental and experimental pa-
Because of the high excitation temperatures attained rameters (e.g., temperature and pressure).
in ICP, the most serious disadvantage of this plasma is Some analytes emit lines that are very close to the lines
the relatively large number of spectral interferences. of other analytes (Fig. 8B). For example, iron emits radia-
Spectral overlap becomes very likely in complex sam- tion at a multitude of lines, any of which might be at almost
ples containing many elements over a wide range of the same wavelength as that emitted by another metal. If
concentrations. The potential spectral overlaps are often the slits on the monochromator are narrow enough and
remedied with the use of high-resolution monochromators if the dispersion of the monochromator is large enough,
(polychromators). spectral lines that are very close to each other will be
The atomic emission signal from a particular element is well separated. The resolution of a monochromator is said
not composed strictly of one wavelength but of a very nar- to be high if it is able to separate close spectral lines.
row range of wavelengths (approximated by Fig. 8). The The highest possible resolution is usually required for
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
TABLE I Elements with Pronounced Spectral Inter- scanning capability of monochromators may be used to
ference for Determination of Be Using 313.042 nm Line correct for these types of interferences. As seen in Table I,
Element Observed changes on Be emissiona (%) the presence of some elements (e.g., presence of Cr, Ta,
and U while measuring the emission from the Be 313.042
Cr −120 line) may cause an over-correction of the background
Ta −20 emission.
U −15 Spectral interferences can be corrected if the magnitude
Ag 0 of the interference is known as a function of the concen-
Gd +30 tration of the interfering element. A correction factor may
Th +75 be calculated and ratioed to the signal size at the ana-
Ti +160 lyte wavelength (concentration ratio method). This is best
a Positive changes in Be emission intensities are due to direct accomplished by making simultaneous measurements of
line or wing overlaps. Negative deviations are due to interference both the signal at the analyte wavelength and the concen-
at a background correction point. tration of the interfering element at another wavelength.
The main requirement is that the concentration of the in-
terfering element can be measured at the other analytical
atomic emission spectrometry, but most of the monochro- wavelengths without any spectral or other interferences.
mators that are commonly used in emission spectrome- Hence, polychromators are ideally suited for this proce-
try cannot resolve two lines that are closer than about dure. It is also assumed that both the interferent signal,
3 pm apart. which causes the spectral interference at the analyte wave-
If two spectral lines from two metals are not satis- length, and the interferent signal at the second wavelength
factorily separated, then a spectral interference occurs. behave in the same way in the plasma. This turns out to
Spectral interferences can still occur even with a large, be an acceptable method of correction for spectral inter-
high-resolution monochromator. The resulting measured ferences due to the lack of chemical interferences within
signal not only represents the element of interest, but also the plasma.
a component of the signal resulting from the interfering An internal standard is frequently used to correct for
element. In general, the signal is larger than it should drift in the signal size as a function of time. An internal
be, which results in an inaccurate measure of the ele- standard is an element of known concentration in a sample
ment concentration. As an example, a list of elements that is present in all examples of the sample. The instru-
with spectral interference for determination of Be is given ment monitors its signal, often simultaneously with the
in Table I. measurements described, and if it drifts in size an appro-
The monochromators that are used for ICP optical emis- priate correction is applied to the analyte signal to com-
sion spectrometry (ICP–OES) often include automatic pensate for the drift. This helps to ensure that the instru-
scanning capability to measure the background at each ment remains accurately calibrated for the duration of the
side of the analytical atom or ion line. The primary ap- analytical run.
proach to discriminating against spectral line interferences
is through the use of a high-resolution monochromator. If
this fails, then one recourse is to find another atomic or
ionic emission line that is sufficiently sensitive but which TABLE II Analytical Wavelengths Used in ICP-AES Analysis
is not affected by a spectral interference. Examples of ac- Primary Alternate Decreased
ceptable alternative emission lines for some elements are analytical analytical sensitivity
presented in Table II. Alternatively, it is possible to re- Analyte wavelength (nm) wavelength (nm) factor
move the element that is causing the spectral interference
Ag 328.068 338.289 2.0
by use of solvent extraction, or another chromatographic
Au 242.795 267.595 1.8
technique, prior to sample introduction.
Be 313.042 234.861 1.2
The atomic emission from inductively coupled plasma
Cd 214.438 228.802 1.1
frequently resides on a substantial background emission.
Cu 324.754 224.700 1.4
The majority of background continuum emission is due
Ga 294.364 417.206 1.4
to bremsstrahlung radiation (deceleration of fast-moving
In 230.606 325.609 1.9
electrons) or an electron-ion recombination process. An-
Ni 221.647 232.003 1.3
other cause of the background radiation is the molecu-
Pb 220.353 216.999 2.1
lar band emission from the OH species originating from
Zn 213.856 202.548 2.2
the aqueous samples (Fig. 8C). As shown in Fig. 8D, the
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
inert gas (argon or neon). The cathode is made of the el- Another popular approach to glow discharge spec-
ement that is to be determined in a sample. Occasionally, troscopy is to use rf power instead of traditional dc power
these plasma sources may be used for determination of sources. The main advantage of rf-GD is its ability to sput-
elements in a solution matrix. In these instances, a few ter nonconductive samples, hence elemental analysis for
microliters of a solution are deposited and dried onto the polymers and ceramics becomes a matter of simple solids
cathode surface. The chamber is then evacuated and back- analysis.
filled with a small amount of Ar gas (1–10 torr). For dried
solutions, the analyte emission is a transient signal that is
4. Other Plasmas
integrated over time and compared to integrated intensities
for a standard solution. Microwave-induced plasmas (MIP) are generated using
The Grimm source is the most popular GD design for the microwave region of electromagnetic radiation (1–
spectrochemical determination of analytes in conductive 3 GHz). The microwave is generated by a magnetron
solids (Fig. 10). The plasma is a restricted discharge on source, and the energy is coupled into an inert gas (e.g.,
the cathode yielding uniform sputtering and well-defined He) containing the sample, using a waveguide and a cav-
craters. The calibration curves are always established with ity. About 50–200 W of power are needed to sustain the
known analyte concentrations in similar matrices. plasma. MIP is an ideal source for analysis of nonmetals
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
FIGURE 10 Two different glow discharge cells based on the hollow cathode and the Grimm-type design.
but it is not the most robust source for analysis of aqueous from the light source is related to the transmittance T ,
solutions. MIP has been successfully, applied to gas chro- where:
matography applications as an atomic emission source.
T = I /I0 (3)
Laser-induced plasma spectroscopy (LIPS), which is
also known as laser-induced breakdown spectroscopy Io and I are the initial intensity of the source and the source
(LIBS), is a versatile tool for elemental analysis. A pulsed intensity after the atom cell, respectively.
laser beam (lasting a few nanoseconds) is focused onto a The path length b in the atom cell (Fig. 1F) needs to be
target at power densities in excess of 100 MW/cm2 . The relatively long to maximize the amount of light absorbed
plasma is formed when the irradiance of the laser beam is by the analyte. The amount of light absorbed depends on
high enough to generate an electric field at the focal vol- the fundamental constant k , the atomic absorption coeffi-
ume that exceeds the breakdown threshold of the target cient. These parameters are related to transmittance in the
medium. Laser plasmas with well-defined spatial resolu- following manner:
tion can be formed in almost any medium. Good analytical
I /I0 = e−k b (4)
results may be obtained for LIBS instruments equipped
with time-resolved detection technology. Laser-induced The absorbance A is the parameter that is usually de-
plasmas can be optimized to yield a detection limit of tens termined with an AAS instrument, where:
of µg/g for certain elements in ideal matrices.
A = −log I /I0 = log 1/T = k b log e = 0.434k b (5)
The constant k prime is proportional to a number of pa-
IV. ATOMIC ABSORPTION rameters, including the number of atoms per unit volume,
SPECTROMETRY the Einstein transition probability for the absorption pro-
cess, and the energy difference (Fig. 1) between levels 0
A. Theoretical Background and 1. The Beer-Lambert law combines these constants
into one constant a to yield the following equation:
Figure 1B illustrates the basic principle of atomic absorp-
tion spectrometry (AAS). The amount of light absorbed A = abc (6)
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
FIGURE 11 A dual-beam flame atomic absorption instrument with the corresponding timing diagram.
Hence, all backgrounds, including any atomic emission and complexing agents such as EDTA (ethylenediaminete-
spectral interferences, are subtracted out automatically. traacetic acid). Releasing agents increase the magnitude
The monochromator and detection system: The mono- of the signal in real samples. They can also change the
chromator for AAS does not need to have the high reso- magnitude of the signal from the standard solutions.
lution necessary for AES, because the selectivity of AAS Spectral interferences: The elegance of atomic absorp-
depends on the line width of the light source as discussed tion lies in the very high selectivity of the technique. The
above. The role of the monochromator is to reject the ma- HCL only emits light that is characteristic of the elements
jority of the background radiation as well as selecting only in the cathode. While in the atom cell there may be many
the pertinent HCL emission line for AAS measurements, elements present, only one element that corresponds to the
thus monochromators for AAS have a moderate resolution cathode material absorbs the light, because only the light
of 0.02–2.0 nm. The detection system is based on a pho- characteristic of that element is emitted by the light source
tomultiplier tube, and the readout electronics are similar and reaches the detector. The HCL acts like a probe to de-
to single-element AES instruments. termine the concentration of only one metal. This means
that virtually no spectral line interferences affect AAS.
The technique is often much simpler to use than AES, and
2. Interference Effects
the instrumentation can be relatively inexpensive while
Physico-chemical interference effects: The flame atom- providing high sensitivity and selectivity.
ization process is affected by chemical interferences that
prevent facile formation of neutral metal species of M.
C. Graphite Furnace Atomic
For example, oxides of many metals are likely to form in
Absorption Spectrometry
the flame environment. Oxygen is abundant in any flame
as a consequence of the oxidant (air, oxygen, or nitrous Graphite furnace AAS, also known as electrothermal
oxide) used in the combustion process. The main con- atomic absorption spectrometry (ETV–AAS), is one of the
cern is whether or not there is enough energy in the flame most sensitive techniques available for routine elemental
to break the bond between the metal and the oxygen to analysis. This technique is capable of the determination
release the metal. If the flame does not have enough en- of picogram (10−12 g) amounts in a few microliters of
ergy, most of the metal oxides remain intact and very little sample.
free metal is available. Phosphates also cause chemical
interference due to formation of a stable compound in the
1. Instrumentation
air-acetylene flame that breaks down slowly to release the
metal atom. This reduces the expected signal size signif- The arrangement for this technique is similar to that for
icantly. In striving to discriminate against chemical inter- flame AAS, except that a graphite furnace replaces the
ferences, the main aim is to obtain some assurance that the flame. The atomizer consists of a graphite tube about 3 cm
signal measured by the detection system does indeed quan- long, with a 6-mm internal diameter and wall thickness of
titatively represent the concentration of the analyte in any about 1 mm. Two different graphite tube designs are shown
particular matrix. We try to calibrate the instrument with in Fig. 13. The tube is fixed between two electrodes and
a known concentration of the analyte in a deionized wa- is subjected to a low-voltage (up to 12 V), high electrical
ter matrix and then compare directly the signal from a real current. The power supply can be programmed to heat the
sample with the calibration curve. The real sample compo- furnace to several pre-selected temperatures. The furnace
nents in the matrix, such as phosphate in a biological ma- has an inert gas passing through and around it to min-
trix during the determination of calcium, cause the signal imize the ingress of oxygen from the air, which would
to be smaller (a depression) or sometimes larger than the lead to rapid oxidative degradation of the furnace tube.
signal obtained from the same concentration of analyte in Five to 100 µl of the sample solution are placed in the
the calibration solution. One way to circumvent chemical furnace and dried at 100◦ C. The sample is then pyrolyzed
interferences is to use releasing agents. Lanthanum chlo- to break down or volatilize the matrix component. The
ride, which is usually added (1% weight/volume) to all pyrolysis temperature is set high enough without uninten-
sample solutions, is a good example of a releasing agent. tionally vaporizing the analyte (about 400–1200◦ C). After
The lanthanum oxide and phosphate compounds are often pyrolysis, the furnace is rapidly heated to the atomization
more stable than the same compounds of other metals. In temperature to volatilize the analyte into the atom cell. The
the flame, the lanthanum oxide or phosphate forms in pref- atomization temperature can be anywhere in the range of
erence to the analyte oxide or phosphate and thus releases a few hundred to about 2700◦ C. The analyte then absorbs
the analyte atom. Other releasing agents that have been light from the hollow cathode lamp to give the atomic ab-
found to work in many situations are strontium chloride sorption. This signal is a transient (Fig. 13E), because the
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
FIGURE 13 Graphite furnace electrothermal atomizers and their corresponding temperature profiles and analytical
signals. (A) and (C) correspond to furnace geometry and the tube temperature profile for transversely heated atomiz-
ers. (B) and (D) correspond to furnace geometry and the tube temperature profile for longitudinally heated atomizers.
The appearance time of the analytical signal for sample atomization from the furnace wall or the platform is highlighted
in (E).
sample is pulse vaporized and the atomic vapor diffuses center of the tube, it may condense at the cooler ends of
rapidly out of the tube. the graphite furnace.
Regardless of the furnace design, the sample is usually
vaporized from a small graphite platform that is placed
2. Tube Design and Heating Characteristics
inside the graphite tube (Fig. 13A and B). The sample is
Figure 13 illustrates two popular tube designs for elec- placed on the platform through a small hole in the top
trothermal vaporization. The transversely heated graphite center of the furnace tube. Figure 13E illustrates the dif-
tube (Fig. 13A), also known as an integrated contact ferent atomization profiles expected when the sample is
cuvette, is heated through the electrical contacts at the vaporized either from the furnace wall or the platform.
side of the atomizer tube. This allows for a more uniform This figure helps to illustrate the function of the platform
heating profile along the length of the graphite tube, as and explain why it is a useful tool. The graphite tube takes
shown in Fig. 13C. In longitudinally heated furnaces a finite time to heat up, as indicated by the tube-wall tem-
(Fig. 13B), the atomizer is heated through the electrical perature versus time plot. The vapor temperature in the
contacts at the ends of the graphite tube. Although simpler furnace follows the tube-wall temperature, and if the sam-
in design, these furnaces tend to have a significant thermal ple is placed on the bottom wall of the tube, it heats up at
gradient along the length of the graphite tube (Fig. 13D). the same rate as the tube. In most cases, the sample is va-
Consequently, because the analyte is vaporized at the porized while the tube is still heating up and results in the
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
appearance of the analyte transient quite early in the heat- centration. Because of the fast heating rates employed in
ing cycle. The main problem in the early appearance of the electrothermal vaporization, the transient from the furnace
atomization profile is with the gas-phase chemistry. The lasts for a very short time and it is necessary to have an
sample is vaporized into a cooler environment, which may instrument with relatively fast (ms time-scale) electronics.
not allow for decomposition of small molecules yielding
a sufficient number of free atoms. But, if the sample is de-
6. Physico-Chemical Interferences
posited on a small graphite platform (inside the graphite
tube), it will be vaporized into the hot environment of The processes that lead to breakdown of the matrix, va-
the furnace later than in the wall atomization case. The porization, and complete release of the analyte as a free
platform heats up later than the tube because of its ther- element vary from analyte to analyte and matrix. These
mal mass and little thermal contact with the tube. As a differing conditions are ultimately responsible for the in-
result, the vaporized sample enters a much hotter envi- terferences experienced by the analyte. These effects are
ronment than in the wall atomization case. The platform well studied and the role of ubiquitous species such as car-
often provides a sufficient delay in vaporization so that bon and oxygen are well documented with regard to the
many samples, upon vaporization, see a constant vapor vaporization process. Even though an attempt is made to
temperature even though each matrix may release the an- maintain an oxygen-free environment, the matrix usually
alyte at slightly different times. One analytical scenario contains oxygen. In addition, oxygen ingress through the
in which the platform vaporization may not be suitable is dosing hole has been observed even during the atomiza-
when refractory elements are to be determined. Refrac- tion stage. Compounds of the analyte, such as oxides or
tory elements have a very high vaporization temperature, halides (formed due to the presence of the matrix) may
and platform atomization will result in premature furnace not break down sufficiently to release all the analyte into
failure due to excessively high temperatures. the source radiation.
In a typical atomization cycle, the sample is vaporized Interference effects can often be minimized by the use
with a heating rate of about 1300–1500 K/s. After reaching of chemicals, which are added to the sample to modify
the desired temperature, the furnace is held at a constant the matrix so that interferences are controlled. A matrix
temperature for 3–4 seconds while the absorption pulse is modifier can both modify the matrix so that the analyte is
measured. vaporized relatively late during the atomization stage and
allow a higher pyrolysis temperature to be used without
3. Pyrolytic Coating of the Graphite Tube premature release of the analyte. Matrix compounds can
then be broken down during the pyrolysis step so that they
Atomic absorption spectrometry furnace tubes are sup- are less likely to cause interference during the atomization
plied with a hard pyrolytic graphite surface. This helps stage. Some typical matrix modifiers are ascorbic acid,
to reduce the porosity of the tube, which minimizes ad- magnesium nitrate, and palladium solutions. Modifiers are
sorption of hot metal vapors and slows down the rate of added in relatively high amounts (about 250 µg).
oxidation of the graphite and the accompanying mechan-
ical degradation of the furnace tube.
7. Background Correction
4. Inert Gas Requirements Small molecules, such as sodium chloride, that are present
An inert gas is used to flush the furnace during analy- during the atomization stage of a furnace analysis are often
sis. This prevents oxidation of the furnace. Usually an able to absorb or scatter radiation from the light source and
argon flow of about 300 cm3 /min. is maintained through cause a spurious absorption signal. This is called spectral
the furnace during the dry and pyrolysis stages of the anal- band interference. In addition, smoke and other particles
ysis. The flow is stopped or reduced during the atomiza- that are present during atomization can cause scatter of the
tion stage to minimize the gas expulsion from the furnace, radiation. This again causes an apparent spurious absorp-
which would reduce the absorbance signal. tion signals. These signals can be corrected effectively by
the use of one of two different methods of background
correction.
5. The Detection System
Deuterium lamp background correction (two-source
The transient signal is usually measured quantitatively by background correction): Two light sources are used in
evaluating the peak area (integrating) rather than by a mea- a deuterium lamp background corrected instrument. The
surement of peak height. While interference may affect first is the usual hollow cathode lamp, and the second is the
position and height of the absorption peak, the area of the deuterium lamp. The latter emits light over a broad range
transient often does not change for a given analyte con- of wavelengths. The deuterium lamp does not normally
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
give a significant atomic absorption signal from the ana- V. ATOMIC FLUORESCENCE
lyte element; however, it does give a scatter signal from SPECTROMETRY
particles within the furnace and an absorption signal from
molecular species. The modulation chopper is mirrored Atomic fluorescence is an extremely sensitive technique
so that, as the chopper spins, it alternatively irradiates the for determination of elements in samples. We should reit-
furnace by using either the HCL or the deuterium lamp. erate that in atomic fluorescence an external light source
Hence, when the HCL is irradiating the furnace, a signal is used to excite the analyte atoms. An ideal light source
is obtained that is composed of the true analyte absorp- for AFS must be much more intense than a hollow cathode
tion signal plus the spurious background signal. When lamp to achieve improvements in sensitivity. As a result,
the deuterium lamp is irradiating the furnace, only the pulsed hollow cathode lamps and lasers are frequently
spurious background signal is obtained. The subtraction used in AFS measurements. Excitation with a light source
of these two signals gives the actual analyte absorption such as a hollow cathode lamp, which only emits radiation
signal. The deuterium lamp works best for elements that specific for the element of interest, makes AFS virtually
have analytical wavelengths in the ultraviolet region of the completely free from spectral interferences. In addition,
spectrum. For background correction in the visible region AFS is like AES in that a multi-element analysis can be
a tungsten–halide lamp may be used. achieved by putting several light sources around the atom
Zeeman background correction: This method of back- cell, as discussed below.
ground correction makes use of the fact that a magnetic
field is able to split the atomic energy levels of an analyte.
A. Theoretical Background
A magnet is placed around the furnace, with the lines of
force of the field perpendicular to the direction of propa- This discussion assumes that the spectral line width of the
gation of the light beam. In this configuration the atomic light source is narrow relative to the absorption profile of
energy levels of an analyte can be split from the normal the analyte atoms, as illustrated in Fig. 12A. The atoms
situation into three energy levels (σ1 , σ2 , and π). The π absorb light from the source, and some of the energy is
energy level absorbs only light of a particular polariza- re-emitted as fluorescence, while various collisional pro-
tion, which is a natural consequence of the splitting of the cesses in the atom cell deplete the remainder of energy.
energy levels. The light source is deliberately polarized The ratio of the amount of light absorbed to that emitted
so that no light can be absorbed by the π energy level. is called the quantum efficiency (Y ) and is ideally equal to
Then two measurements are made. First, the magnetic one. In its simplest form, the equation for fluorescence ra-
field is switched off, and a measurement of signal plus diant power ( f ) resembles general expressions in atomic
background (A) is obtained. In this case, the light-source absorption (because for optically unsaturated systems the
polarization is irrelevant because atoms that are not in a fluorescence signal is a function of the initial source radi-
magnetic field absorb light of any polarization. Second, ant power, 0 ):
the magnetic field is switched on, causing the atoms to
f = 0 (1 − e−kl ) (7)
split into the various components. When the field is on,
no analyte absorption of light occurs, because the light In the above equation, k is the absorption coefficient and
source is not of the correct depolarization even though l is the optical path length for the atom cell. Therefore,
it clearly is still at the correct wavelength. The σ compo- in AFS the signal size is directly proportional to both the
nents are not involved because they are at the wrong wave- light-source intensity and the atom concentration. Calibra-
length to absorb energy from the light source. Therefore, tion curves for AFS with HCL excitation are linear with
the signal obtained (B) results only from light absorbed or a slope of 1 (on a logarithmic plot) at low concentrations
scattered by background species and particles. The simple and bend back towards the concentration axis with a lim-
subtraction (A–B) then gives the background-free analyte iting slope of –0.5 at high concentrations. The curvature
signal. This method has been proven to give reliable and at high concentration is related to self-absorption in the
accurate background correction for all elements that are atom cell.
normally analyzed by AAS. It is particularly easy to ap-
ply because only one light source is used which removes
1. Light Source
many alignment problems that causes difficulties in a two-
source system. The background measurement is made at The multi-element capability of AFS is realized by use of
exactly the same wavelength as the source and analyte several light sources. The HCLs used for AFS are special
wavelengths. In contrast, two-source correction measures high-intensity versions of the ones used for atomic absorp-
the background over the range of wavelengths that exit the tion and up to 12 of them can be arranged in a practical
monochromator. experimental arrangement. Lasers are also frequently used
P1: FYK/LSX Revised Pages P2: FWQ/FPW QC: FYD
Encyclopedia of Physical Science and Technology en001d42 April 28, 2001 15:9
The linear dynamic range of an ICP–MS instrument is lyzers. Some common isobaric interferences observed in
typically about 4–6 orders of magnitude, while the sensi- ICP–MS are listed in Table III.
tivity is often 3 orders of magnitude better than the ICP–
AES technique. While the detection limits for many el-
ements with ICP–MS are better or comparable to AFS VII. COMPARISONS OF THE ATOMIC
and graphite furnace atomic absorption spectrometry, the SPECTROMETRIC TECHNIQUES
simultaneous multi-elemental detection afforded by ICP–
MS clearly places this technique in the lead for analytical It is unlikely that any one of the above analytical tech-
utility and versatility. niques is capable of replacing all other methodologies.
Interferences in ICP–MS are similar to chemical inter- Some techniques, however, have all but disappeared (e.g.,
ferences observed for ICP–OES. In addition, a number flame atomic emission), and others have been employed
of isobaric interferences caused by molecules with sim- for unique projects (i.e., LEI, for selective isotopic detec-
ilar mass-to-charge ratios of some analyte ions also in- tion). Frequently, a technique is selected based on the spe-
fluence the analytical results. These interferences can be cific analytical needs and instrument availability. A gen-
minimized by judicious use of acids in sample matrices, eral comparison of these techniques is outlined in Table IV.
optimized plasma power settings, use of different sam- The information provided in this table is a gross oversim-
ple introduction schemes, and use of altermate mass ana- plification of many experimental factors and it should be
I. Introduction
II. The Auger Process and Auger
Electron Spectroscopy
III. Applications in Materials Science
IV. Conclusions
787
P1: GKD/LOW Revised Pages P2: FPP
Encyclopedia of Physical Science and Technology en001f-44 May 7, 2001 15:8
Let us first consider the external surfaces. The most in which results could be misinterpreted. One is that the
common types of experiments that are performed are those surface will be covered with enough oxygen and carbon
in which the surface is cleaned and then heated to allow from the environment to mask the surface species that may
segregation. The heating allows solid state diffusion to oc- have originally caused a problem. The second is that con-
cur, and one finds that with time the composition of the taminants from handling of the material may show up and
surface can be significantly altered as a result of this seg- mislead the researcher in his or her effort to find the root
regation. Figure 3 shows an example of such changes in cause of a problem.
composition. The base material is an Fe-Cr-Ni 304 stain- In contrast, the direct studies of segregation to internal
less steel that has bulk concentrations of phosphorus, sul- interfaces have been extremely important in the solution
fur, and nitrogen that are below 300 ppm. However, as of many engineering problems. The basic segregation pro-
seen in Fig. 3, the concentration of these elements on the cess is similar to that which occurs for surface segregation.
surface can be quite significant. The solid is heated to a sufficiently high temperature and
The most common use of these types of studies has been solid state diffusion begins to occur. However, there is an
to simply detect elements present on the surface during a important difference in the way grain boundary segrega-
surface science experiment or to determine what elements tion is detected compared with the way surface segregation
will tend to segregate to interfaces when the materials are is studied. For the surfaces, the change in the composition
heated. Some investigations have attempted to use surface is simply monitored by impinging the electron beam on
segregation to predict the segregation that will occur to the surface while it is being heated. However, grain bound-
internal interfaces such as grain boundaries, but there are aries are internal interfaces and they must be exposed after
numerous studies which show that the differences between the segregation has occurred. What is done is to take ad-
grain boundary segregation and surface segregation are vantage of the fact that segregation often weakens the grain
great enough that such studies can, at best, give qualitative boundaries. The samples can then be fractured apart along
trends. their grain boundaries in the high-vacuum spectrometer,
One of the concerns of this type of experiment is that it and this fracture surface becomes the one that is analyzed
must be done in high vacuum, whereas most applications by Auger electron spectroscopy. However, the fracture
are in an environment of some type. Thus, one must be may not be completely intergranular, and this application
careful in the application of the results of these experi- is one where the development of scanning Auger electron
ments to situations where aggressive media are the cause spectrometers has provided a great advantage. One can
of the problem, such as would be the case for corrosion obtain an image of the fracture surface and select regions
or high-temperature oxidation. Another word of caution on it that are intergranular and analyze only them. One can
must be given about the analysis of surfaces that have been then compare the compositions of these regions with those
exposed to the atmosphere. Because of the extreme sur- from regions where the fracture goes through the grains.
face sensitivity of Auger spectroscopy, there are two ways The reason why Auger analysis of grain boundaries has
been so important is because the segregation often controls
metallurgical properties. We have already mentioned that
segregation often makes the boundaries preferred fracture
paths, and, in general, intergranular fracture is a very low
energy, undesirable fracture mode. Prior to the develop-
ment of Auger electron spectroscopy, it was suspected
that segregation of impurity elements, such as phospho-
rus, tin, and antimony, caused the embrittlement. Figure 4
shows the change in the ductile to brittle transition temper-
ature of Ni-Cr steel plotted as a function of the amount of
segregation of phosphorus and antimony. This transition
temperature measures the brittleness in steels, and as the
transition temperature increases, the brittleness of the steel
increases. This plot shows that there is a good correlation
between brittleness and segregation.
Another example of the effect of segregation can be
FIGURE 3 The segregation kinetics at 650◦ C for sulfur, phos-
to affect the response of the material to an external envi-
phorus, and nitrogen in a commercial 304 stainless steel alloy. In
this figure, changes in the peak height ratio are used to express ronment. Numerous studies have shown that phosphorus
changes in the composition of the surface. [From Briant, C. L., and segregation to grain boundaries of steel enhances its cor-
Mulford, R. A. (1982). Metall. Trans. A, 13A, 745.] rosion nitrate environments, nitric acid, picric acid, and
P1: GKD/LOW Revised Pages P2: FPP
Encyclopedia of Physical Science and Technology en001f-44 May 7, 2001 15:8
355
P1: LDK Revised Pages
Encyclopedia of Physical Science and Technology EN002E-79 May 17, 2001 20:28
conventional electrophoretic techniques have been dem- with an electrolyte. A schematic of a CE system is pre-
onstrated: zone electrophoresis, isotachophoresis, isoelec- sented in Fig. 1. The capillary ends are immersed in
tric focusing, and sieving separations. Unlike conven- electrolyte-filled reservoirs containing electrodes con-
tional electrophoresis, however, the separations are nected to a high-voltage power supply. A sample is in-
usually performed in free solution without the require- troduced at one end of the capillary (the inlet) and ana-
ment for a casting a gel. As in HPLC, detection is lytes are separated as they migrate through the capillary
accomplished as the separation progresses, with resolved towards the outlet end. As separated components migrate
zones producing an electronic signal as they migrate past through a section at the far end of the capillary, they are
the monitor point of a concentration-sensitive (e.g., ul- sensed by a detector, and an electronic signal is sent to a
traviolet absorbance or fluorescence) detector; therefore, recording device.
the need for staining and destaining is eliminated. Data As in conventional gel electrophoresis, the basis of sep-
presentation and interpretation is also similar to HPLC; arations in capillary electrophoresis is differential migra-
the output (peaks on a baseline) can be displayed as an tion of analytes in an applied electric field. The elec-
electropherogram and integrated to produce quantitative trophoretic migration velocity ν E P (in units of cm/sec)
information in the form of peak area or height. In most will depend upon the magnitude of the electric field, E,
commercial CE instruments, a single sample is injected and electrophoretic mobility, µEP , of the analyte:
at the inlet of the capillary and multiple samples are
νEP = µEP E (1)
analyzed in serial fashion. This contrasts to conventional
electrophoresis in which multiple samples are frequently In a medium of a given pH, mobility of an analyte is given
run in parallel as lanes on the same gel. This limitation by the expression:
of CE in sample throughput is mitigated by the ability
µEP = q/6π ηr (2)
to process samples automatically using an autosampler.
Several multicapillary systems have been introduced to where q is the net charge of the analyte, η is the vis-
increase sample throughput; most of these are designed cosity of the medium, and r is the Stoke’s radius of the
specifically for high-throughput DNA sequencing. analyte. Since the Stoke’s radius is related to molecular
Compared to its elder cousins, CE is characterized mass and therefore to frictional drag, mobility will in-
by high resolving power, sometimes higher than elec- crease inversely with molecular weight and directly with
trophoresis or HPLC. The use of narrow-bore capillar- increasing charge. Mobility in units of cm2 sec−1 V−1 can
ies with excellent heat-dissipation properties enables the be determined from the migration time and field strength
use of very high field strengths (sometimes in excess by:
of 1000 volts/centimeter), which decreases analysis time
µEP = (I /t)(L/V ) (3)
and minimizes band diffusion. When separations are per-
formed in the presence of electroosmotic flow (EOF), the where I is the distance from the inlet to the detection
plug-flow characteristics of EOF also contribute to high point (termed the effective length of the capillary), t is the
efficiency. In contrast, the laminar flow properties of liq- time required for the analyte to reach the detection point
uid chromatography increase resistance to mass transfer, (migration time), L is the total length of the capillary, and
reducing separation efficiency. V is the applied voltage.
Because of its many advantages, CE is finding increas- In contrast to most forms of gel electrophoresis, the ve-
ing use as a tool in analytical chemistry. In some cases it locity of an analyte in capillary electrophoresis will also
may replace HPLC and electrophoresis, but more often it depend upon the rate of electroosmotic flow. This phe-
is used in conjunction with existing techniques, providing nomenon is observed when an electric field is applied to
a different separation selectivity, improved quantitation, a solution contained in a capillary with fixed charges on
or automated analysis. It was the adaptation of CE tech- the capillary wall. Typically, charged sites are created by
nology in the design of 96-channel DNA sequencers that ionization of silanol (SiOH) groups on the inner surface
enabled a draft of the human genome sequence to be com- of the fused silica. Silanols are weakly acidic and ionize
pleted in less than two years. at pH values above about 3. Hydrated cations in solution
associate with ionized SiO− groups to form an electrical
double layer: a static inner Stern layer close to the surface
I. PRINCIPLES OF CAPILLARY and a mobile outer layer known as the Helmholtz plane.
ELECTROPHORESIS Upon application of the field, hydrated cations in the outer
layer move toward the cathode, creating a net flow of the
As the name implies, capillary electrophoresis separates bulk liquid in the capillary in the same direction (Fig. 2).
species within the lumen of a small-bore capillary filled The rate of movement is dependent upon the field strength
P1: LDK Revised Pages
Encyclopedia of Physical Science and Technology EN002E-79 May 17, 2001 20:28
FIGURE 1 General schematic of a capillary electrophoresis system. Capillary internal diameters can range from 25
to 150 µm, and capillary lengths can range from 20 cm to over 1 meter. In some cases, sample may be injected at the
capillary outlet, and the short segment between the outlet and the detection window serves as the separation path.
FIGURE 2 Electroosmotic flow in an uncoated fused silica capillary. The negative charges represent ionized silanol
groups on the silica surface, and the positive charges represent hydrated electrolyte cations.
P1: LDK Revised Pages
Encyclopedia of Physical Science and Technology EN002E-79 May 17, 2001 20:28
length should not exceed 5% of the total capillary length. The small internal diameters of fused silica capillar-
Sample zone sharpening can be achieved by preparing ies have been the key to high-resolution CE separations.
the sample in an electrolyte of lower conductivity than The smaller the capillary bore, the greater the surface-to-
the analysis buffer. Under these conditions, there is a dis- volume ratio and the more efficiently heat is removed from
continuity in field strength at the sample:buffer boundary the tube. The optimal capillary inside diameter (i.d.) for
such that ions migrating rapidly from a region of higher most applications is 50 to 75 µm. Capillaries with smaller
field strength become focused at the boundary. This fo- diameters are subject to plugging, and the high surface-
cusing or stacking effect not only produces narrow zones to-volume ratio increases the risk of adsorption. Capillar-
for increased resolution, but also increases zone concen- ies with larger internal diameters may exhibit significant
tration for enhanced sensitivity. Sensitivity enhancement loss in resolution due to thermal effects, and may require
is directly proportional to the ratio of sample to buffer operation at low field strengths or with low-conductivity
conductivity. Sample zone sharpening can also be accom- buffers.
plished by transient isotachophoretic preconcentration, as
discussed below.
D. Adsorption (hads )
B. Diffusion and Electrodispersion (hdiff +cond ) Wall interactions have been the greatest obstacle to achiev-
ing satisfactory resolution and reproducibility in capillary
Band broadening due to axial diffusion should be reduced electrophoresis, particularly in zone electrophoresis and
by shortening analysis time (i.e., by operating at high field isoelectric focusing. Approaches to minimizing or elimi-
strengths with short capillaries). On the other hand, higher nating adsorption include use of buffer additives and dy-
field strengths will generate more Joule heat, which in- namic or covalently-coupled capillary coatings.
creases diffusion rates. Electrodispersive band broaden-
ing arises from conductivity differences between the zone
and the background electrolyte. If this difference is large,
diffusion at one boundary of the zone is neglible due to III. CAPILLARY ELECTROPHORESIS
the zone-sharpening effect caused by the discontinuity in INSTRUMENTATION
field strength. In this case, the peak will be asymmetric;
that is, it will exhibit fronting or tailing depending on A variety of commercial CE instruments are available,
whether conductivity of the zone is greater or less than from simple modular systems consisting of a power sup-
that of the background electrolyte. This phenomenon is ply, detector, and injection device to fully integrated au-
readily observable when analyzing highly charged small tomated systems under computer control. This discussion
molecules such as inorganic ions, and satisfactory peak will focus on the features of automated CE instrumenta-
shapes can only be obtained by carefully matching sam- tion.
ple and electrolyte conductivities. The effect is much less
noticeable in other CE separations, except at high sample
A. Power Supply
concentrations.
Power supplies capable of delivering constant voltage with
high precision up to 30 kV are standard throughout the
C. Joule Heating (hjoule )
industry. Most systems offer, in addition, constant cur-
When an electric field is applied to a capillary containing rent operation at up to 300 µA; constant current operation
an electrolyte, Joule heat is generated uniformly across may be desirable in systems without adequate tempera-
the circumference of the tube. Since heat can only be re- ture control. High voltage is applied at the capillary inlet,
moved at the margin of the tube, a temperature gradient with the outlet (detector) at ground potential. When sepa-
exists across the radius of the tube. As noted in Eq. (2), rations are performed in uncoated fused silica capillaries
mobility is inversely related to viscosity, which decreases in the presence of EOF, the inlet (high-voltage) electrode
with temperature. Mobility increases approximately 2.5% is the anode and EOF carries analytes towards the cathode
for each degree rise in temperature. Therefore, the temper- (“normal” polarity). Separations performed in coated cap-
ature gradient creates a mobility gradient across the tube illaries without EOF may require “reversed” polarity. In
radius which contributes to band broadening. It should some separations, the direction of EOF may be reversed
also be noted that, in addition to this mobility gradient, by using capillaries coated with positively charged poly-
Joule heat can compromise resolution by increasing dif- mers or by using osmotic flow-modifying additives in the
fusion and convection; these are minimized by the use of background electrolyte. In these cases, reversed polarity
viscous buffers. is also used.
P1: LDK Revised Pages
Encyclopedia of Physical Science and Technology EN002E-79 May 17, 2001 20:28
employ UV or UV–Vis absorbance as the primary mode of through the capillary is dispersed by a grating onto an array
detection. The simplest approach is the use of line-source of photodiodes, each of which samples a narrow spectral
lamps or continuum-source lamps with wavelength selec- range. In fast-scanning detectors, monochromatic light is
tion by filters. Better flexibility is obtained using a con- collected from the source using a movable grating and slit
tinuum (e.g., deuterium lamp) source with wavelength se- assembly and directed to the capillary; light transmitted by
lection by a grating monochromator. the capillary is detected by a single photodiode. Scanning
All commercial CE absorbance detectors employ on- is accomplished by rapidly rotating the grating through an
tube detection, in which a section of the capillary itself angle to “slew” across the desired spectral range. Photodi-
is used as the detection cell. This permits detection of ode array detectors are more sensitive in full-scan mode,
separated zones with no loss in resolution. Most capil- but the high light intensity can damage coated or gel-filled
laries used for CE are coated with with a polymer (usu- capillaries.
ally polyimide) which protects the fused silica capillary
and provides it with mechanical stability. Since the poly-
2. Fluorescence
mer is not optically transparent, it must be removed from
the detection segment to form a “window,” and this win- Fluorescence detection offers the possibility of high sen-
dow must be accurately positioned in the optical path to sitivity and, in the case of complex samples, improved
achieve good sensitivity. This segment of bare capillary selectivity. However, this mode of detection requires that
is very fragile and is subject to breakage during manipu- the analyte exhibit native fluorescence or contain a group
lation and installation of the capillary. Capillaries with a to which a fluorophore can be attached by chemical deriva-
UV-transparent coating are commercially available which tization. The number of compounds that fall into the for-
eliminate this problem; however, the coating is not resis- mer category are small, and while many analytes contain
tant to some coolants (e.g., fluorinated hydrocarbons) used derivatizable groups (e.g., amino, carboxyl, hydroxyl),
in liquid-cooled CE systems. most derivatization chemistries are limited by one or more
In on-tube detection, the internal diameter of the cap- disadvantages (slow reaction kinetics, complicated reac-
illary forms the detection light path. In accordance with tion or cleanup conditions, poor yields, interference by
Beer’s law, the sensitivity of a concentration-sensitive de- matrix components, derivative instability, interference by
tector is a direct function of the length of the light path. reaction side products or unreacted derivatizing agent).
Therefore, in comparison to an HPLC detector with a When compared to fluorescence detectors for HPLC,
1-cm path length, detector signal strength should be re- the design of a fluorescence detector for CE presents some
duced 200-fold in a CE system equipped with a 50-µm- technical problems. In order to obtain acceptable sensitiv-
i.d. capillary. Concentration sensitivity can be improved ity, it is necessary to focus sufficient excitation light on the
by employing focusing lenses to collect light at the capil- capillary lumen. This is difficult to achieve with a conven-
lary lumen, by detecting at low wavelengths (where most tional light source, but is easily accomplished using a laser.
analytes have greater absorbance), and by using sample- The most popular sources for laser-induced fluorescence
focusing techniques during the injection process. How- (LIF) detection are the argon ion and helium–neon lasers,
ever, even under ideal conditions, the concentration limit which are stable and relatively inexpensive. These lasers
of detection (CLOD) is about 10−6 M. Sensitivity in op- produce emission lines close to the excitation wavelengths
tical detectors can be improved by the use of extended- for several common fluorophores. The CLOD for a laser-
pathlength capillaries. These include capillaries with in- based fluorescence detector can be as low as 10−12 M.
creased diameters in the window segment (“bubble cell” The most successful applications of CE–LIF detection
capillaries) or in-line cells with a section of capillary po- have been in the analysis of carbohydrates and nucleic
sitioned axially along the light path (“Z-cells”). acids. Carbohydrates can be derivatized with agents such
Several commercial CE systems incorporate scanning as aminopyrene trisulfonic acid, which introduces a flu-
absorbance detectors. Scanning detection enables on-the- orophore and also confers electrophoretic mobility to the
fly acquisition of spectra as analytes migrate through the analyte. Double-stranded nucleic acids can be detected
detection point; this information can assist in the identi- using intercalating dyes and by covalent attachment of
fication of peaks based on spectral patterns, in detection nucleotides carrying fluorophores. The fluorescently la-
of peak impurities by variation in spectral profiles across beled primers or chain terminating nucleotides which are
a peak, or in determation of the absorbance maximum of widely used in automated slab gel-based DNA sequencers
an unknown compound. Two different designs are used are also used in capillary-based sequencers. In one multi-
to accomplish scanning detection in CE instruments. In capillary sequencer, a novel sheath flow system is used for
photodiode array (PDA) detectors, the capillary is illu- LIF excitation of separated bands with imaging of emitted
minated with full-spectrum source light; the light passing fluorescence on a charge-coupled device (CCD) detector.
P1: LDK Revised Pages
Encyclopedia of Physical Science and Technology EN002E-79 May 17, 2001 20:28
Mobility and selectivity in CZE are most profoundly af- ity monitoring of protein therapeutics; clinical analysis of
fected by analyte charge, and selection of the electrolyte serum proteins; and quality control of wines. The CZE
pH is the most effective method of controlling a CZE sep- separation of peptides originating from tryptic digestion
aration. A wide variety of buffers have been employed of cytochrome C is shown in Fig. 4. In addition to these
in CZE, and a buffer is selected to provide good buffer- applications, three important variations of CZE are of par-
ing capacity at the desired pH, low UV absorbance, and ticular note.
low conductivity. In addition to the buffer, other compo-
nents may be added to the electrolyte to control EOF,
1. Capillary Ion Analysis
reduce solute–wall interactions, or to modulate the mo-
bility or solubility of an analyte. Additives for CZE in- Inorganic ions are a particular challenge for CZE because
clude neutral salts, organic amines, surfactants, organic of their high mobilities and lack of UV absorbance. Inor-
solvents, and chiral selectors. Secondary equilibria intro- ganic ions can be detected in the UV region using indi-
duced by additive–analyte interactions are very important rect detection. In this technique, a UV-absorbing species
for achieving resolution in CZE. is used as the background electrolyte, and analytes are
detected as negative peaks. High-mobility anions are an-
alyzed using reversed-charge CZE. Capillary ion analysis
B. Applications of Capillary is routinely used in the analysis of environmental samples,
Zone Electrophoresis industrial brines, and food products.
The great flexibility of CZE has enabled its successful
application to a wide variety of analytes ranging from in-
2. Affinity Capillary Electrophoresis
organic ions to macromolecules. A very incomplete listing
of these includes analysis of ionic drugs and metabolites Affinity capillary electrophoresis (ACE) is used to char-
in pharmaceutical formulations and body fluids; analysis acterize receptor–ligand interactions. The technique de-
of foods, food supplements, and nutriceuticals; analysis pends upon a shift in the electrophoretic mobility of the
of milk and cereal proteins; quality control and stabil- receptor upon complexation with a charged ligand. The
FIGURE 4 Separation of a tryptic digest of cytochrome C by capillary zone electrophoresis. Separation was performed
in a neutral-coated capillary using 100-mM sodium phosphate (pH 2.5).
P1: LDK Revised Pages
Encyclopedia of Physical Science and Technology EN002E-79 May 17, 2001 20:28
FIGURE 5 Separation of enantiomers of epinephrine and norepinephrine. Separation was performed in a neutral-
coated capillary using 100-mM sodium phosphate (pH 2.5) and β-cyclodextrin.
distribution of bound vs. free receptor at various ligand and an acidic anolyte solution. The technique is almost
concentrations can be used to determine binding kinetics. universally applied to separation of proteins, which are
Applications of ACE include determination of enzyme– resolved according to their isoelectric points. Capillary
effector binding, antibody–antigen binding, and DNA– isoelectric focusing (CIEF) is similar in concept to con-
carcinogen interactions. ventional gel IEF; carrier ampholytes form a pH gradient
under the influence of an electric field and proteins migrate
electrophoretically to the point in the gradient at which
3. Chiral Analysis
they are isoelectric. In both IEF formats, very high resolu-
Enantiomers can be resolved by incorporating a chi- tion is achieved due to the electrophoretic focusing effect,
ral selector in the background electrolyte. Typical chi- which counters protein diffusion away from the zone at
ral selectors are cyclodextrins, proteins such as α1 -acid equilibrium. In contrast to gel IEF (in which focused pro-
glycoprotein, and macrocyclic antibiotics such as van- teins are fixed and detected by staining in situ within the
comycin. Charged enantiomers may be separated by dif- gel), capillary IEF with on-tube detection requires mobi-
ferential formation of inclusion complexes with neutral lization of focused protein zones past the detection point.
cyclodextrins, while uncharged enantiomers may be re-
solved using a charged cyclodextrin such as cyclodextrin-
A. Separation Conditions for Capillary
sulfobuytlether. Separation of enantiomers of epinephrine
Isoelectric Focusing
and norepinephrine using a cyclodextrin chiral selector is
shown in Fig. 5. Capillary IEF can be carried out as a single- or two-step
process. In single-step CIEF, a mixture of sample proteins
and ampholytes is introduced as a plug at the capillary
VI. CAPILLARY ISOELECTRIC FOCUSING inlet, and an electric field and a mobilizing force are ap-
plied simultaneously. Formation of the pH gradient and
Isoelectric focusing (IEF) is an equilibrium technique in protein focusing occur while the zone is in transit from
which amphoteric compounds are focused in a stable pH the capillary inlet to the detection point. The mobilizing
gradient at the point where they have no net charge. The pH force can be pressure, vacuum, gravity, or electroosmotic.
gradient is contained between a basic catholyte solution Although single-step CIEF is simple in concept, achieving
P1: LDK Revised Pages
Encyclopedia of Physical Science and Technology EN002E-79 May 17, 2001 20:28
reproducible migration patterns can be difficult, particu- pholyte blends are used which generate gradients of over
larly when using electroosmotic mobilization. 8 pH units (e.g., pH 3–10); for high-resolution separation
In two-step CIEF, focusing and mobilization are per- of a few proteins, narrow-range ampholyte blends are used
formed separately. Initially, the entire capillary is filled to generate gradients of 1 to 2 pH units.
with a mixture of sample and ampholytes, high voltage
is applied, and the focusing progress is monitored by the B. Applications of Capillary
decrease in operating current. Once equilibrium has been Isoelectric Focusing
established, the mobilization step is begun. The mobilizing
force may be pressure, vacuum, gravity, or electrophoretic. The high resolving power of CIEF enables separation of
In all cases, the electric field is maintained to prevent defo- protein species with small differences in isoelectric points.
cusing of zones during mobilization. Electrophoretic mo- Capillary IEF has been applied to the analysis of protein
bilization is accomplished by changing the composition of glycoforms, protein clips, and deamidation products. The
the anolyte or catholyte, usually by the addition of a salt, technique has also been used for analysis of hemoglobin
buffer, or zwitterion solution. This change in composi- variants and glycated hemoglobins. It provides a rapid and
tion produces a progressive shift in pH down the capillary, convenient method for estimating the isoelectric point of
causing electrophoretic migration of focused proteins past an unknown protein. On-line CIEF–MS is being inves-
the detection point (Fig. 6). In two-step CIEF, the presence tigated as an instrumental counterpart to analytical two-
of EOF disrupts the CIEF process, resulting in loss of reso- dimensional gel electrophoresis in proteomic studies.
lution and reproducibility. For this reason, two-step CIEF
is performed in capillaries coated to reduce or eliminate VII. CAPILLARY SIEVING TECHNIQUES
EOF. Because the entire capillary is filled with sample in
two-step CIEF, zone concentrations at equilibrium will be Sieving techniques are required for separation of species
higher and detection sensitivity will be greater compared which have no differences in mass-to-charge ratio such as
to single-step CIEF. nucleic acids and sodium dodecylsulfate (SDS)–protein
Resolution in CIEF is governed primarily by the com- complexes. Sieving systems include cross-linked or linear
position of the ampholyte mixture. For screening and for polymeric gels cast in the capillary or replaceable polymer
analysis of complex mixtures of proteins, wide-range am- solutions.
FIGURE 6 Separation of proteins by two-step CIEF using electrophoretic mobilization. The ampholytes generated
a gradient from pH 3 to 10 after a focusing time of 300 sec in a neutral-coated capillary. Cathodic mobilization was
initiated by replacing the catholyte (40-mM NaOH) with an alkaline zwitterion solution. pI, Isoelectric point.
P1: LDK Revised Pages
Encyclopedia of Physical Science and Technology EN002E-79 May 17, 2001 20:28
FIGURE 7 Separation of SDS–protein complexes by polymer-sieving CE. Separation was obtained in an uncoated
capillary using a commercial polymer-sieving applications kit developed for analysis of SDS–protein complexes (Bio-
Rad Laboratories; Hercules, CA).
A. Capillary Gel Electrophoresis is the ability to closely control the porosity by changing
the ratio of monomer to cross-linker. Polyacrylamide gel-
Polyacrylamide gels can be cast within a capillary using
filled capillaries are incompatible with absorbance detec-
the same chemistry employed for casting conventional
tion at short UV wavelengths. Capillary gel electrophore-
polyacrylamide gels. However, preparation and use of
sis (CGE) has been used for high resolution of synthetic
gel-filled capillaries is complicated by bubble formation
antisense nucleotides such as phosphothioate nucleotides.
caused by volume changes during polymerization, and by
extrusion of the gel from the capillary. To circumvent this,
gels are covalently attached to the capillary wall and poly-
B. Replaceable Polymer Solutions
merized in a segmental fashion. Care must be exercised
in their use to avoid bubble formation and contamina- Sieving separations can be accomplished using solu-
tion of the gel bed. An advantage of gel-filled capillaries tions of linear or branched hydrophilic polymers. These
P1: LDK Revised Pages
Encyclopedia of Physical Science and Technology EN002E-79 May 17, 2001 20:28
FIGURE 8 Separation of synthetic oligonucleotides by polymer-sieving CE. Separation was obtained in a neutral-
coated capillary uisng a commercial polymer-sieving applications kit developed for analysis of oligonucleotides (Bio-
Rad Laboratories; Hercules, CA).
polymers have low UV absorbance at short wavelengths, ity toward the detection point where they are detected
and can be replenished between analyses by pressure. as steps with zone length proportional to concentration.
Polymer solutions act by forming entangled networks When UV-transparent spacers are added to the sample, ITP
above a certain polymer concentration, which mimics the zones may appear as isolated peaks and the detector trace
sieving properties of cross-linked gels. Polymers used for will resemble a CZE electropherogram. Isotachophoresis
sieving separations include alkylated celluloses, polyethy- is rarely used as a separation method but is occasionally
lene oxide, and dextrans. Replaceable polymer systems used transiently as an on-line preconcentration technique
have been developed for separation of SDS–protein com- prior to CZE separation.
plexes ranging form 10 to 200 kDa (Fig. 7). Replaceable
polymer solutions have found wide use in nucleic acid sep-
arations, with applications including purity determination IX. MICELLAR ELECTROKINETIC
of synthetic oligonucleotides (Fig. 8), separation of DNA CHROMATOGRAPHY
restriction fragments, DNA typing using microsatellite or
short tandem repeat (STR) sequences, mutation analysis of As the name implies, micellar electrokinetic chromatog-
amplified genetic sequences, and, most importantly, DNA raphy (MEKC) is a chromatographic technique in which
sequence analysis. Because of their excellent precision, samples are separated by differential partitioning between
ease of use, and high resolution, polymer sieving systems two phases (Fig. 9). The technique is usually performed in
have virtually replaced gel-filled capillaries for size-based uncoated capillaries under alkaline conditions to generate
separations by capillary electrophoresis. a high electroosmotic flow. The background electrolyte
contains a surfactant at a concentration above its critical
VIII. ISOTACHOPHORESIS
micelle concentration (CMC); surfactant monomers are in flow when a voltage is applied to the system, pumping
equilibrium with micelles. The most widely used MEKC mobile phase through the column. Since EOF is plug-like
system employs sodium dodecylsulfate (SDS) as the sur- rather than laminar in nature, efficiencies in CEC can be
factant. The sulfate groups of SDS are anionic, so both sur- much higher than in HPLC. In contrast to pressure-driven
factant monomers and micelles have electrophoretic mo- capillary liquid chromatography, which requires pressures
bility counter to the direction of EOF. Sample molecules of several thousand psi to pump mobile phase through
will be distributed between the bulk aqueous phase and the the column, CEC generates no column backpressure. Like
micellar phase, depending upon their hydrophobic char- MEKC, CEC is used primarily for small molecules.
acter. Hydrophilic neutral species with no affinity for the
micelle will remain in the aqueous phase and reach the
detector in the time required for EOF to travel the effec- SEE ALSO THE FOLLOWING ARTICLES
tive length of the column. Hydrophobic neutral species
will spend varying amounts of time in the micellar phase ELECTROPHORESIS • LIQUID CHROMATOGRAPHY • MASS
depending on their hydrophobicity, and their migration SPECTROMETRY
will therefore be retarded by the anodically moving mi-
celles. Charged species will display more complex behav-
ior, and their migration will include contributions from BIBLIOGRAPHY
electrophoretic mobility and electrostatic interaction with
the micelles in addition to hydrophobic partitioning. The Camilleri, P., ed. (1993). “Capillary Electrophoresis: Theory and Prac-
selectivity of MEKC can be expanded with the introduc- tice,” CRC Press, Boca Raton, FL.
Deyl, Z., and Struzinsky, R. (1991). J. Chromatogr. 569, 63.
tion of chiral selectors to the system. MEKC is used al- Grossman, P. D., and Colburn, J. C., eds. (1992). “Capillary Elec-
most exclusively for small molecules such as drugs and trophoresis: Theory and Practice,” Academic Press, San Diego, CA.
metabolites, pesticides, herbicides, vitamins, etc. Guzman, N. A. (1993). “Capillary Electrophoresis Technology,” Dekker,
New York.
Kuhr, W. G. (1990). Anal. Chem. 62, 403R.
Kuhr, W. G., and Monnig, C. A. (1992). Anal. Chem. 64, 389R.
X. CAPILLARY ELECTROKINETIC Landers, J. P., ed. (1996). “Handbook of Capillary Electrophoresis,” 2nd
CHROMATOGRAPHY ed., CRC Press, Boca Raton, FL.
Li, S. F. Y. (1992). “Capillary Electrophoresis: Principles, Practice, and
Like MEKC, capillary electrokinetic chromatography Applications,” Vol. 52, Elsevier Science, Amsterdam.
Lunte, S. M., and Radzik, D. M., eds. (1996). “Pharmaceutical and
(CEC) is a chromatographic technique performed with CE
Biomedical Applications of Capillary Electrophoresis: Progress in
instrumentation. It employs fused silica capillaries packed Pharmaceutical and Biomedical Analysis,” Vol., Elsevier Science,
with 1.5- to 5-µm microparticulate porous silica beads, Oxford.
usually derivatized with a hydrophobic ligand such as C18. Monnig, C. A., and Kennedy, R. T. (1994). Anal. Chem. 66, 280R.
Mobile phases are similar to those used for conventional Wehr, T., Rodriguez-Dı́az, R., and Zhu, M. (1998). “Capillary Elec-
reversed-phase HPLC (e.g., mixtures of aqueous buffers trophoresis of Proteins,” Dekker, New York.
Weinberger, R. (1992). “Practical Capillary Electrophoresis,” Academic
and an organic modifier such as acetonitrile). The silica Press, Boston, MA.
surface of the derivatized beads has sufficient densities of Vindevogel, V. (1992). “Introduction to Micellar Electrokinetic Chro-
ionized silanol groups to generate a high electroosmotic matography,” Hüthig, Heidelberg.
P1: GRB Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry
Donald T. Sawyer
Texas A&M University
I. Fundamentals
II. Electrode Potentials and Potentiometry
III. Controlled-Potential Electrolysis
and Voltammetry
IV. Electron-Transfer Processes
V. Electrochemical Characterization of Molecules
VI. Industrial Electrosynthesis
VII. Batteries and Fuel Cells
VIII. Corrosion; Cathodic Protection
161
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
162 Electrochemistry
THE TECHNOLOGY for the interconversion of chem- tween the anode and the cathode), which is the defining
ical energy and electrical energy has been utilized since difference for electrochemistry and electronics, is due to
the mid-19th century. This conversion is accomplished by the incompatibility of electrons and electrolyte solutions.
ionic-current flow in an electrolyte solution between two During the past four decades, the dynamics and mech-
electrodes connected to each other via an external circuit anisms of electron-transfer processes have been studied
with an electrical load or current source. Batteries, fuel via the application of transition-state theory to the kinet-
cells, and corrosion processes convert the energy of chem- ics for electrochemical processes. As a result, the kinetics
ical reactions into electrical energy. Electrolysis, electro- of both the electron-transfer processes (from solid elec-
plating, and some forms of electroanalysis reverse the trode to the solution species) as well as the pre- and post-
direction of conversion, using electrical energy to produce electron-transfer homogeneous processes can be charac-
a net chemical change. The basic principles and quanti- terized quantitatively.
tative relationships (voltage, current, charge conductance, By the use of various transient methods, electrochem-
capacitance, and concentration) for electrochemical phe- istry has found extensive new applications for the study
nomena were empirically elucidated by Michael Faraday of chemical reactions and adsorption phenomena. Thus,
and other European scientists before the discovery of the a combination of thermodynamic and kinetic measure-
electron (J. J. Thompson, 1893) and the development of ments can be utilized to characterize the chemistry of het-
chemical thermodynamics (G. N. Lewis, 1923). Building erogeneous electron-transfer reactions. Furthermore, het-
on this foundation, the utilization of electrochemical phe- erogeneous adsorption processes (liquid-solid) have been
nomena for thermodynamic characterization and analysis the subject of intense investigations. The mechanisms of
of molecules and ions (electroanalytical chemistry) be- metal-ion complexation reactions also have been ascer-
gan at the beginning of this century [potentiometry (1920) tained through the use of various electrochemical impulse
and polarography (1930)]. Relationships that describe the techniques.
techniques of potentiometry and polarography derive di- The so-called Renaissance of electrochemistry has
rectly from solution thermodynamics. In the case of po- come about through a combination of modern electronic
larography, there is a further dependence on the diffusion instrumentation and the development of a more molecular-
of ionic species in solution. The latter is the basis of con- based theory implemented with the data processing and
ductivity measurements, another area that traces its ori- computational power of computers. Within the area of
gin to the 19th century. These quantitative relationships physical chemistry, numerous thermodynamic studies of
make it possible to apply electrochemistry to the detailed unstable reaction intermediates have made use of mod-
characterization of chemical species and processes in the ern electrochemistry. In addition, extensive studies of the
solution phase. kinetics of electron-transfer processes in aqueous and
Electrochemistry is the science of electron transfer nonaqueous media have been accomplished. The electro-
across a solution/electrode interface. At the cathode, chemical characterization of adsorption phenomena has
electrons (from the electrode) are transformed within the been of immense benefit to the understanding of catalytic
interface via reaction with ions or molecules to produce processes.
reduced molecules or ions (e.g., H+ −
3 O + e → H· + H2 O; Some of the most exciting applications of electrochem-
− − − −
H2 O + e → H· + HO ; · O2 · + e → O2 ·; CuII (bpy)2+ 2 istry have occurred in the areas of organic and inorganic
+ e− → CuI (bpy)+ − II −
2 ; Fe Cl3 + e → Fe Cl3 ). [Note: Al-
III
chemistry and of biochemistry. The applications have
though the traditional formulation of the hydro- ranged from mechanistic studies to the synthesis of un-
nium ion (H3 O+ ) is pervasive in the chemical literature, stable or exotic species. The control of an oxidation or
the positive charge is equally distributed among the reduction process through electrochemistry is much more
three hydrogens, which prompts the formulation used precise than is possible with chemical reactants. Within the
here (H+ 3 O).] At the anode, molecules or ions (from the area of inorganic chemistry, electrochemistry has been es-
solution) are transformed within the interface to pro- pecially useful for the determination of formulas of coordi-
duce electrons (at the electrode surface) and oxidized nation complexes and the electron-transfer stoichiometry
ions and molecules (e.g., 2 H2 O → H+ −
3 O + HO· + e ; of new organometallic compounds. Electrochemical syn-
II − −
Fe Cl3 → Fe Cl3 + e ). The resultant electrons move
III
thesis is increasingly important to the field of organometal-
from the anode through the wires of the external circuit to lic chemistry.
the cathode as electronic current (amperes; coulombs per During the past 50 years, numerous exciting extensions
second). Within the solution phase the current is carried of electrochemistry to the field of analytical chemistry
by the ions of the supporting electrolyte (positive ions have occurred. A series of selective-ion potentiometric
toward the cathode and negative ions toward the anode). electrodes have been developed, such that most of the
The limitation of ionic current in the solution phase (be- common ionic species can be quantitatively monitored in
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry 163
aqueous solution. A highly effective electrolytic moisture The flux of ox that is reduced at the electrode (vc ) has
analyzer provides continuous online assays for water in the dimensions of moles per unit area (cm2 ) per second,
gases. Another practical development has been the voltam- and
metric membrane electrode for dioxygen (·O2 ·), which re- vc = d(Cox )o /A dt = kc (Cox )o = jc /n F), (7)
sponds linearly to the partial pressure of ·O2 ·, either in the
gas phase or in solution. The use of an immobilized en- where (Cox )o is concentration (mol/cm3 ) at the electrode
zyme (glucose oxidase) on an electrode sensor to assay surface, A is area of the electrode (cm2 ), and jc is net
glucose in blood is another extension of electrochemistry cathodic (reductive) current density (A/cm2 ).
to practical analysis. At equilibrium (E = 0),
(vc )E=0 = (Cox )o [κkT / h] exp − G ‡c E=0 RT ,
(8)
I. FUNDAMENTALS
where κ is the transmission coefficient within the Ac-
Reductive electrochemistry involves the transfer of elec- tivated Complex Theory, k is the Boltzmann constant,
trons from the electrode surface (cathode) into the double- and h is the Planck constant, which, in combination
layer interface. Within the latter, the electrons react with with Eq. (7), gives an expression for the heterogeneous
the most electrophilic component of the interface; e.g., electron-transfer rate constant when E is zero,
H+
3 O(aq) in acidic aqueous solutions and H2 O(aq) in basic (kc )E=0 = [κkT / h] exp − G ‡c E=0 RT . (9)
aqueous solutions. From electrochemistry the respective
standard reduction potentials in water are In turn, combination of these equations gives expressions
for the flux of ox reduction (vc ) and cathodic current den-
e − + H+
3O [H·] + H2 O E ◦ , −2.10 V vs NHE (1) sity ( jc ) for a potential difference E.
−
e + H2 O [H·] + HO − ◦
E , −2.93 V vs NHE (2)
vc = (Cox )o [κkT / h] exp − G ‡c E=0 RT
and in acetonitrile are × exp(−αn a EF/RT )
e −
+ H+
3O [H·] + H2 O ◦
E , −1.58 V vs NHE (3) = (Cox )o (kc )E=0 exp(−αn a EF/RT ); (10)
− − ◦
e + H2 O [H·] + HO E , −3.90 V vs NHE. (4) jc = (Cox )o n F(kc )E=0 exp(−αn a EF/RT ). (11)
The expressions for the reverse anodic process (oxida-
A. Electron-Transfer Dynamics (Kinetics tion) of Eq. (5) (red → ox + n e− ; ka ) follow from similar
and Thermodynamics) arguments:
The rate of electron transfer at an electrode/solution in- va = (Cred )o (ka )E=0 exp[(1 − α)n a EF/RT ], (12)
terface for the direct reduction of an oxidized species
− ja = (Cred )o n F(ka )E=0 exp[(1 − α)n a EF/RT ]. (13)
2 +e →
(ox) to its reduced state (red ) [e.g., CuII (bpy)2+
I +
Cu (bpy)2 ], When the cathodic current density ( jc ) is equal to the an-
kc odic current density ( ja ), the net current flow across the
ox + ne− ka
red, (5) electrode/solution interface is zero and the net flux of ox
is a function of the concentration of the oxidized species and red is zero. For this unique condition (zero net current)
and its heterogeneous electron-transfer rate constant (kc ; the current densities represent the equilibrium exchange
cathodic process). The latter is a function of the potential current density ( jo ),
difference across the electrode/solution interface (E), jc = ja = jo , (14)
which is directly proportional to the activation energy which is associated with the equilibrium potential differ-
for reduction (−G ‡c ). Only a fraction of E is effec- ence, E e . Thus,
tive for accelerating the rate of reduction, which is repre-
sented by a symmetry parameter, α [transfer coefficient, jo = jc = (Cox )o n F(kc )E=0 exp(−αE e F/RT ) (15)
0.0 < α < 1.0 (usually about 0.5)], jo = ja = (Cred )o n F(ka )E=0 exp[(1 − α)E e F/RT ].
G ‡c total = G ‡c E=0 + G ‡c E (16)
= G ‡c E=0 + αn a EF, (6) The difference between E and E e is the activation
overpotential (η),
where n a = number of electrons in the rate limiting step;
usually one. η = E − E e . (17)
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
164 Electrochemistry
These relationships can be combined to give an expres- Marcus/Hush/Gerischer type) are discussed in Bard and
sion for the net current density ( j), which by definition Faulkner (2001).
is equal to ja − jc in terms of the activation overpoten-
tial (substitute η + E e = E), which is referred to as the
Butler-Volmer equation II. ELECTRODE POTENTIALS
AND POTENTIOMETRY
j = ja − jc = jo {exp[(1 − α)n a Fη/RT ]
− exp[−αn a Fη/RT ]}. (18) Use of the potential of a galvanic cell to measure the con-
centration of an electroactive species developed later than
Both kc and ka vary exponentially with potential a number of other electrochemical methods. In part, this
difference: was because a rational relation between electrode potential
kc = (kc )E=0 exp[−αn a EF/RT ], (19) and the concentration of an electroactive species required
the development of thermodynamics, and, in particular, its
ka = (ka )E=0 exp[(1 − α)n a EF/RT ]. (20) application to electrochemical phenomena. The work of
At equilibrium ( j = 0) and for the special case when J. Willard Gibbs in the 1870s provided the foundation for
(Cox )o = (Cred )o , the Nernst equation. The latter provides a quantitative re-
lationship between potential and the ratio of effective ther-
(kc )E=0 = (ka )E=0 = ks . (21) modynamic concentrations [activities] for a redox couple
This simple rate constant (ks ) is the defined value of kc ([ox]/[red]) and is the basis for potentiometry and poten-
and ka at the formal potential, E ◦ . Then, tiometric titrations. The utility of potentiometric measure-
ments for the characterization of ionic solutions was es-
(kc )E=0 = ks exp[−αn a F(E − E ◦ )/RT ] (22) tablished with the invention of the glass electrode in 1909
◦
(ka )E=0 = ks exp[(1 − α)n a F(E − E )/RT ] (23) for a selective potentiometric response to hydronium-ion
concentrations. Another milestone in the development of
j = n Fks {(Cred )o exp[(1 − α)n a F potentiometric measurements was the introduction of the
× (E − E ◦ )/RT ] hydrogen electrode for the measurement of hydronium-
ion concentrations; one of many important contributions
− (Cox )o exp[−αn a F(E − E ◦ )/RT ]}. by Professor Joel Hildebrand. Subsequent development
(24) of special glass formulations has made possible electrodes
that are selective to different monovalent cations. The idea
Redox couples in aqueous solutions at room temperature
is so attractive that intense effort has led to the develop-
have ks values that range from about 1.0 cm s−1 down to
ment of electrodes that are selective for many cations and
essentially zero (<10−10 cm s−1 ); e.g., ks = 0.1 cm s−1
anions, as well as several gas- and bioselective electrodes.
6 /Fe (CN)6
for FeIII (CN)3− II 4−
and 10−5 cm s−1 for
The use of these electrodes and the potentiometric mea-
Cr (OH2 )6 /Cr (OH2 )6 .
III 3+ II 2+
surement of pH continue to be among the most important
At the equilibrium potential, E = E e , j = 0, Coxo
=
applications of electrochemistry.
Cox , and Cred = Cred , where C denotes the concentration
b o b o
Electrochemistry 165
For such a half reaction the free energy is given by the More rigorous methods for the calculation of aqueous ac-
relation tivity coefficients are available.
[red ] The reaction of an electrochemical cell always involves
G = G ◦ + RT ln , (29) a combination of two redox half reactions such that one
[ox]
species oxidizes a second species to give the respective
where −G indicates the tendency for the reaction to go to redox products. Thus, the overall cell reaction can be ex-
the right; R is the gas constant and in the units appropriate pressed by a balanced chemical equation
for electrochemistry has a value of 8.317 J mol−1 K−1 ; T
is the temperature of the system in K; and the logarithmic a ox1 + b red2 c red1 + d ox2 , K equil . (36)
terms in the bracketed expression represent the activities
(effective concentrations) of the electroactive pair at the However, electrochemical cells are most conveniently
electrode surface. The free energy of this half reaction is considered as two individual half reactions, whereby each
related to the electrode potential, E, by the expression is written as a reduction in the form indicated by Eqs. (28)–
(32). When this is done and values of the appropriate
−G = n FE; −G ◦ = n FE ◦ . (30) quantities are inserted, a potential can be calculated for
each half-cell electrode system. Then that half-cell reac-
The quantity G ◦ is the free energy of the half reaction tion with the more positive potential will be the positive
when the activities of the reactant and product have values terminal in a galvanic cell, and the electromotive force of
of unity and is directly proportional to the standard half- that cell will be represented by the algebraic difference
cell potential for the reaction as written. It also is a measure between the potential of the more positive half-cell and
of the equilibrium constant for the half reaction, assuming the potential of the less positive half-cell,
the activity of electrons is unity,
E cell = E (more positive) − E (less positive) = E 1 − E 2 . (37)
−G ◦ = RT ln K . (31)
Insertion of the appropriate forms of Eq. (32) into Eq. (37)
An extensive summary of E ◦ values is presented in the
gives an overall expression for the cell potential,
compilation by Bard, Parsons, and Jordan (1985); the
most important are tabulated in Sawyer, Sobkowiak, and RT [ox1 ]a [red2 ]b
Roberts (1995). Standard potentials are thermodynamic E cell = E 1◦ − E 2◦ + ln . (38)
nF [ox2 ]d [red1 ]c
quantities that usually are evaluated via caloriometry
for a cell reaction (e.g., 2 H2 + ·O2 · → 2H2 O; E cell◦
= The equilibrium constant for the chemical reaction ex-
◦ ◦
[E O2 − E H+ /H2 ]) and the relationship of Eq. (30). pressed by Eq. (36) is related to the difference of the stan-
When Eqs. (29) and (30) are combined, the resulting dard half-cell potentials by the relation
Nernst expression relates the half-cell potential to the ef- ln K equil = (n F/RT )(E 1◦ − E 2◦ ). (39)
fective concentrations (activities) of the redox couple,
To apply potentiometric measurements to the determi-
RT [red ] RT [ox]
E = E◦ − ln = E◦ + ln . (32) nation of the concentration of electroactive species, a num-
nF [ox] nF [red] ber of conditions have to be met. The basic measurement
The activity of a species is indicated as the symbol of system must include an indicator electrode, which is capa-
the species enclosed in a bracket. This quantity is equal ble of monitoring the activity of the species of interest, and
to the concentration of the species times a mean activity a reference electrode, which gives a constant, known half-
coefficient, cell potential to which the indicator electrode potential
can be referred. The voltage resulting from the combina-
[M a+ ] = a M a+ = γ± C M a+ . (33) tion of these two electrodes must be measured in a man-
Although there is no straightforward and convenient ner that minimizes the amount of current drawn by the
method for evaluating activity coefficients for individual measuring system. For low-impedance electrode systems,
ions, the Debye-Huckel relationship permits an evalua- a conventional potentiometer is satisfactory. However,
tion of the mean activity coefficient (γ± ) for ions at low electrochemical measurements with high-impedance elec-
concentrations (usually below 0.01 M), trode systems, and in particular the glass-membrane elec-
√ trode, require the use of an exceedingly high-input-
µ impedance measuring instrument (usually an electrometer
log γ± = −0.509z 2 √ , (34)
1+ µ amplifier with a current drain of less than 10−12 A). Be-
cause of the logarithmic nature of the Nernst equation, the
where z is the charge on the ion and µ is the ionic strength
measuring instrumentation must have considerable sensi-
1 tivity. For example, a one-electron half reaction of 25◦ C
µ= Ci z i2 . (35)
2 gives a voltage change of 59.1 mV for a 10-fold change
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
166 Electrochemistry
in the concentration of the electroactive species. Another metal as a surface. However, during the past 50 years, it
important point is that the potential response is directly has become evident that platinum electrodes are not as in-
dependent on the temperature of the measuring system. ert as long believed and that their potentiometric response
Thus, if the correct temperature is not used in the Nernst is frequently dependent on the history of the surface and
expression, large absolute errors can be introduced in the the extent of its activation. The evidence is convincing that
measurement of the activity for an electroactive species. platinum electrodes, and, in fact, all metal electrodes, are
covered with an oxide film that changes its characteristics
with time. Nonetheless, the platinum electrode continues
B. Electrode Systems to enjoy wide popularity as an “inert” indicator of redox
The indicator electrodes for potentiometric measurements reactions and of the activities of the ions involved in such
traditionally have been categorized into four separate reactions.
classes. “First-class” electrodes consist of a metal im- For those redox couples that involve a metal ion plus
mersed in a solution that contains the metal ion. These the metal, the logical electrode system is the metal itself.
electrode systems provide a direct response to the ion or In other words, if the measured quantity is to be cupric
species to be measured, ion [copper (II)], a practical indicator electrode is a piece
of copper metal. All second-class electrodes involve an
RT active metal in combination with an insoluble compound
E = E◦ + ln [M n+ ]; M n+ + ne− M (s).
nF or salt. Thus, the silver/silver-chloride electrode actually
(40) is a silver/silver-ion electrode system that incorporates the
Therefore, the primary electrode reaction includes the means to control the silver-ion concentration through the
sensed species. Such electrodes give a direct response ac- chloride-ion concentration [Eq. (41)].
cording to the Nernst equation for the logarithm of the A number of the most common potentiometric electrode
activity of the species. systems and their applications are summarized in Table I.
Electrodes classified as “second-class” electrode sys- One of the most important and extensively used indicator
tems are those in which the electrode is in direct contact electrode systems is the glass-membrane electrode that is
with a slightly soluble salt of the electroactive species used to monitor hydronium-ion activity. Although devel-
such that the potentiometric response is indicative of the oped in 1909, it did not become popular until reliable elec-
concentration of the inactive anion species. Thus, the trometer amplifiers were developed in the 1930s (modern
silver/silver-chloride electrode system, which is represen- pH meters use high-input-impedance digital voltmeters).
tative of this class of electrodes, gives a potential response Figure 1 gives a schematic representation of this elec-
that is directly related to the logarithm of the chloride ion trode and indicates that the primary electrode system is
activity, a silver/silver-chloride (or mercury/mercurous-chloride)
RT electrode in contact with a known and fixed concentra-
E = E◦ − ln [Cl− ]; AgCl (s) + e− Ag (s) + Cl− , tion of hydrochloric acid (usually about 0.1 M). When
nF
(41)
even though it is not the electroactive species. This is true
because the chloride-ion concentration, through the solu- TABLE I Redox Potentials (E ◦ ) for the MI (OH2 )+
2 /M and
bility product, controls the activity of the silver ion, which MI OH/M, HO− Couples of Cu, Ag, and Au in H2 O and in MeCN
(0.1 M Tetraethylammonium Perchlorate)
is measured directly by the potentiometric silver-electrode
system. E ◦ , V vs NHEa
Because any potentiometric electrode system ultimately M MI (solv)+
n /M MI OH/M, HO−
must have a redox couple (or an ion-exchange process
in the case of membrane electrodes) for a meaning- A. H2 O Cu +0.52 −0.36
ful response, the most common form of potentiometric Ag +0.80 +0.34
electrode systems involves oxidation-reduction processes. Au +1.7
Hence, to monitor the activity of ferric ion [iron(III)], an H+
2 O·/H2 O;
HO·/HO− (GC) +2.72 +1.89
excess of ferrous ion [iron(II)] is added such that the con-
B. MeCN Cu +0.19 −0.79
centration of this species remains constant to give a di-
Ag +0.54 −0.30
rect Nernstian response for the activity of iron(III). For
Au +1.58 −0.19
such redox couples the most common electrode system
H+
2 O·(MeCN)/H2 O;
has been the platinum electrode. This tradition has come
HO·/HO− (GC) +3.2 +0.92
about primarily because of the historic belief that the plat-
inum electrode is totally inert and involves only the pure a SCE = +0.24 V vs NHE.
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry 167
168 Electrochemistry
such that changes in pH would be proportional to changes ence potentials that do not vary more than a few millivolts.
in the partial pressure of NH3 . These electrodes as well as The silver/silver chloride electrode also finds application
SOx -, NOx -, HF-, H2 S-, and HCN-sensing electrodes are in nonaqueous titrations, although some solvents cause
commercially available. the silver-chloride film to become soluble. Aqueous refer-
The gas-sensing electrodes also are used for the poten- ence electrodes are as convenient for nonaqueous systems
tiometric measurement of biologically important species. as any of the prototypes that have been developed to date.
An enzyme is immobilized at or near the gas probe. The When there is a need to rigorously exclude water, double-
gas sensor measures the amount of characteristic gas pro- salt bridges (aqueous/nonaqueous) are a convenient
duced by the reaction of the analyzed substance with solution.
the enzyme. For example, an enzyme electrode for urea For measurement of redox couples, a frequently over-
[NH2 C(O)NH2 ] determination is constructed by the im- looked but convenient reference electrode is a conven-
mobilization of urease onto the surface of an ammonia- tional glass pH electrode (assuming the sample solution
selective electrode. When the electrode is inserted into a system contains a constant level of acidity). Such an elec-
solution that contains urea, the enzyme catalyzes its con- trode provides an extremely inert and stable reference po-
version to ammonia, tential that is completely indifferent to most redox species.
urease However, the glass electrode requires the use of an elec-
NH2 C(O)NH2 + H2 O 2 NH3 + CO2 . (43)
trometer amplifier such as that contained in pH meters.
The generated ammonia diffuses through the gas-
permeable membrane and is detected by the probe of the
C. Applications of Potentiometry
electrode. A steady-state signal is reached when the rate
of NH3 diffusion from the electrode equals the rate of its Although all potentiometric measurements (except those
generation. involving membrane electrodes) ultimately are based on a
In general, a necessary part of a potentiometric mea- redox couple, the method can be applied to oxidation-
surement is the coupling of a reference electrode to the reduction processes, acid-base processes, precipitation
indicating electrode. The ideal reference electrode has a processes, and metal-ion complexation processes. Mea-
number of important characteristics: (1) a reproducible surements that involve a component of a redox couple
potential, (2) a low-temperature coefficient, (3) the capac- require that either the oxidized or reduced conjugate of
ity to remain unpolarized when small currents are drawn, the species to be measured be maintained at a constant
and (4) inertness to the sample solution. If the reference and known activity at the electrode. If the goal is to mea-
electrode must be prepared in the laboratory, a convenient sure the activity of silver ion in a solution, then a silver
and reproducible system is desirable. wire coupled to the appropriate reference electrodes makes
Although the standard half-cell reactions are all refer- an ideal potentiometric system. Likewise, if the goal is
enced to the standard hydrogen electrode (NHE or SHE), to monitor iron(III) concentrations with a platinum elec-
this is an exceedingly awkward reference electrode. It has trode, a known concentration of iron(II) must be present
been selected because its potential falls in the middle of in the sample solution such that potential changes are only
those for the most common half reactions in water and dependent on the iron(II) concentration.
because with rigorous care highly reproducible potentials Table I summarizes a number of redox couples that are
can be duplicated by equally careful workers in other lab- well behaved in aqueous solutions and provide a means for
oratories. Furthermore, because it consists of a platinized monitoring the indicated species by potentiometric mea-
platinum electrode over which hydrogen gas is passed in surements. This can be either in the form of monitoring
combination with a known activity of hydronium ion, it a titration or as a direct absolute measurement of activity.
can be combined in a number of cases with other half-cells Although the tabulations of standard potentials infer that
without a liquid junction. These three factors undoubtedly the listing should be much more comprehensive, most of
justify its selection as the ultimate reference electrode for the couples tabulated are not well behaved in an electro-
fundamental measurements. The electrode reaction is a chemical sense and do not provide a Nerstian response
typical redox half reaction which includes the oxidized under normal laboratory conditions. The vast majority of
and reduced forms of hydrogen. By controlling the hy- the data tabulated is based on calorimetric data.
drogen partial pressure at a fixed level, this becomes an The major application of the potentiometric method
indicating electrode for hydronium-ion activity. is for acid-base measurements in both aqueous and non-
For most potentiometric measurements, either the sat- aqueous solvent systems. Although the glass electrode is
urated calomel reference electrode or the silver/silver- universally the most common indicating electrode sys-
chloride reference electrode is used. These electrodes can tem for such measurements, many other electrodes have
be made compact, are easily produced, and provide refer- been developed. However, except in extremely specialized
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry 169
circumstances, none of these provides the reliability and sure of metal-ion activity. If one is concerned purely with
precision that is afforded by the glass electrode. In the ab- the actual activity of free metal ion, these measurements
sence of interfering substances, the quinhydrone electrode are meaningful. However, if the measurement is taken to
(an equimolar combination of quinone and hydroquinone represent the total metal-ion content of the solution, both
with a gold-foil electrode) provides a simple monitoring as a free solvated ion and as its various complexes, then
system for measurements up to pH 8. However, the pres- highly erroneous conclusions can be made. Thus, the abil-
ence of oxidizing or strongly reducing ions in the sample ity to monitor the concentration of ligands should be rec-
system will interfere, as is true for the hydrogen gas elec- ognized as a pitfall if one does not take account of this in
trode and most other systems that are an alternative to the the use of potential measurements to monitor metal-ion
glass electrode. concentrations.
Potentiometric redox measurements are often per- A recent and rapidly developing extension of poten-
formed in nonaqueous or mixed-solvent media. For such tiometry is in the area of membrane-type indicator elec-
solvents, various potentiometric sensors have been devel- trodes. These include (1) specialized glass electrodes that
oped, which, under rigorously controlled conditions, give respond to ions other than the hydronium ion, (2) ion-
a Nernstian response over a wide ranges of activities (par- exchange membranes, and (3) single-inorganic-crystal
ticularly in buffered solutions). There are some experi- membranes. Each year the selectivity and reliability are
mental limitations, e.g., solvent purification and handling improved for this important class of electrode. In partic-
or use of a reference electrode without salt bridges, but ular, the development of ion-exchange membranes that
there are important advantages. Solutes may be more sol- provide selective response for a number of anions has
uble in such media, and redox properties of the species made new areas of analysis amenable to potentiometric
may be altered in comparison with aqueous solutions. measurements. This has been particularly important for
Measurements of pH in nonaqueous solvents almost with- the biomedical field where nondestructive, highly specific
out exception use the glass electrode in combination poteniometric measurements are desirable. Furthermore,
with an appropriate reference electrode, frequently the the potentiometric method, because of its continuous na-
silver/silver-chloride electrode. In general, the response ture, is particularly attractive to those concerned with in
of the glass electrode follows the Nernst expression in vivo monitoring of biological substances.
nonaqueous solvents and is an accurate representation of Because potentiometry (through the Nernst equation)
the changes in activity of hydronium ion. Unfortunately, gives a response that is proportional to the logarithm of
few, if any, standard buffers are available to calibrate pH the activity of the electroactive ion, the accuracy and pre-
meters for nonaqueous measurements. Thus, nonaqueous cision are more limited than for many methods that give
pH measurements are only meaningful for monitoring the a direct proportional response. Thus, for a one-electron
course of an acid-base titration or relative to some refer- redox process, an order of magnitude change in activity
ence measurement made within the individual laboratory. gives a potential change of 59.1 mV (at room temperature),
Little, if any, confidence can be attached to absolute pH a 10% change in activity gives a change of 2.5 mV, and a
measurements in nonaqueous systems. The application of l% change in activity gives only a change of 0.25 mV.
ion-selective electrodes in nonaqueous media has been Potentiometry has found extensive application over the
limited, but the response for several cations and anions past 50 years as a means to evaluate various thermody-
is usable, especially when 10–20% of water is added to namic parameters. Although this is not the major appli-
pure nonaqueous solvent. Limited use of liquid-membrane cation of the technique today, it still provides one of the
electrodes in such media arises from the solubility of most convenient and reliable approaches to the evalua-
electrode-system components in organic media. tion of thermodynamic quantities. In particular, the ac-
Second-class electrodes, that is those whose response tivity coefficients of electroactive species can be evalu-
is dependent on the change in concentration of an anion ated directly through the use of the Nernst equation (for
which gives an insoluble salt with the metal ion of the species that give a reversible electrochemical response).
indicator electrode, provide a general means for moni- Thus, if an electrochemical system is used without a junc-
toring the concentrations of anions. Some of those half tion potential and with a reference electrode that has a
reactions, which are well behaved electrochemically and well-established potential, then potentiometric measure-
provide means for the potentiometric monitoring of anion ment of the constituent species at a known concentration
species, are summarized in Table I. This table also includes provides a direct measure of its activity. This provides a
a tabulation of redox reactions that are useful to moni- direct means to evaluate the activity coefficient (assuming
tor the concentration of ligands that can complex metal the standard potential is known accurately for the con-
ions. Consideration of these indicates one of the difficul- stituent half reaction). If the standard half-reaction poten-
ties with absolute potentiometric measurements as a mea- tial is not available, it must be evaluated under conditions
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
170 Electrochemistry
where the activity coefficient can be determined by the at the half-equivalence point; the half-reaction potential,
Debye-Huckel equation. assuming activities are equal to concentrations, is given
Another important application is the use of potentio- directly by the potential of the indicator electrode relative
metric measurements for the evaluation of thermodynamic to the reference electrode. If the latter is a standard hydro-
equilibrium constants. In particular, the dissociation con- gen electrode, the measured potential is equal to the E ◦ for
6 /Fe (OH2 )6 couple. The evaluation of
the FeIII (OH2 )3+ II 2+
stants for weak acids and weak bases in a variety of sol-
◦
vents are evaluated conveniently with a pH electrode mea- the E for a half reaction provides a direct measure of the
suring system. The most precise approach is to perform free energy for the half reaction relative to the free energy
an acid-base titration such that the titration curve can be for the reduction of hydronium ion to hydrogen gas. Like-
recorded. Obviously, one could measure the pH of a known wise, a combination of any pair of E ◦ values or of the free
concentration of a weak acid and obtain a value of its energy values permits the evaluation of the equilibrium
hydronium-ion activity which would permit a direct eval- constant and the standard free energy for a redox reaction.
uation of its dissociation constant. However, this would From a practical standpoint it is often useful to have the
be a one-point evaluation and subject to greater errors observed potential in the medium of measurement for the
than by titrating the acid halfway to the equivalence point. condition of equal concentrations of the oxidized and re-
The latter approach uses a well-buffered region where the duced species of a half reaction. Such potentials are known
pH measurement represents the average of a large num- as formal potentials, E ◦ , rather than standard potentials
ber of data points. Similar arguments can be made for the and are not purely thermodynamic quantities. The term
evaluation of solubility products and stability constants of “formal potential” comes from the tradition of having the
complex ions. supporting electrolyte at a one formal concentration. How-
In the use of potentiometry for the evaluation of stability ever, other stated solution conditions also are included in
constants for complex ions, the expressions can become many listings. Thus, the indicated potential is what one
extremely complicated if multi-equilibria are present. For would expect at the half-equivalence point under actual
a simple one-to-one complex, a direct potentiometric titra- titration conditions. In other words, activity corrections
tion curve again provides the most satisfactory route to an have not been made.
accurate evaluation of the constant. The curve looks sim-
ilar to that for an acid-base titration, and the appropriate
III. CONTROLLED-POTENTIAL
point to pick is the half-equivalence point. If the complex is
ELECTROLYSIS AND VOLTAMMETRY
extremely stable, then the amount of free metal ion at this
point on the titration curve (ligand titrated with metal ion)
For chemists, the second important application of elec-
is sufficiently low that it can be disregarded. Assuming it
trochemistry (beyond potentimetry) is the measurement
is a stable complex, at the first half-equivalence point the
of species-specific [e.g., iron(III) and iron(II)] concentra-
concentration of complexed metal ion will be equivalent to
tions. This is accomplished by an experiment whereby the
that of the free ligand. The potential will give a direct mea-
electrolysis current for a specific species is independent of
sure of the free metal ion and allow the stability constant
applied potential (within narrow limits) and controlled by
for the complex to be evaluated at the half-equivalence
mass transfer across a concentration gradient, such that
point, e.g.,
it is directly proportional to concentration (i = kC). Al-
[AgI (en)+ ] though the contemporary methodology of choice is cyclic
Kf = . (44)
AgI (OH2 )+
2 [en]
voltammetry, the foundation for all voltammetric tech-
niques is polarography (discovered in 1922 by Professor
Potentiometry also is a direct means to evaluate the
Jaroslov Heyrovsky; awarded the Nobel Prize for Chem-
standard potential for half reactions (E ◦ ) and has been ap-
istry in 1959). Hence, a historical approach is used with
plied for appropriate reversible systems. Such measure-
a recognition that cyclic voltammetry will be the primary
ments require corrections for activity coefficients or ex-
methodology for most chemists.
trapolation of the data to infinite dilution. Again, direct
measurements in which equal molar concentrations of the
oxidized or reduced species are introduced into the sys- A. Principles and Fundamental Relations
tem provide a simple approach to such evaluations and
1. Diffusion to a Planar Electrode
are as precise as those obtained by less direct methods.
However, E ◦ values also can be extracted from poten- The basic approach in controlled-potential methods of
tiometric titration data. For example, in the titration of electrochemistry is to control in some manner the poten-
FeII (OH2 )2+ IV 4+ III
6 ion with Ce (OH2 )n ion, the Fe (OH2 )6
3+
tial of the working electrode while measuring the resultant
II 2+
ion concentration equals the Fe (OH2 )6 concentration current, usually as a function of time. When a potential
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry 171
sufficient to electrolyze the electroactive species com- This relationship holds for any electrochemical process
pletely is applied to the electrode at (t = 0), the concen- that involves semi-infinite linear diffusion and is the basis
tration at the electrode surface is reduced to zero and an for a variety of electrochemical methods (e.g., polarogra-
electrode process occurs, for example, phy, voltammetry, and controlled-potential electrolysis).
Equation (50) is the basic relationship used for solid-
ox + ne− red, (45)
electrode voltammetry with a preset initial potential on
where ox and red represent an oxidized and reduced form a plateau region of the current-voltage curve. Its applica-
of an electroactive species. Passage of current requires tion requires that the electrode configuration be such that
material to be transported to the electrode surface as well semi-infinite linear diffusion is the controlling condition
as away from it. Thus, relationships must be developed for the mass-transfer process.
which involve the flux and diffusion of materials; this is
appropriately accomplished by starting with Fick’s second
law of diffusion, B. Voltammetry
∂C(x,t) D∂C(x,t)
= , (46) 1. Polarography
∂t ∂x2
where D represents the diffusion coefficient, C represents The most extensively studied form of voltammetry has
the concentration of the electroactive species at a dis- been polarography (first described by Heyrovsky in 1922,
tance x from the electrode surface, and t represents the with the quantitative relationships of current, potential,
amount of time that the concentration gradient has ex- and time completed by the early 1930s with the assistance
isted. Through the use of Laplace transforms with initial of associates such as Ilkovic). The potential-time depen-
and boundary conditions; dence that is used for polarographic measurements is pre-
sented in Fig. 2 (solid line). The potential is scanned from
for t = 0 and x ≥ 0 C = C b, E 1 to E 2 to obtain a current response that qualitatively
for t ≥ 0 and x → 0 C → C b, and quantitatively characterizes the electroactive species
present. The vast body of data from polarographic mea-
for t > 0 and x = 0 C = 0. surements can be adopted by other electroanalytical meth-
Equation (46) can be solved to give a relationship for con- ods. Moreover, pulse polarographic methods and anodic
centration in terms of parameters x and t, stripping analysis, which are still used for determination
x of trace amounts of metal ions, are closely related to po-
C(x,t) = C b erf 1/2 1/2 , (47) larography. The unique characteristic of polarography is
2D t
its use of a dropping mercury electrode, such that the elec-
where C b is the bulk concentration of the electroactive trode surface is continuously renewed in a well-defined
species. and regulated manner to give reproducible effective elec-
By taking the derivative of Eq. (47) for the proper trode areas as a function of time. The diffusion cur-
boundary condition, namely at the electrode surface rent equation [Eq. (50)] can be extended to include a
(x = 0), the diffusion gradient at the electrode surface is dropping mercury electrode by appropriate substitution
expressed by the relation for the area of the electrode. Thus, the volume of the
∂C Cb drop for a dropping mercury electrode is given by the
= 1/2 1/2 1/2 . (48) relationship
∂ x (0,t) π D t
This flux of material crossing the electrode boundary can 4 3 mt
be converted to current by the expression V= πr = , (51)
3 d
∂C(0,t)
i = n FAD , (49) where r is the radius of the drop of mercury, m is the mass
∂x
flow rate of mercury from the orifice of the capillary, t is
where n is the number of electrons involved in the elec-
the life of the drop, and d is the density of mercury under
trode reaction, F is the faraday, and A is the area of the
the experimental conditions. When this equation is solved
electrode. When Eq. (48) is substituted into this relation a
for r and the latter is substituted into the equation for the
complete expression for the current that results from semi-
area of a sphere, an expression for the area of the dropping
infinite linear diffusion is obtained (the Cottrell equation
mercury electrode drop as a function of the experimental
for a planar electrode),
parameters is obtained,
n FAC b D 1/2
i= . (50)
π 1/2 t 1/2 A = (4π )1/3 32/3 m 2/3 t 2/3 d −2/3 . (52)
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
172 Electrochemistry
FIGURE 2 Potential-time profile that is used for polarography and linear-sweep voltammetry (solid line) and cyclic
voltammetry (both solid and dashed lines).
This then can be substituted into Eq. (50) to give a calcu- rial produced by the electrolysis reaction, these quantities
lated diffusion current for the dropping mercury electrode, can be directly related to the concentration of the species
at the electrode surface. For a generic reduction process
(i t )calc = 462nC b D 1/2 m 2/3 t 1/6 . (53)
[Eq. (45)], the potential of the electrode is given by the
Actually, experimental results indicate √that the constant Nernst equation
in Eq. (53) is too small by a factor of 7/3, due to the
growth of the mercury drop into the solution away from RT Cox(0,t)
E = E◦ + ln . (56)
the capillary orifice. Thus, the correct diffusion current nF Cred(0,t)
expression for a dropping mercury electrode is
Initially, the solution contains only ox (concentration,
i = 706nC b D 1/2 m 2/3 t 1/6 , (54) b
Cox ). When a potenital is applied and reductive current
which gives the current at any time up to the lifetime of the flows, the oxidized form of the electroactive species dif-
drop. If the drop time, td , is substituted, the well-known fuses toward the electrode. From Eq. (63) it follows that
Ilkovic equation results, b 1/2 2/3 1/6
1/6 i = 706n Cox − Cox(0,t) Dox m td
i d = 706nC D b 1/2
m 2/3 td , (55)
1/6
= i d − 706nCox(0,t) Dox m td .
1/2 2/3
(57)
where i d is in A if D is in cm2 s−1 , C b is in mol cm−3 , m
in mg s−1 , and t is in s. Alternatively, i d is in µA when
Hence,
C b is in mM.
The polarographic current-potential wave that is illus- id − i
Cox(0,t) = 1/2
. (58)
trated by Fig. 3 conforms to the Nernst equation for re- 706n Dox m 2/3 t 1/6
versible electrochemical processes. However, it is more
convenient to express the concentrations at the electrode After reduction of ox, its reduced form (red ) diffuses either
surface in terms of the current, i, and diffusion current, into the bulk of solution or into the mercury to form an
i d . Because i d is directly proportional to the concentra- amalgam. In either case,
tion of the electroactive species in the bulk and i at any 1/2 2/3 1/6
point on the curve is proportional to the amount of mate- i = 706n Cred(0,t) − Cred
b
Dred m td . (59)
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry 173
But the reduced form of the electroactive species was 2. Linear Sweep and Cyclic Voltammetry
not present in the solution before electrolysis, therefore,
The potential-time relation for voltammetric measure-
b
Cred = 0 and
ments is presented in Fig. 2. With linear-sweep voltamme-
i try, the potential is linearly increased between potentials
Cred(0,t) = 1/2
. (60)
706n Dred m 2/3 t 1/6 E 1 and E 2 . Cyclic voltammentry is an extension of linear-
sweep voltammetry with the voltage scan reversed after
Substitution of Eqs. (58) and (60) into Eq. (56) gives the current maximum (peak) of the reduction process has
been passed. The voltage is scanned negatively beyond the
RT Dred 1/2 RT i d − i
E = E◦ + ln + ln . (61) peak and then reversed in a linear positive sweep. Such a
nF Dox nF i
technique provides even more information about the prop-
At the half-height of a polarographic wave (i = i d /2), the erties and characteristics of the electrochemical process
corresponding potential is defined as the half-wave poten- and also gives insight into any complicating side processes
tial (E 1/2 ). Therefore, Eq. (61) takes the form such as pre- and post-electron-transfer reactions as well
RT id − i as kinetic considerations. Whereas in classical polarog-
E = E 1/2 +ln . (62) raphy the voltage-scan rate is about 1 V min−1 , linear-
nF i
sweep voltammetry uses scan rates up to 100 V s−1 for
For the reduction of a simple solvated metal ion to its conventional microelectrodes (and up to 10,000 V s−1 for
amalgam, E 1/2 is given by ultra-microelectrodes; 10−6 m diameter).
RT γion Da
1/2 Figure 4 illustrates the shape of a cyclic voltammo-
E 1/2 = E ◦ + ln 1/2
, (63) gram with an electrode of fixed area. The voltammogram
nF γa Dion is characterized by a peak potential, E p , at which the cur-
where γion is the activity coefficient for the ion and γa is rent reaches a maximum value, and by value of the peak
the activity coefficient for the amalgamated species. The current, i p . When the reduction process is reversible the
diffusion coefficients for the amalgam and ionic species peak current is given by the relation
also are a part of this expression. The standard reduction i p = 0.4463n FA(Da)1/2 C b (64)
potential is for reduction of the ion to the amalgamated
species. These expressions also hold for the reduction of with
an ion to a lower oxidation state, but require that the ap- n Fν nν
a= = at 25◦ C, (65)
propriate value be used. RT 0.026
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
174 Electrochemistry
RT 0.0285
E p = E 1/2 − 1.11 = E 1/2 − at 25◦ C
nF n
(69)
Another useful parameter of the voltammetric curves is the
half-peak potential, E p/2 , which is the potential at which
the registered current reaches half its maximum value and
is used to characterize a voltammogram. For a reversible
process, E 1/2 is located halfway in between E p and E p/2 .
The ratio of the peak current for the cathodic process
relative to the peak current for the anodic process is equal
to unity (i p,c /i p,a = 1) for a reversible electrode process.
To measure the peak current for the anodic process the ex-
trapolated baseline going from the foot of the cathodic
wave to the extension of this cathodic current beyond
the peak must be used as a reference, as illustrated by
Fig. 4.
For the condition
0.141
|E λ − E p/2 | ≥ , (70)
n
where E λ is the extent of the voltage sweep, the differ-
FIGURE 4 Linear voltage-sweep voltammogram with reversal of
sweep direction to give a cyclic voltammogram. The initial sweep ence in the peak potentials between the anodic and ca-
direction is to more negative potential. thodic processes of the reversible reaction is given by the
relationship
0.059
where ν is the scan rate in volts per second. This relation |E p | = |E p,a − E p,c | = at 25◦ C, (71)
results from the set of differential equations for Fick’s n
second law of diffusion (with the appropriate initial and which provides a rapid and convenient means to determine
boundary conditions for ox and red ). Thus, in terms of the number of electrons involved in the electrochemical
the adjustable parameters the peak current is given by the reaction. For a reversible system, i p is a linear function of
Randles-Sevcik equation √
ν, and E p is independent of ν.
i p = 2.69 × 105 n 3/2 AD 1/2 C b ν 1/2 at 25◦ C, (66)
where i p is in A, A is in cm2 , D is in cm2 s−1 , C b is in mol C. Controlled-Potential Bulk Electrolysis
cm−3 , and ν is in V s−1 .
Because of the extensive amount of data available from the
Nicholson and Shain revolutionized the voltammetric
polarographic and voltammetric literature, the optimum
experiment with their elegant development and demon-
conditions for macroscopic electrolyses often are estab-
stration of linear-sweep and cyclic voltammetry. In their
lished. In particular, controlled-potential electrolysis at a
approach, the current-potential curve is presented as
mercury pool can be approached with predictable success
i = n FAC b (πDa)1/2 χ (at). (67) on the basis of available polarographic information for
the system of interest. An electrolysis can be accelerated
From tabulations for the relation between π 1/2 χ ( at) and by maximizing the electrode surface area and minimizing
n(E − E 1/2 ) and converting the term π 1/2 χ (at) to χrev , the thickness of the diffusion layer. However, the same
Eq. (67) takes the form electrode material must be used as in polarography. Thus,
i = n FAC b (Da)1/2 χrev . (68) a conventional approach in controlled-potential electrol-
ysis is the use of a mercury pool stirred as vigorously as
For a given potential (E) the value of χrev is obtained from possible with a magnetic stirring bar to minimize the con-
tabulations. centration gradient. Under such conditions the decay of
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry 175
the current as well as the decay of the concentration of the in the electrode reaction be known, and the concentration
electroactive species is given by the relation of the electroactive species and the area of electrodes be
known. With these conditions satisfied, diffusion coeffi-
it Ct −D A
= = exp t , (72) cients can be evaluated rapidly over a range of tempera-
i t=0 Ct=0 Vx tures and solution conditions.
where V is the volume of the solution to be electrolyzed Voltammetric methods also provide a convenient ap-
and x is the thickness of the concentration gradient. proach to establish the thermodynamic reversibility of an
Thus, the current and concentration decay exponentially. electrode reaction and for the evaluation of the electron
Under idealized conditions, 90% of the electroactive stoichiometry for the electrode reaction. As outlined in
species will be electrolyzed in approximately 20 min. In- earlier sections, the standard electrode potential, the dis-
creases in the temperature as well as in the electrode area sociation constants of weak acids and bases, solubility
relative to the solution volume will accelerate the rate of products, and the formation constants of complex ions can
electrolysis. The fundamentals of the controlled-potential be evaluated from polarographic half-wave potentials, if
bulk electrolysis are discussed in recent monographs (see the electrode process is reversible. Furthermore, studies of
Bibliography). half-wave potentials as a function of ligand concentration
provide the means to determine the formula of a metal
complex.
D. Applications of Controlled-Potential Methods
The techniques of voltage sweep and cyclic voltamme-
To date, the most extensive application of electrochemical try provide the analytical and physical chemical capabili-
methods with controlled potential has been in the area of ties of classical voltammetry and, in addition, provide the
qualitative and quantitative analysis. Because a number of means to perform these measurements much more rapidly
monographs have more than adequately reviewed the liter- for a broader range of conditions. Cyclic voltammetry is
ature and outlined the conditions for specific applications, particularly useful for the rapid assessment of thermody-
this material is not covered here. In particular, inorganic namic reversibility and for the evaluation of the stoichiom-
applications of polarography and voltammetry have been etry for the electrode reaction.
discussed in great detail in the classic treatise by Kolthoff
and Lingane.
An important specialized type of voltammetric system IV. ELECTRON-TRANSFER PROCESSES
is a self-contained cell for the determination of ·O2 · in the
gas or solution phases. This is the so-called Clark elec- In electrochemical cells, electron transfer occurs within
trode, which consists of a platinum or gold electrode in the the electrode/solution interface, with electron removal
end of a support rod that is covered by an ·O2 ·-permeable (oxidation) at the anode and electron introduction (reduc-
membrane (polyethylene or Teflon) such that a thin film tion) at the cathode. The current through the solution is
of electrolyte is contained between the electrode surface carried by the ions of the electrolyte, and the voltage limits
and the membrane. A concentric tube provides the support are those for electron removal from and electron insertion
for the membrane and the means to contain an electrolyte into the solvent/electrolyte {e.g., H2 O/(H+ −
3 Oaq )(ClO4aq );
I + −
solution in contact with a silver-silver chloride reference [Na (OH2 )6 ](Claq )}
electrode. The Clark device has found extensive applica- −e−
tion to monitor ·O2 · partial pressure in blood, in the atmo- 2 H2 O H2 O(H2 O·)+ H+
3O
sphere, and in sewage plants. By appropriate adjustment
of the applied potential, it gives a voltammetric current +[HO·](E ◦ )pH 0 , + 2.72 V vs NHE
plateau that is directly proportional to the ·O2 · partial pres- (E ◦ )pH 7 , +2.31 V (73)
sure. The membrane material prevents interference from −e −
electroactive ions as well as from surface-contaminating HO− [HO·] (E ◦ )pH 14 , +1.89 V (74)
biological materials. −e −
176 Electrochemistry
by the presence of substrate atoms (rather than be from at much lower potentials [Eq. (73); at pH 0, +2.72 V vs
them). NHE]. At a silver electrode, water oxidation is facilitated
For example, with water at pH 0 the process of Eq. (73) via formation of an AgI –OH+2 bond; the shift in oxidation
is shifted −4.82 V when hydrogen atoms are present, potential from +2.72 to +0.80 V is a measure of the bond-
−e− formation energy (−G BF ),
[H·] + 2 H2 O (H2 O)H+
3O H+
3 O + H2 O −G BF = (+2.72 − 0.80) 23.1 = 44.4 kcal mol−1 . (78)
◦
E , −2.10 V. (76) −
At pH 14 the anodic process is the oxidation of HO ,
e−
A. Oxidative Electrochemistry: Metals, HO− HO· E ◦ , +1.89 V
Metal Complexes, Lewis Bases (E ◦ )pH 7 , +2.30 V, (79)
1. Metals I
which is facilitated via formation of an Ag –OH bond
The transformation of metal electrode surfaces via ele- [−G BF = (1.89 − 0.34) 23.1 = 35.8 kcal mol−1 ]
ctrooxidation to their metallo-oxides, solvated ions, and −e−
Ag (s) + HO− AgI −OH (s)
metal complexes is fundamental to most anodic elec-
trochemical processes (batteries, electrorefining, anodic E◦ , +0.34 V
stripping analysis, and reference electrodes). Although (E ◦ )pH 7 , +0.75 V. (80)
this is traditionally represented as the removal of one (or
more) valence electron from a metal atom at the electrode The data of Eqs. (77) and (79) can be combined to give a
surface to give a metal ion [e.g., Ag (s) − e− → Ag+ ; E ◦ , value for the solubility product (K sp ) for Ag–OH (s),
+0.80 V vs NHE], the gas-phase ionization potential [e.g., AgI (OH2 )+ −
2 + HO AgI −OH (s) + 2 H2 O
Ag · (g) − e− → Ag+ (g); IP, 7.6 eV] is far greater than the I
observed oxidation potential. The difference is attributed Ag (OH2 )+ −
2 [HO ] = K sp ;
to the solvation energy for the metal ion [e.g., Ag+ + log K sp = (0.34 − 0.80)/0.059 = −7.8.
nH2 O → Ag+ (aq); −G(aq) ≈ 70–100 kcal mol−1 ].
(81)
However, such a sequential path would not obviate the
7.6-V energy barrier for the initial step and is in conflict In the presence of chloride ion, metal electrodes facil-
with the observed thermodynamic reversibility for many itate its oxidation [Eq. (75); E ◦ , +2.41 V vs NHE] via
metal/solvated-metal-ion redox couples. formation of metal-chlorine covalent bonds, e.g.,
All reactions, and particularly redox processes, occur −e−
via the easiest and lowest energy pathway that is avail- Ag (s) + Cl− Ag−Cl (s) E ◦ , +0.22 V. (82)
able (mechanistically feasible) to the system. In the case Hence, the differential bond-formation energy
of a metal electrode/electrolyte interface undergoing an- [(−G BF )] [Ag–Cl bond energy, minus the energy
odic transformations, the electrons can come from (a) sur- required to break an Ag–Ag bond at the Ag (s) surface)] is
face metal atoms (energy limit; first ionization potential), given by the difference in oxidation potentials [Eqs. (75)
(b) solvent molecules (energy limit; oxidation potential and (82)],
of solvent), (c) electrolyte anions (energy limit; oxidation
potential of anions), and (d) base ligands (energy limit; ox- (−G BF ) = (2.41 − 0.22) 23.1 = 50.6 kcal mol−1 .
idation potential of ligand). All metal electrodes are elec- (83)
trochemically transformed via path b, c, or d and never
Because the escape energy for an Ag· atom from Ag (s) is
via path a. This general conclusion is illustrated for silver
68 kcal mol−1 , a reasonable approximation for the break-
and copper electrodes in aqueous and acetonitrile (MeCN)
age of a single bond is 22.6 kcal mol−1 [(1/3) 68]. When
solutions that contain inert electrolyte, chloride ion (Cl− ),
combined with Eq. (83), this gives a reasonable value for
or bipyridine (bpy).
−G BF ,
In aqueous solutions at pH 0, the silver electrode facil-
itates oxidation of water Ag· + Cl· Ag–Cl −GBF = 50.6 + 22.6
−e− = 73.2 kcal mol−1 .
Ag (s) + 2 H2 O AgI (OH2 )+
2
(84)
E ◦ , +0.80 V vs NHE. (77)
The literature value for the dissociative bond energy
The gas-phase ionization potential for a silver atom is (HDBE ) of Ag–Cl (g) is 81.6 kcal mol−1 , which is equiv-
7.6 eV. In contrast, water is oxidized (gives up an electron) alent to an estimated −G BF value of 73.8 kcal mol−1
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry 177
[−G BF = HDBE − TS = 81.6 − 7.8 (est) = 73.8 kcal and CeIII (OH2 )3+ +
6 , each with one, two, and three M–OH2
mol−1 ]. Thus, the proposition that metal electrode ox- covalent bonds, respectively],
idations are solvent or ligand centered with potentials
−e− II
that reflect the metal-solvent/ligand bond-formation free CuI (OH2 )+
6 + H2 O (H2 O)+
5 Cu
(E ◦ )pH 0 ,+0.16 V
energies (−G BF ) is supported by independent bond-
energy data. The data of Eqs. (77) and (82) provide a meas- − OH+
2 (OH2 ) (H2 O)+
5 Cu −OH + H3 O
II +
178 Electrochemistry
E b1/2 , V vs NHEc
Ligand
(L or L− )a L/L+ · (L− /L·) ZnII L3 MnII L3 FeII L3 CoII L3
Electrochemistry 179
TABLE III Apparent Metal-Ligand Covalent-Bond- The electrochemical reduction of permanganic acid
Formation Free Energies (−∆G BF ) for Several Man- [HOMnVII (O)3 ], which is traditionally represented as a
ganese, Iron, and Cobalt Complexes
metal-centered electron transfer to change Mn7+ to Mn6+ ,
a
Complex −∆G BF , kcal mol−1 is another example of a ligand-centered process,
A. Manganese
(O)3 MnVII –OH + H+
3O+e
−
MnVI (O)3 + 2 H2 O
(8Q)2 MnIII –8Q 6
(acac)2 MnIII –acac 9 −GBF = 28 kcal mol−1 (pH 1). E◦ , +1.45 V (106)
(PA)2 MnIII –PA 22
[(bpy)2 MnIII –bpy]3+ >23b Comparison of this with the reduction of free hy-
B. Iron droxyl radical (HO·) [Eq. (104)] provides a measure
(8Q)2 FeIII –8Q 15 of the HO–MnVII (O)3 bond energy [−G BF = (2.66 −
(acac)2 FeIII –acac 23 1.45) 23.1 = 28 kcal mol−1 ]. The other strong oxidants
3+
(PA)2 FeIII –PA 31 [(HO)2 CrVI IV
2 (O)5 and HOCe (OH2 )5 ] that are used for
[(bpy)2 FeIII –bpy]3+ >29b aqueous redox titrations are reduced by a similar path,
[(Ph3 PO)3 FeIII –OPPh3 ]3+ >30b
[(MeCN)4 FeIII –OH2 ]3+ 23 HO–CrVI (O)2 OCrVI (O)2 OH + H+
3O+e
−
C. Cobalt
+1.30 V (O)2 CrV OCrVI (O)2 OH + 2 H2 O
(8Q)2 CoIII –8Q 16
(acac)2 CoIII –acac 21 −G BF , 31 kcal mol−1 (pH 1) (107)
(PA)2 CoIII –PA 35 + −
[(bpy)2 CoIII –bpy]3+ >46b 5 + H3 O + e
HO–CeIV (OH2 )3+
a (−G BF ) = [E 1/2[ZnL − /ZnL 2 (L·)] − E 1/2(M L − /M(·L)L 2 ) ] +1.66 V 6 + H2 O
CeIII (OH2 )3+
3 3
× 23.1 kcal mol−1 . −GBF , 23 kcal mol−1 (pH 1). (108)
b (−G ) = [E
BF p,a (ZnL/ZnL + ·) − E p,a(M L/M−L + ) ] × 23.1 kcal
mol−1 ; L = (bpy)3 or (Ph3 PO)4 .
An important point in these electron-transfer reductions
is that the primary electron acceptor is the hydronium ion
e (H+ 3 O), which is transformed to a hydrogen atom (H·) that
reacts with HO· (either free or bound via a covalent bond
to the metal center). Thus, in the reactions of Eqs. (103),
(104), and (106)–(108), the oxidant in each is the hydro-
nium ion (H+
H2O
H HO 3 O) and the reduction potential is determined
by the H–OH bond energy (−G BF ) of the product H2 O,
(100)
minus the metal–OH bond energy [Eqs. (106)–(108)].
Hence, reductive electrochemistry converts electrons (e− ) Under alkaline conditions MnVII O− 4 is reduced via di-
via the solution matrix at the interface to atoms and an- rect electron addition to one of the bound oxygen atoms,
ions. The solution outside the inner double layer never
−
is exposed to an electron. Some examples of such inner- OMnVII (O)3 + e− −
OMnVI (O)2 O−
double-layer electron transfer include
E ◦ , +0.55 V vs NHE. (109)
− − ◦
H2 O + e [H·] + HO E , −2.93 V vs NHE
(101) The extent of the stabilization of the oxygen atom in
H+ +e −
[H·] + H2 O ◦
E , −2.10 V MnVII O−
4 is indicated by the reduction potential for a free
3O (102)
·O· atom,
(H2 O)25 + FeIII –OH + H+
3O+e
−
[·O·] + e− ·O− (E◦ )pH 14 , +1.43 V. (110)
pH 1 ◦
Fe II
(OH2 )2+ + H2 O E , +0.71 V (103)
H2 O 6
In summary, the electron-transfer reactions for metals
pH 1
[HO·] + H+
3O+e
−
H2 O
2 H2 O E◦ , +2.66 V (104) and metal complexes are ligand centered (or solvent cen-
tered). In each case the potential for the oxidation of free
ligand is decreased in the presence of metal or reduced-
−GBF (H2 O)2+
5 Fe –OH = [2.66 − 0.71] × 23.1
III
metal complex by an amount that is proportional to the
= 45 kcal mol−1 (105) metal-ligand bond energy (−G BF ).
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
180 Electrochemistry
Electrochemistry 181
Electrochemical measurement of pHa via Eq. (116) senses Substitution of Eq. (122) into Eq. (116) gives
hydronium-ion activity rather than its concentration.
E Pt = 0.059 log (K a )sol + 0.059 log [HA(sol)]/[A− (sol)]
Hence, electrochemical evaluations of dissociation con-
stants (K HA ) yield thermodynamic quantities − (0.059/2) log PH2 . (124)
−
HA + H2 O H+
3O+A K HA This equation in conjunction with voltammetric measure-
+ − ments of half-wave potentials (E 1/2 ) for the reduction of
K HA = H3 O [A ]/[HA]. (119)
Bronsted acids at a platinum electrode in any solvent per-
mits the evaluation of (K a ) sol [pK a (sol)],
1. Hydronium-Ion (H+
3 O) Reduction E 1/2 = 0.059 log (K a )sol + 0.059 log κ A− κH2 κHA
Although the NHE is fundamental to electrochemistry,
= −0.059 pK a + ε, (125)
it does not represent the primary electron-transfer step
for hydronium-ion reduction at an inert (glassy-carbon) where κ A− , κH2 , and κHA are parameters that relate to diffu-
electrode, sion coefficients, activity coefficients, and PH2 for a given
GC experimental system. Because solvent has such a major
H+
3 O(aq) + e
−
H·(aq) E◦ , −2.10 V vs NHE, and selective effect on the activity of H+ 3 O(aq), the differ-
(120) ences between concentration-based dissociation constants
(pK ) and activity-based constants (pK a ) are dramatic. For
where GC means a glassy-carbon electrode. The −2.10-V example, phenol in MeCN (pK , 26.6; pK a , 16.0) and in
difference in standard potential (E ◦ ) between the latter Me2 SO (pK , 16.4; pK a , 20.8) exhibits a reversal; it disso-
and that for the NHE [Eq. (115)] is due to the platinum ciates more in Me2 SO than in MeCN, but is more acidic
electrode, which stabilizes the hydrogen atom (H·) via [greater H+ 3 O(aq) activity] in MeCN. Other examples in-
formation of a Pt–H covalent bond, clude (Et3 NH)Cl in MeCN (pK , 18.5; pK a , 10.0) and in
H+
3 O(aq) + Pt (s) + e
−
Pt–H (s) E◦ , 0.000 V Me2 SO (pK , 10.5; pK a , 12.7) (again, greater dissociation
in Me2 SO and greater acidity in MeCN); PhC(O)OH in
(−G BF ) = [0.00 − (−2.10)] 96.48 kJ mol−1 (eV)−1 MeCN (pK , 20.7; pK a , 7.9), in Me2 SO (pK , 11.1; pK a
= 203 kJ mol−1 = 48 kcal mol−1 , (121) 13.6), in DMF (pK , 11.6; pK a , 11.5), and in H2 O (pK , 4.2;
pK a , 3.2); and H2 O in MeCN (pK a , 30.4) and in Me2 SO
where −G BF is the free energy of bond formation. (pK , 31.4; pK a , 36.7).
In this system the platinum electrode surface does not
consist of free platinum atoms, but must undergo ho- 3. Oxidation of Dissolved Dihydrogen (H2 )
molytic Pt–Pt bond breakage [HDBE , 24 kcal mol−1
per Pt· (est)] before a Pt–H bond is formed. Thus, the Molecular hydrogen (H2 ; HDBE , 104 kcal mol−1 ) is re-
Pt–H bond-formation energy (−G BF ) is estimated to be sistant to electrochemical oxidation at inert electrodes
72 kcal mol−1 on the basis of the electrochemical data [for (glassy carbon or passivated metals; Ni, Au, Hg, Cu). At
the gas-phase Pt–H molecule, HDBE ≤ 80 kcal mol−1 or passivated Pt and Pd, dissolved H2 only exhibits broad, dif-
(−G BF ≈ 72 kcal mol−1 )]. Determination of the reduc- fuse, anodic voltammetric peaks with irreproducible peak
tion potential for H+
3 O(aq) at other metal electrodes (M)
currents that are not proportional to the partial pressure of
provides a convenient means to estimate M–H bond ener- dissolved H2 (PH2 ). However, with freshly preanodized Pt
gies [(−G BF )]. and Pd electrodes, well-defined oxidation peaks for H2 are
obtained, which have peak currents that are proportional
to PH2 . The surface conditioning produces a fresh reactive
2. Brønsted-Acid (HA) Reduction and Evaluation
metal-oxide surface [PtII (OH)2 (s)], which upon exposure
of pK a(sol)
to H2 becomes an oxide-free metal surface (Pt∗ ). In turn,
Brønsted acids (HA) undergo dissociation in any solvent the clean surface reacts with a second H2 to form two Pt–H
to yield the solvated aqua-hydronium ion [(sol)n H+3 O(aq)] bonds,
{from residual H2 O; or [(sol)n Hsol+ ] for basic solvents}, fast
2 Pt∗ (s) + H2 PH2 2 Pt–H (s) Keq , ∼1 atm−1 .
which further dissociates to H+ 3 O(aq) and the solvated
conjugate base [A− (sol)n ] (126)
HA (sol) + H2 O H+ −
3 O(aq) + A (sol) (Ka )sol The value of K eq is estimated on the basis of the Pt–H
(122) bond-formation energy from metallic platinum
and [(−G BF ), 48 kcal mol−1 ] and the dissociative
H+ −
3 O(aq) = (Ka )sol [HA (sol)]/[A (sol)]. (123) bond energy for H2 (G DBE , ∼96 kcal mol−1 ).
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
182 Electrochemistry
This activated platinum electrode system is the equiv- sensors for ·O2 · are the most common analytical method-
alent of an NHE and, therefore, conforms to the Nernst ology for its assay in blood, gas streams, biotreaters, and
expression of Eq. (116). Rearrangement gives process streams.
2
log PH2 = −E Pt (2/0.059) + log H+
3 O(aq) , (127)
1. Molecular Oxygen
which responds to the overall reaction
The ground state of molecular oxygen (·O2 ·, 3 − g ) has
K eq two unpaired electrons in degenerate 2πg orbitals and is re-
H2 PH2 + 2 Pt∗ (s) 2 Pt–H (s) ferred to as dioxygen by most contemporary biologists and
−2e − biochemists. When dioxygen is reduced by electron trans-
2 H+
3 O(aq) + 2 Pt (s). (128) fer, a series of intermediate basic dioxygen and monooxy-
H2 O
gen species are produced that may take up one or two hy-
Although there is a long-standing tradition to write an dronium ions (H+ −
3 O) from the media (O2 ·, HOO·, HOO ,
−
− −
electron-transfer sequence with electron removal (ioniza- HOOH, ·O , HO·, HO , and H2 O). The thermodynamics
tion) from the hydrogen atom to produce a proton [Pt–H (s) for the various reduction steps in aqueous solutions have
→ Pt + H+ + e− ] followed by hydration to give the ob- been evaluated by numerous techniques, but all are fun-
served hydronium ion [H+ + H2 O → H+ 3 O(aq)], the ion- damentally based on the calorimetry associated with the
ization potential for a free hydrogen atom (H·) is 13.6 eV reaction
and even greater for bound hydrogen [Pt–H (s)]. Electron
removal from an aqueous solution at pH 0 with a GCE ·O2 · (g) + 2 H2 (g) 2 H2 O (l) −G◦ = nE◦cell F.
(free of H2 and Pt–H) occurs via the lowest energy path, (131)
which is oxidation of the solvent
If redox potentials are relative to the normal hydrogen
−e− H2 O
H2 O H+
2 O· H+
3 O + HO·
electrode,
E ◦ , +2.72 V vs NHE. (129) 2 H+
3 O + 2e
−
H2 (g) E◦H+ O/H2≡0.000, (132)
3
In the presence of Pt–H (s) (and H2 ), this process is facil- then the standard redox potential for the four-electron re-
itated via direct formation of an H–OH bond, duction of dioxygen (E O◦ 2 /H2 O ) can be calculated from
−e− the calorimetric data for the reaction of Eq. (131) under
H2 O + Pt–H (s) Pt (s) + H+
3 O + HO–H E ◦ , 0.00 V standard-state conditions,
H2 O
anions and radicals, quinones) is its most important and via reduction, hydrolysis, and disproportionation steps.
unique application. In general, electrochemical measure- Thus, the most effective means to activate ·O2 · is the addi-
ments provide the only direct means for the evaluation of tion of an electron (or hydrogen atom; H+ −
3 O + e → H·),
the electron-transfer thermodynamics of oxygen species which results in significant fluxes of several reactive oxy-
and oxygen-containing molecules. Also, amperometric gen species.
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry 183
TABLE IV Standard Reduction Potentials for Dioxygen Species in Water [O2 , 1 atm (1 mM)] (Formal Potentials for 1 M O2 )
1.66 V vs NHE
pH 0 (1 M H3O)
1.13
0.73 (0.79)
1.229 (1.27)
1.20
pH 7
0.64
0.31 (0.37)
0.815 (0.86)
0.65
pH 14 (1 M HO)
0.09
0.03 (0.01)
0.401 (0.44)
2. Aprotic Media dia decrease. The heat of hydration (−Haq ) for gaseous
O− −1
2 · is 100 kcal mol , which is consistent with the unique
In the absence of proton sources dioxygen is reversibly
strong solvation of anions by water. Hence, if the E O◦ /O− ·
reduced to superoxide ion,
values for the ·O2 ·/O−
2
2 · couple are affected primarily by
2
−
·O2 · + e− O−
2· E◦ , −0.64 V vs NHE. (134) the degree of solvation of O2 · (that is, the solvation en-
The second reduction is an irreversible one-electron ergy for ·O2 · is assumed to be small and about the same
process, for different solvents), then the relative solvation energies
H2 O
for O− 2 · are H2 O Me2 SO > DMF > py ∼ MeCN.
O−
2·+e
−
HOO− + HO− Ep,c , −1.8 V (135)
− − −
HO + HOO + ·O2 · 2 O 2 · + H2 O 3. Hydrogen Peroxide (HOOH, H2 O2 )
−9 −1 −1
K , 10 M atm . (136)
The electrochemical reduction of HOOH in pyridine
The reduction potential for the ·O2 ·/O−2 · couple shifts to yields a superoxide ion. That a reduction process gen-
more negative values as the solvating properties of the me- erates a species with a higher oxidation state than HOOH
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
184 Electrochemistry
TABLE V Formal Reduction Potentials for 1 M Dioxygen Species in Acetonitrile [O2 at 1 atm (8.1 mM)]
2.17 V vs NHE
pH 8.8
1.64 [1 M (H3O)ClO4]
1.30 2.28
1.31
1.79
1.06
pH 10.0
[1:1 (Et3NH)Cl/Et3N]
0.64
0.19 1.17
0.20
0.68
0.50
pH 30.4
[1 M (Bu4N)OH]
1.04
0.95 0.11
0.90
0.53
is surprising. The cyclic voltammogram for the reduc- oxide ion. The products and the observed electron stoi-
tion of HOOH in pyridine at a platinum working elec- chiometries for the electrochemical reduction of HOOH
trode exhibits a broad cathodic peak at −0.95 V vs NHE are consistent with a mechanism in which the primary step
and anodic peaks for the reverse scan at −0.50 and is a one-electron transfer,
−0.15 V vs NHE. The latter is characteristic of elec-
trolytically generated H2 . Controlled-potential reduction HOOH + e− HOO− + 1/2 H2
of HOOH in pyridine at −1.0 V vs NHE (with argon de- E◦ , −1.0 V vs NHE (137)
gassing) results in a solution that exhibits an anodic cyclic
voltammogram that is characteristic of O− 2 ·. ESR studies followed by a chemical reaction with another hydrogen
of the reduced solution confirm the presence of a super- peroxide molecule.
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry 185
HOOH + HOO− [HOOOH] + HO− Some of the considerations for electron-transfer pro-
py cesses that have been discussed are fundamental to
k,3 × 109 M−1 s−1
1/n [·pyOH]n + H2 O + O−
2· (138) the electrochemistry of these examples. Thus, reduc-
tive processes always involve the most electrophilic
The resultant HO· is trapped by the pyridine solvent to
(acidic, positive-charge density) center (substrate or sub-
yield a stable solution of O− 2 ·. strate/matrix combination) that produces the least basic
In acetonitrile there is no evidence of a superoxide ion,
(nucleophilic) product. Under acidic conditions the pri-
either by ESR or by cyclic voltammetry, from the elec-
mary reactant often is the hydronium ion (H+ 3 O) to give a
trochemical reduction of HOOH. This can be explained
hydrogen atom that couples with the substrate via covalent
by the slow rate of reaction of HO· with CH3 CN (k, 4 ×
bond formation, e.g.,
106 M−1 s−1 ), which favors direct disproportionation.
[HOOOH] + HO− H2 O + [HOOO− ] PhN NPh + 2 H+
3 O + 2e
−
186 Electrochemistry
e
2. Sulfur Dioxide (SO2 ) 2 PhNHNHPh PhNNHPh R NH2
(R NH) Ep,a, 0.6 V vs SCE
Electrochemistry 187
El + e− El − ·
CO2
Hg2(l) HC(O)O HO
E red (159)
[O(O)C HgII HgII C(O)O(s)] CO2 Often, the solution matrix contains Lewis acids (e.g.,
HOC(O)O
H+
3 O) that are more electrophilic than the substrate
Hg2(l) CO CO 3
2 molecule and, in combination, are even more electrophilic,
(157) e, H3O
El H3O e [HEl] HE lH (Ered)EC.
2 HO
to be the result of an irreversible two-electron reduction
process. In most cases it is an ECEC process in which each
2
CO3 2 H 2O. electron transfer (E part of the ECEC mechanism) to the
Lewis acid (H+ 3 O) is reversible to give a product (H·) that
(158)
forms a covalent bond with the substrate (H–El) (the C
Again, the activated gold surface stabilizes the carbon part of the mechanism).
radical intermediate [·C(O)OH] and facilitates the second Conversely, nucleophilic molecules (Nu) [Lewis bases;
electron-transfer oxidation of HO− to HO· via coupling e.g., catechols, hydroquinones, phenols, alcohols, and thi-
to the carbon radical. ols (and their anions); aromatic hydrocarbons and amines
(benzene, toluene, pyridine, bipyridine, etc.)] can be oxi-
dized (1) by direct electron-transfer oxidation [Eq. (161)]
D. Organic Molecules [with the electron coming from the Highest-Occupied-
Most organic reactions are Lewis acid/base processes that Molecular-Orbital (HOMO)] or (2) by coupling with the
involve the interaction of a nucleophilic center with an oxidation product of H2 O (or HO− ), hydroxyl radical
electrophilic center. Because electrochemistry provides (HO·) [Eq. (162)].
the ultimate nucleophile via the electrons at the cathode
−e−
surface and the ultimate electrophile via the electron holes Nu: Nu+ · E ox (161)
at the anode surface, it is the ideal methodology for the
characterization of the electrophilicity and nucleophilic- The potential (E ox ) for those nucleophiles that undergo
ity of molecules. Thus, the carbon centers of saturated direct electron-transfer oxidation at an inert electrode is
hydrocarbons (e.g., CH4 ) are resistant to electrochem- a quantitative measure of their nucleophilicity (the more
ical reduction and oxidation because of their inert na- negative the potential the more nucleophilic the molecule;
ture (all valence electrons are stabilized in sigma bonds; see Table VI for representative values). In many cases wa-
an absence of any Lewis acid/base character). However, ter in the solvent matrix (or as the solvent) is more nucle-
organic molecules with electrophilic components [e.g., ophilic than the substrate molecule and, in combination,
alkyl-, aryl-, and acyl- halides; carbonyl groups; unsatu- is even easier to oxidize, which often results in an ECEC
rated and aromatic hydrocarbons; nitro groups; Brønsted oxidation process.
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
188 Electrochemistry
−
eaq −3.9 −2.9 H2 O+ · +3.2 +2.7
K· −2.9 −2.9 Ph+ ·CH2 OH +2.2
Na· −2.1 −2.7 AuI (OH2 )+
6 +1.6 +1.8
(Cl8 TPP+ ·)FeIV O +1.5
(Compound I)
Li· −2.0 FeIII (bpy)3+
3 +1.3 +1.1
(TPP− ·)Co− −1.7 HO· +0.9 +1.9
H· −1.6 −2.1 FeIII (PA)3 +0.4
(TPP− ·)Fe− −1.4 (TPP)FeIII (py)+
2 +0.4
(TPP)Fe− −0.8 (TPP)FeIII Cl +0.2
(TPP)Co− −0.6 MV2+ −0.2
O−
2· −0.7 −0.2 (Cl8 TPP)FeIV O −0.3
(Compound II)
PhCH2 S− 0.0 AQ (Anthraquinone) −0.6
HOO− 0.0 +0.7 O2 −0.7 −0.2
PhO− +0.3 CCl4 −0.9
Me3 N +0.7 PhCH2 Br −1.4
HO− +0.9 +1.9 PhCl6 −1.4
MeC(O)O− +1.5 t-BuI −1.5
PhOH +1.7 H3 O+ −1.6 −2.1
pyridine +2.0 PhCH2 Cl −1.7
Cl− +2.2 +2.4 n-BuI −1.9
HOOH +2.3 +1.0 c-C6 H11 Br −1.9
H2 O +3.0 +2.3 t-BuBr −2.0
n-BuBr −2.2
n-BuCl −2.5
PhCl −2.6
H2 O −3.9 −2.9
e e, 2 H2O substrate and coupling (R–R) or (b) to a Brønsted acid
Nu: 2 H2O [HO Nu] Nu(OH)2 (Eox)EC
H3O H3O (HA; H2 O) to replace the C–X bond with a C–H bond.
For example,
(162)
1. Alkyl- and Aryl-Halides
Because the halo-groups of organic molecules have large
electronegativities and electron affinities, all halo-carbon n-BuI e [n-BuI] Ep,c, 2.05 V vs SCE
(a) e
molecules are electrophilic. Their electrochemical re- n-BuI,
n-Bu–Bu-n 2 I
duction potential is a measure of their electrophilicity (b) H2O, e
n-BuH I HO
(and electron affinity), which is illustrated in Fig. 6 for
(163)
hexachlorobenzene (C6 Cl6 ), 1,2,3,4-tetrachlorobenzene
(1,2,3,4-C6 H2 Cl4 ), and n-butyl iodide (n-BuI). Table VII In both cases the overall process is an irreversible two-
summarizes the reduction potentials for several alkyl- electron reduction via either (a) an EE path or (b) an ECEC
halides and aryl-chlorides. path; the first electron transfer is the most difficult and
In the absence of other Lewis acids, alkyl-halides un- depends on the substrates electrophilicity. In the presence
dergo direct electron addition at the electrode surface with of an hydronium ion, the primary electron transfer will be
subsequent stimulated electron transfer (a) to a second to the most electrophilic center, e.g.,
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
Electrochemistry 189
PhCl6 e [PhCl
6]
e, H2O
HPhCl5 Cl HO
(165)
TABLE VII Redox Potential for Alkyl Halides (RX ) and Aryl
Chlorides (Ar Clx in Dimethyl Formamide at a Glassy Carbon
Electrodea
Alkyl halides
E p,c ,b V vs SCEc
RX I Br Cl
CH3− −2.10
n-C4 H9− −2.05 −2.41
sec-C4 H9− −1.92 −2.35
t-C4 H9− −1.78 −2.25
c-C6 H11− −2.03 −2.48
PhCH2− −1.68 −1.90
ClH2 C− −2.05
Cl2 HC− −1.99
FIGURE 6 Cyclic voltammograms for chlorinated aromatic
molecules and n-butyl iodide in dimethylformamide (0.1 M TEAP) Cl3 C− −1.13
at a glassy-carbon electrode (area, 0.062 cm2 ): (a) 1.1 mM hex- FCl2 C− −1.71
achlorobenzene (C6 Cl6 ); (b) 2.3 mM 1,2,3,4-C6 H2 Cl4 ; (c) 20 mM (F3 C)Cl2 C− −1.31
n-BuI. PhCl2 C− −1.47
[( p-ClPh)2 HC]Cl2 C− (DDT) −1.50
[( p-ClPh)2 FC]Cl2 C− (F-DDT) −1.44
H3O n-BuI e [n-BuIH ]
Alkyl halides
H2O H3O, e
n-BuH HI .
E p,c , V E p,c , V
H2O Ar Clx vs SCE Ar Clx vs SCE
(164) PhCl(C6 H5 Cl) −2.7 PhCl5 −1.6
Thus, the reduction of n-BuI is the equivalent of the addi- 1,2-PhCl2 −2.5 PhCl6 −1.4
tion of two hydrogen atoms [H·] (generated via the elec- 1,2,3-PhCl3 −2.2 Cl5 Ph-PhCl5 −1.5
trochemical reduction of the two hydronium ions). 1,2,3,4-PhCl4 −1.9
With aryl-chlorides (e.g., PhCl6 ) a similar EEC process a mM solutions in DMF [0.1 M (Et4 N)ClO4 ].
occurs as each chlorine atom is replaced with a hydrogen b E p,c, the reduction-peak potential.
atom (Fig. 6 and Table VII). c Saturated calomel electrode (SCE) vs NHE, +2.44 V.
P1: GRB Final Pages
Encyclopedia of Physical Science and Technology EN005M-206 June 15, 2001 20:25
190 Electrochemistry
2. Quinones, Semiquinones, and Catechols Neutral catechols (H2 Cat) and hydroquinones (H2 Q) are
much more resistant to electron removal because of their
All molecules with unsaturated bonds (olefins, acetylenes,
release of a proton via the solvent matrix; the more basic
aromatics, carbonyls, quinones, etc.) have a degree of
the solvent, the less positive the oxidation potential.
electrophilicity and electron affinity. Within a class, the
extent of conjugation increases the electron affinity (re-
e e
duction of benzene occurs at a less negative potential H2Cat [HSQ] Q (Ep,a)DMF, 0.97 V
sol sol
than 1-butene), and the presence of unsaturated carbon-
Hsol
Hsol (Ep,a)AN, 1.19 V
oxygen (carbonyl) functions within a conjugated system
(e.g., quinones; O ) also enhances the electron affinity
O
(173)
of the molecule. In acetonitrile, the residual H2 O is the stronger base, which
The cyclic voltammogram for 3,5-di-tert-butyl-o- results in the formation of H+ 3 O.
quinone (3,5-DBTQ) has a first reduction that is a re- The electrochemistry of quinones is surprisingly sim-
versible one-electron process, followed by a second one- ilar to that of dioxygen. It is as if a conjugated carbon
electron reduction, which can be reversible in rigorously link is inserted between two oxygen atoms (Q → SQ− · →
anhydrous media to give catechol dianion (3,5-DTBC2− ). HCat− vs ·O2 · → O− −
2 · → HOO ; Q → H2 Cat vs ·O2 · →
HOOH).
e−
3,5-DTBQ + e− 3,5-DTBSQ− . 3,5-DTBC2−
(167)
3. Carbonyl Groups, Olefins,
Although DTBC2− is a strong base that is hydrolyzed by and Aromatic Hydrocarbons
residual H2 O,
Unsaturated carbon centers possess a limited degree of
− − electrophilicity and will accept an electron at potentials
DTBC 2−
+ H2 O DTBCH + HO , (168)
significantly more negative than their chloro derivatives
it is also a strong reductant that can reduce H2 O in DMF. (usually at least −2.5 V vs SCE in rigorously anhydrous
solvents). When water is present, its reduction is syner-
DTBC2− + H2 O DTBSQ− · + 1/2 H2 + HO− gistically facilitated via unsaturated carbon. For example,
(169) in acetonitrile at a glassy-carbon electrode,
In many cases, the second reduction step of quinones
[Eq. (167)] is irreversible and due to the facilitated reduc- H2O e [H] HO Ep,c, –3.9 V vs SCE
tion of residual H2 O.
(174)
3,5-DTBSQ− · + H2 O + e− 3,5-DTBCH− + HO−
(170) e NP Ep,c, 2.6 V vs SCE
+
In the presence of hydronium ions (H3 O), the reduction
of the quinones is an irreversible two-electron process (175)
(ECEC), with the first step being the more difficult (re-
e
quiring the more negative potential). NP H2O e [HNP] H2O H2NP
3,5-DTBQ H 3 O e [ 3,5-DTBSQH ] HO HO
e, H3O
H2O 3,5-DTBCH2 Ep,c, ~ 2.0 V vs SCE (176)
H2O
Electrochemistry 191
E. Organometallic Molecules
192 Electrochemistry
FIGURE 7 Cyclic voltammograms: (a) 3 mM [FeII (MeCN)4 ](ClO4 )2 ; (b) 3 mM FeIII Cl3 ; (c) 3 mM FeIII (acac)3 ; (d)
3 mM FeII (Cp)2 ; (e) 3 mM FeVIII (CO)5 in MeCN (0.1 M tetraethylammonium perchlorate [TEAP]). Conditions: scan
rate, 0.1 V s−1 ; ambient temperature; glassy-carbon working electrode (area, 0.09 cm2 ); saturated calomel electrode
(SCE) vs NHE, +0.244 V.
Electrochemistry 193
194 Electrochemistry
Electrochemistry 195
(1.3 vs 2.0 V), (3) a constant cell voltage during discharge, The alkaline cell has lower internal resistance, longer shelf
and (4) almost unlimited life. In contrast to the sulfuric- life, and a larger open-circuit voltage (about 1.9 V) than
acid electrolyte of the lead-acid battery, the Edison battery the dry cell.
uses concentrated potassium hydroxide (KOH) as the elec-
troyte. The cathode is nickel dioxide [NiIV O2 (s)] and the D. High-Performance Batteries
anode is metallic iron. As the cell reaction confirms, the (Nickel/Cadmium, Silver/Zinc, Lithium)
KOH concentration remains constant during discharge,
which accounts for the constant cell voltage during dis- 1. Nickel/Cadmium
charge (only water is consumed during discharge and re- This is the most common of the rechargable batteries and
stored during recharge). has extensive use in “cordless” power tools, electric razors,
Cathode (+ terminal): NiIV O2 (s) + 2 H2 O + 2e− and electronics. It has similarities to the Edison cell, but
better performance in its areas of application.
NiII (OH)2 (s) + 2 HO− (200)
Cathode: NiIV O2 (s) + 2 H2 O + 2e−
Anode (−terminal): Fe (s) + 2 HO−
NiII (OH)2 (s) + 2 HO− (209)
FeII (OH)2 (s) + 2e− (201)
Anode: Cd (s) + 2 HO− CdII (OH)2 (s) + 2e− (210)
Cell: NiIV O2 (s) + Fe (s) + 2 H2 O
Cell: NiIV O2 (s) + Cd (s) + 2 H2 O
NiII (OH)2 (s) + FeII (OH)2 (s) (202)
NiII (OH)2 (s) + CdII (OH)2 (s) (211)
C. “Dry-Cell” Batteries The main problem with Ni/Cd batteries is their reduced
charge capacity if they are not fully discharged before
1. LeClanché Cell
recharging. Because of the toxicity of cadmium, responsi-
This traditional dry cell consists of a carbon-rod cathode ble disposal of worn-out batteries is important (but often
(positive terminal) immersed in a moist paste of MnIV O2 , ignored).
ZnII Cl2 , NH4 Cl, and powdered carbon, which is contained
in a metallic zinc-can anode (negative terminal). The volt- 2. Silver/Zinc
age (without load) of these cells is about 1.6 V, which have
limited shelf life because of corrosion of the zinc can and Forms of this battery type range from those used in hear-
increased internal resistance. ing aids, watches, and cameras to those used as portable
power sources in space vehicles. The cathode/anode and
Cathode: 2 MnIV O2 (s) + H2 O+2e−
electrolyte system is similar to that for the alkaline cell,
−
2 O3 (s) + 2 HO
MnIII (203) except the MnIV O2 cathode material is replaced with much
− more expensive silver oxide (AgI2 O). However, it has high
Anode: Zn (s) + 2 NH4 Cl + 2 HO
stability, a long shelf life, and a large charge density with
ZnII (NH3 )2 Cl2 + 2 H2 O+2e− (204) a stable voltage (about 1.8 V).
Cell: 2 Mn O2 (s) + Zn (s) + 2 NH4 Cl
IV Cathode: AgI2 O (s) + H2 O + 2e− 2 Ag (s) + 2 HO−
196 Electrochemistry
Cathode: MnIV O2 (s) + LiX + e− modynamically favored reactions are accelerated by the
presence of salts and acids to cause rapid corrosion:
MnIII (O)OLi (s) + X− (215)
Fe (s) + 2 H2 O/CO2 6 + H2
FeII (OH2)2+
Anode: Li (s) + X− LiX + e− (216)
+ 2 HO(O)O− , (221)
Cell: MnIV O2 (s) + Li (s) MnIII (O)OLi (s) (217)
2 Fe (s) + ·O2 · + 4 H2 O/CO2 + 10 H2 O 2 Fe (OH2)2+
II
6
All metals (M) react with atmospheric oxygen (·O2 ·) to + 2 H2 O/CO2 K (227)
form surface films of metal oxides (MOx ). When this film ◦
◦ ◦
E cell = E Fe(II)/Fe − E Al(III)/Al
is formed under controlled conditions, it produces an in-
ert ( passivated ) surface that precludes further reaction = −0.44 V − (−1.83 V) = 1.39 V
and corrosion. However, the oxide films on copper alloys ◦
E cell = [0.05915/n] log K
and structural steel undergo dissolution when exposed
to aqueous media that contain ·O2 · and salts, acids, or log K = [6/0.05915]1.39 = 141; K = 10141
bases, and the surface no longer is protected and corrodes
Hence, the steel structure is protected until the aluminum
(dissolves).
anodes are consumed.
Electrochemistry 197
ELECTRON TRANSFER REACTIONS • FUEL CELLS, APPLI- Bard, A. J., Parsons, R., and Jordan, J., eds. (1985). “Standard Potentials
STATIONARY POWER SYSTEMS
CATIONS IN in Aqueous Solution,” Dekker, New York.
Dryhurst, G., Kadish, K. M., Scheller, F., and Renneberg, R. (1982).
“Biological Electrochemistry,” Vol. 1, Academic Press, New York.
Kissinger, P. T., and Heineman, W. R., eds. (1996). “Laboratory Tech-
niques in Electroanalytical Chemistry,” 2nd ed., Dekker, New York.
BIBLIOGRAPHY Kolthoff, I. M., and Lingane, J. J. (1952). “Polarography,” 2nd ed., In-
terscience, New York.
Baizer, M. M., and Lund, H., eds. (1991). “Organic Electrochemistry,” Sawyer, D. T., Sobkowiak, A., and Roberts, J. L., Jr. (1995). “Electro-
3rd ed., Dekker, New York. chemistry for Chemists,” 2nd ed., Wiley-Interscience, New York.
Bard, A. J., and Faulkner, L. R. (2001). “Electrochemical Methods,” Weinberg, N. L., ed. (1974, 1982). “Technique of Electroorganic Syn-
2nd ed., Wiley, New York. thesis,” Parts I, II, and III, Wiley-Interscience, New York.
P1: GLQ/LSK P2: FQP Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
I. Basic Principles
II. Experimental Aspects
III. Isotropic Hyperfine Analysis
IV. Relation of Hyperfine Constants
to Spin Densities
V. Second-Order Hyperfine Effects
VI. Anisotropic Hyperfine Interaction
VII. g Anisotropy
VIII. Spin Relaxation
IX. Double-Resonance and Time-Domain ESR
X. Applications
XI. New Developments
331
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
and transition-metal and rare-earth ions. The specificity and gives the unit of spin angular momentum. The two en-
of electron spin resonance spectroscopy for only species ergy levels associated with an electron in a magnetic field
containing unpaired electrons is particularly valuable for are thus given by + 12 gβH and − 12 gβH . The difference
the study of chemical reaction intermediates. between these two energy levels is gβH , so that the tran-
sition energy is given by h ν = gβH . This basic electron
spin resonance transition is illustrated in Fig. 1.
I. BASIC PRINCIPLES Typically, electron magnetic resonance is carried out in
a magnetic field of about 3000 G (gauss) or 0.3 T (tesla).
In general, the energy of a system or, more specifically, This corresponds to an energy absorption frequency of
of a molecule depends on momentum. The kinetic energy about 9 GHz. This frequency is in the microwave range.
of a molecule is proportional to the square of the linear As for any spectral transition, the number of systems or
momentum, and the rotational energy of the molecule is electrons in the upper and lower energy states at thermal
proportional to the square of the rotational or angular mo- equilibrium is given by a Boltzmann distribution.
mentum. An isolated electron can be regarded as a point
particle that has no classical angular momentum. How-
ever, experimentally it is found that isolated electrons in II. EXPERIMENTAL ASPECTS
a magnetic field do absorb a quantitized amount of en-
ergy, which means that they must have at least two energy Figure 2 shows a block diagram of a typical electron spin
levels. These are not translational energy levels, because resonance (ESR) spectrometer. It can be seen that this
the amount of energy absorbed does not depend on the diagram is analogous to that for an optical absorption
kinetic energy of the electron. However, the magnitude spectrometer, consisting of a source of electromagnetic
of the energy absorbed does depend on the magnitude of optical radiation, a sample cavity, and a detector of the op-
the magnetic field to which the electrons are exposed. As tical radiation intensity. In the block diagram of the ESR
an explanation of the existence of these magnetic energy spectrometer, the microwave irradiation is produced by a
levels, it is postulated that an electron has an intrinsic angu- special microwave tube called a klystron or a solid-state
lar momentum called spin angular momentum. When this device called a Gunn diode. The microwaves are transmit-
spin angular momentum interacts with a magnetic field, ted through a rectangular waveguide or through a special
two different energy levels are produced whose difference coaxial cable to an isolator, which is a ferrite device al-
accounts for the absorption of energy by the unpaired elec- lowing microwaves to travel in only one direction. This
tron system. component is used so as to prevent reflection of the mi-
The Hamiltonian energy operator for the electron spin crowaves back into the klystron tube or Gunn diode. The
transition we have just discussed is given by microwaves then pass through a device called an attenua-
tor, which simply controls the intensity of the microwaves
spin = gβS · H, (1)
and consists of a resistive but electrically conducting ele-
where spin is the spin Hamiltonian energy operator; S ment that penetrates to different extents into a waveguide.
is the spin angular momentum; H is the magnetic field; Although not shown, there also are devices available to
gβ is a proportionality constant, where g = 2.0023 and is shift the phase of the microwaves if that is necessary. The
called the g factor or spectroscopic splitting factor, which microwaves then pass into a circulator, which is a ferrite
is dimensionless; and β is the Bohr magneton, which has a device that routes the microwaves in one port and out an
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
FIGURE 2 Block diagram of a typical ESR spectrometer. KLY, klystron or Gunn diode source of microwaves; ISOL,
isolator; ATTEN, attenuator; Circ, three-port circulator; CAV, resonant cavity; M, magnetic field modulation coils;
XL, detector crystal; PSD, phase-sensitive detector; and OSC, oscillator. The double lines represent a waveguide
connecting the microwave components, and the single line represents a cable connecting the electronic components.
adjacent port in a unidirectional fashion. This component signal is then passed into the signal channel of the phase-
is particularly useful for a resonance-type experiment in sensitive detector, and this device compares the coding of
which all of the microwaves are absorbed in the sample the signal with that of the reference and passes only signals
cavity. When this equilibrium situation is upset by a sam- coded at the same frequency. This has the effect of greatly
ple absorbing some of the microwaves in the cavity, an im- increasing the signal-to-noise ratio, since noise generated
balance is set up and some microwaves are reflected from in the system is not coded at the magnetic field modula-
the cavity out through the circulator in a unidirectional tion frequency. The phase-sensitive detection procedure
fashion to the detector crystal. In the diagram the ESR also has the effect of differentiating the ESR signal and
cavity shown is a reflection type, which is typically more producing a first-derivative curve instead of an absorption
sensitive than a transmission type, which would be more curve. The signal from the phase-sensitive detector then
analogous to the sample cell of an optical spectrometer. goes to a recorder or another readout device and is usually
The microwaves are detected by a semiconducting crys- referenced to a swept magnetic field so that one obtains
tal diode rectifier mounted in a waveguide. This crystal ESR intensity as a function of magnetic field.
converts the high-frequency microwave electromagnetic
radiation to a direct current voltage, which can then be III. ISOTROPIC HYPERFINE ANALYSIS
amplified and treated by ordinary electronics. For opti-
mum sensitivity, the detector crystal is biased with a little The real power of electron spin resonance spectroscopy
reflected microwave power controlled by the size of the for structural studies is due to interaction of the unpaired
iris, which is the coupling hole allowing the microwaves electron spin with nuclear spins in molecular species. This
to enter the cavity. The size of this coupling hole can be gives rise to a splitting of the energy levels and gener-
simply changed with a screw tip. ally allows the determination of the atomic or molecular
So that the signal-to-noise ratio can be increased greatly, structure of the radical species. In this situation the spin
by approximately a factor of a thousand, magnetic field Hamiltonian of Eq. (1) involves additional terms corre-
modulation is used. This is accomplished with small coils sponding to the nuclear spin interacting with the magnetic
mounted on each side of the sample cavity through which field, to the nuclear spin interacting with the electron spin,
radiofrequency radiation, typically 100 kHz, is passed. and, if the nuclear spin is ≥1, to a nuclear quadrupole in-
This produces a modulation of the static magnetic field, teraction. This more complete spin Hamiltonian may be
which codes the ESR signal at this modulation frequency. written as
Then a device known as a phase-sensitive detector is used,
which is referenced to this magnetic field modulation fre- spin = βH · g · S + hS · An · In − gn βn H · In
n n
quency. The microwave signal is coded at this modulation
frequency, and the crystal detector converts the signal into +h In · Qn · In . (2)
a pulsating dc voltage at this modulation frequency. This n
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
In Eq. (2) the summations are taken over all the nuclei
in the molecular species. The new symbols in Eq. (2) are
defined as follows: gn is the nuclear g factor, which is di-
mensionless; βn is the nuclear magneton, having units of
joules per gauss or per tesla; the nuclear spin angular mo-
mentum operator In ; the electron–nucleus hyperfine ten-
sor An ; the quadrupole interaction tensor Qn ; and Planck’s
constant h.
Note that the nuclear Zeeman term involving the inter-
action of the magnetic field with the nuclear spin angular
momentum operator has a negative sign. This is essen-
tially due to the difference in charge between electrons
and nuclei. Also note that the nuclear g factor can be either
FIGURE 3 Schematic of the first-order spin energy levels of a hy-
positive or negative, while the electron g factor is taken drogen atom, showing successive interactions in the spin Hamil-
as intrinsically positive. In Eq. (2) the electron g factor tonian, the allowed ESR transitions, and the spin wave functions.
has been written in tensor form involving a 3 × 3 matrix
that connects the magnetic field vector and the electron energy levels given in Eq. (5), where mS and mI refer to
spin angular momentum vector. Similarly, the hyperfine the electron and nuclear spin angular momentum quantum
interaction is written in tensor form connecting the elec- numbers:
tron spin and nuclear spin angular momentum vectors. The
E n = gβ H m S − gn βn H m I n + m S h An m I n .
quadrupole interaction is also written in tensor form. n n
First consider the special case of isotropic hyperfine (5)
interaction in which the hyperfine interaction becomes a The wave functions correct to first-order perturbation the-
scalar and can be written in front of the dot product of the ory are just the product functions of the respective electron
nuclear and electron spin angular momentum vectors. For and nuclear spin combinations. The hyperfine interaction
simplicity the electron g factor will also be considered to term involves only the z-components of the electron and
be isotropic and to be a scalar. This simplification typically nuclear spin angular momentum operators when treated
applies to most organic and some inorganic free radicals by first-order perturbation theory.
in liquids and also to a few cases in solids. This occurs This simplified treatment can be applied exactly to a
because rapid tumbling of the molecular species averages hydrogen atom with S = 12 and I = 12 where the corre-
out the anisotropic interactions. Also, since the quadrupole sponding mS and mI values are both ± 12 . The spin wave
interaction is typically small and can only be experimen- functions such as αe βn and the spin energy levels for a
tally resolved in special cases, it will be left out of the hydrogen atom in a magnetic field are shown in Fig. 3.
simplified spin Hamiltonian. The resulting simplified spin The transition probabilities between the first-order spin
Hamiltonian becomes energy levels can be calculated from time-dependent per-
spin = gβH · S − gn βn H · In + h An S · In . (3) turbation theory by standard methods in which the mag-
n n netic dipole moment operator is used. It is found that the
The external magnetic field is unidirectional, and by con- transition moment is only finite for a magnetic dipole mo-
vention it is taken to be in the z-direction. ment operator oriented perpendicular to the magnetic field
Then the energy levels of the spin system are given by direction. This means that in the design of an electron spin
resonance cavity into which the sample is placed, the mi-
E n = n∗ spin n dτ, (4) crowave magnetic field must be arranged to be perpendic-
ular to the external static magnetic field. The second point
where represents the spin wave functions. The spin wave of interest is that transition moments that determine the se-
functions can be taken as products of the electronic and lection rules for magnetic resonance are finite for electron
nuclear spin wave functions. These product functions are spin transitions corresponding to a change in the electron
not exact wave functions for the spin Hamiltonian includ- spin orientation but are zero for nuclear spin transitions.
ing the hyperfine interaction term, but they serve as good These selection rules can be compactly written as
first approximations. The correct treatment of the prob-
lem usually involves some approximation method, and a
mS = ±1 (6)
common one is perturbation theory, which is a standard
and
quantum-mechanical method. Applying first-order pertur-
bation theory to the spin Hamiltonian in Eq. (3) gives the
mI = 0, (7)
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
and 7. After the hyperfine interaction of one nucleus is In spectral analysis, the hyperfine coupling constant A has
considered, all the individual lines are no longer sepa- been treated as an experimental parameter with units of
rated. The superposition leads to fewer lines and a vari- megahertz. The theoretical expression relating the hyper-
able intensity distribution. The splitting between all the fine coupling constant for atoms is given by
lines in the observed spectrum is identical and is equal
h A = 83 πgβg N β N |ψ(0)|2 , (8)
to the coupling constant. The resulting intensity distribu-
tions are characteristic for sets of equivalent nuclei and where the units are joules on both sides of the equation
often simplify analysis. For protons or any nucleus with and |ψ(0)|2 is the probability density of the electron’s be-
spin 12 , n equivalent nuclei give n + 1 lines with a bino- ing at the nucleus. From a quantum-mechanical point of
mial intensity distribution. In the general case the number view, the electron is in contact with the nucleus, and hence
of lines arising from n equivalent nuclei of spin I is given the isotropic hyperfine coupling is called a “contact” in-
by 2nI + 1. teraction. Only s orbitals have finite electron density at
The equivalence of all magnetic nuclei greatly simpli- a nucleus; p, d, and f orbitals have nodes at the nucleus.
fies the spectrum. For several sets of equivalent nuclei, Thus, contact interaction depends on the s-electron charac-
interpretation is simple only if the difference in coupling ter of the unpaired electron, and the hyperfine constant for
constants is large enough to separate the groups. If a radical a given nucleus provides a measure of the contributions
has two groups of three equivalent protons, the spectrum of the s orbitals on the corresponding atom to the total
will be split into four lines by the first group, and each many-electron wave function of the atom or molecule.
of the four lines will be split again into a quartet by the For molecules, Eq. (8) can be rewritten as
second group. If the two coupling constants differ enough,
h A = 83 πgβg N β N ρ̄ N , (9)
the spectrum will appear as four separated quartets. More
often the groups (quartets in our example) will overlap. If where ρ̄ N is the unpaired spin density, usually just called
all the lines are still separated, analysis is tedious but not spin density, at nucleus N with units of reciprocal vol-
too complex. Real difficulty arises when the linewidth of ume. The spin density ρ̄ N is evaluated from the total
the hyperfine lines interferes with their resolution. Unfor- electron wave function and represents the difference be-
tunately, this often occurs when the number of interacting tween the average number of electrons at the nucleus with
protons becomes large. α spin (m s = + 12 ) and the average number with β spin
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
(m s = − 12 ). By convention, the unpaired electron of a free The spin polarization mechanism generates observable
radical is taken to be an α spin; hence spin densities are hyperfine coupling to protons on the carbon containing the
usually positive. However, a particular nucleus may be in p orbital; such protons are called α protons. The two elec-
a region of excess β spin, in which case the spin density trons in the C—Hα sigma bond are spin polarized such that
at that nucleus is negative. the electron nearest the carbon has the same spin as that of
For example, in the benzyl radical the spin density at the unpaired electron, namely, an α spin. This occurs be-
the meta positions is negative while the spin density at cause the exchange interaction between two parallel spins
the other ring positions is positive. The sign of the spin near the carbon nucleus slightly lowers the energy. This
density corresponds to the sign of the hyperfine coupling spin polarization causes the spin orientation of the un-
constant. In the typical ESR spectrum, no sign information paired electron in the p orbital on carbon to be opposite
on the coupling constants is obtained. Nuclear magnetic the spin orientation of the bonding electron largely in the
resonance (NMR) measurements or, under certain condi- hydrogen 1s orbital. Thus the spin density at the proton is
tions, second-order effects in the ESR spectrum are used negative and the hyperfine coupling constant is negative.
to determine signs. Quantitative calculations for one electron in the 2pz
❍
By convention, the spin density ρ̄ N , which has units of carbon orbital in a ✟ Ċ—H fragment show that a negative
reciprocal volume, is usually normalized by division by spin density of −0.05 is induced at each α proton. This
|ψ N (0)|2 to obtain a dimensionless fractional spin density corresponds to −0.05 (507 G) ≈ −25 G, where 507 G is
ρ N , also usually just called spin density. The number ρ N the value for unit spin density on a proton, which compares
represents the fraction of unpaired spin on an atom N . well with −23.0 G observed for the methyl radical and
A proton hyperfine coupling constant of 142 MHz cor- −22.4 G observed for the ethyl radical. The negative sign
responds to a spin density (ρ N ) of 142/1420 = 0.1 at the is confirmed by NMR measurements.
proton where 1420 MHz is |ψ N (0)| In the ethyl radical the β protons on the carbon adja-
. The spin density ρ N1
2
may be positive or negative, but N ρ N = 1 for all spin- 2 cent to the one with the unpaired electron also produce a
radicals. hyperfine coupling that is the same order of magnitude as
Since the isotropic hyperfine coupling constant is di- that of the α-proton coupling. A spin polarization mech-
rectly proportional to the s-electron spin density, it can be anism would have to pass through two bonds to reach the
used to determine orbital hybridization and consequently β protons and would be expected to be weaker than for
radical structures. To apply this, one must know what the α protons. Therefore, an alternative mechanism involving
hyperfine coupling constant is for a 100% s electron on an hyperconjugations seems probable. This mechanism can
atom. Only for H atoms is this known exactly. For other be pictured qualitatively as follows. The unpaired electron
atoms, the best available Hartree-Fock wave functions are is envisioned as occupying a molecular orbital consist-
used to calculate |ψns (0)|2 . Values have been tabulated that ing of contributions from the two pz carbon orbitals. This
are good to ±10% or much better for the lighter elements. molecular orbital will overlap H-atom 1s orbitals that are
As an example, consider the trifluoromethyl radical not in the nodal plane of the pz orbitals and therefore will
CF3 . To determine the s-electron spin density on the car- overlap with the β protons in the ethyl radical. The hy-
bon, one must measure the 13 C hyperfine coupling con- perconjugation mechanism allows some of the unpaired
stant experimentally. This is found to be 271.6 G, which is α-spin density to directly overlap into the β-proton 1s or-
∼24% of a full 2s electron on the carbon. This implies near bitals and thus predicts a positive coupling constant. A
SP3 bonding in the radical and indicates that CF3 is tetrahe- positive sign is observed by NMR. It should be noted that
dral and not planar. In contrast, the 13 C hyperfine coupling a positive sign at the β protons is also predicted by the
constant for the methyl radical CH3 is 38 G, which indi- spin polarization mechanism, but the magnitude from this
cates only ∼3% s character. This is consistent with a near- mechanism is expected to be smaller.
planar structure for CH3 . In fact, the time-average structure In aromatic radicals the unpaired electron is delocal-
of CH3 is planar, but a small amount of s hybridization can ized over the pz carbon orbitals, so that the spin density
arise from out-of-plane vibrations of the H atoms. in any one pz orbital is less than it is on a methyl radical.
Equation (9) shows a direct proportionality between The protons are in the nodal plane of the pz orbitals and
the hyperfine coupling constant and the s-electron spin exhibit hyperfine coupling through the spin polarization
density. Many radicals have the unpaired electron largely mechanism. The proton splitting is directly proportional
localized in a p orbital, but direct or indirect interaction to the pz or molecular π -orbital spin density on the car-
with orbitals of partial s character can lead to a net spin bon to which the proton is attached, as represented by
density at the nuclei. The ethyl radical ·CH2 CH3 illustrates Eq. (10):
two types of mechanisms that lead to hyperfine coupling
with all its protons. AH = Q H ρCπ , (10)
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
where Q H is a constant with a value of about −25 G. Equa- The difference between the allowed transitions shown
tion (10) applies to protons bonded to any atom, except in Fig. 8 is not changed from the first-order situation at
that Q H is dependent on the atom. It has been most exten- constant magnetic field. However, in an actual ESR spec-
sively tested for protons bonded to carbon and to a lesser trometer the magnetic field is swept, and since the second-
extent for protons bonded to nitrogen. order correction term depends on the magnetic field, the
Although proton hyperfine interactions are by far the “apparent” splitting will change in second order. This ob-
most common in aromatic radicals, a great deal of in- served apparent splitting will be larger than the actual hy-
formation about spin densities can also be derived from perfine splitting, and for the case of hydrogen atoms this
hyperfine interactions with 13 C, 14 N, 19 F, 17 O, and so forth. difference is about 2 G.
The simple relation of Eq. (10) for protons does not gen- The second-order effect on the hydrogen-atom spec-
erally apply to these nuclei because the interactions of the trum is a subtle one, in the sense that the nominal appear-
unpaired spin with spin density on adjacent nuclei and ance of the spectrum does not change. Generally in more
with lone-pair π electrons must be considered. complex paramagnetic systems additional lines in the ESR
spectrum appear that are due to second-order effects in the
energy-level diagram. These types of transitions must be
V. SECOND-ORDER HYPERFINE EFFECTS identified in order to interpret a spectrum well enough to
assign the structure of a radical. The “extra” lines that
When the hyperfine coupling constants are large and the commonly occur due to second-order hyperfine effects
linewidth is small, second-order hyperfine effects must be can be divided into two classes. One class arises from the
considered to explain the observed spectra in many cases. splitting of some of the degeneracies of the inner lines
A simple example is the spectrum of a hydrogen atom. of a hyperfine pattern involving several equivalent nuclei.
The second-order effects in the hydrogen spectrum cause The second class involves forbidden transitions where the
a shift of the center of the spectrum to lower field and nuclear spin selection rule is violated and transitions cor-
cause the observed hyperfine splitting to be larger than responding to
m I = ±1 are observed.
the actual value for a field-swept spectrum. Second-order In general, extra spectral lines due to the splitting of
analysis is required because the hyperfine splitting of 507 the degeneracies of equivalent nuclei occur when the
G is a significant fraction of the typical 3300-G magnetic second-order hyperfine correction magnitude is greater
field used to observe the spectrum. The second-order per- than the linewidth. This has been observed only in liquid-
turbation theory correction factor to the energy levels is phase spectra. To second order, the transition energies for
proportional to the square of the hyperfine coupling con- a system with one unpaired electron and one type of de-
stant divided by the observing field. This causes a second- generate nuclei of nuclear spin I is given by
order shift in two of the four hydrogen-atom energy levels,
1 A2
as shown in Fig. 8. The two transition energies are in- E n = gβH + AM I + I (I + 1) − M I2 , (11)
creased, so the entire spectrum is shifted to lower field due 2 gβH
to this effect. For hydrogen atoms the shift is about 18 G, where M I is the total nuclear spin quantum number for the
which corresponds to a change in the apparent g factor of set of equivalent nuclei. For n equivalent nuclei in general
0.0108. I 2 = M I2 , so from Eq. (11) one can see that some splitting
of the first-order degeneracy will result. An example of this
situation occurs for the trifluoromethyl radical in solution,
where the first-order spectrum would predict four equally
spaced lines with relative intensities of 1 : 3 : 3 : 1. Since
Aiso for 19 F is 145 G, which is a large value, second-
order effects are observable. In the actual spectrum at suffi-
ciently high resolution six lines are observed with relative
intensities 1 : (1 : 2) : (1 : 2) : 1, where the (1 : 2) lines are
two closely spaced lines whose splitting is due to second-
order shifts. All lines are shifted slightly to low field, but
different energy levels are shifted by different amounts,
which leads to the additional splitting of about 9 G in this
case.
The other result of second-order hyperfine effects is to
FIGURE 8 Schematic of the first- and second-order energy lev-
els of a hydrogen atom, the first-order allowed ESR transitions, mix the first-order wave functions as shown in Fig. 8 so
and the first- and second-order spin wave functions. The magni- as to partially allow some forbidden transitions that in-
tude of the second-order shift is A2 [4(g β H + gN β N H )]−1 . volve flipping of nuclear spins. When the first-order wave
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
functions are mixed, the normally forbidden transitions diagonal elements of a diagonalized hyperfine tensor are
become weakly allowed, due to components of the oscil- called the principal values.
lating microwave magnetic field parallel to the direction of To determine the anisotropic hyperfine tensor experi-
the external magnetic field, instead of perpendicular to the mentally, one must usually use a single crystal, although
external magnetic field as is the case for ordinary allowed if there is only one interacting nucleus or possibly more
ESR transitions. In the practical situation there is usually with sufficiently large anisotropy, it is sometimes possible
a small component of the applied microwave magnetic to determine the tensor from powder spectra. For single
field in the direction parallel to the external magnetic field, crystals it is necessary to measure the angularly dependent
which can bring about such nominally forbidden transi- hyperfine splitting in three mutually perpendicular planes
tions. These types of transitions are seldom seen in liquids with respect to the external magnetic field. From this data
but are relatively common in solids. there are well-known procedures to obtain the hyperfine
tensor in the axis system chosen for measurement. Then,
VI. ANISOTROPIC HYPERFINE as outlined above, this tensor may be diagonalized and the
INTERACTION principal values with their associated direction cosines
may be determined.
The general spin Hamiltonian was given by Eq. (2), in The physical interpretation of the anisotropic hyperfine
which the interaction parameters were written in the gen- principal values is given by the classical magnetic dipolar
eral tensor form. The total hyperfine tensor A can be rep- interaction between the electron and nuclear spin angular
resented by a 3 × 3 matrix that connects the three com- momenta. This interaction energy is given by
ponents each of the electron spin angular momentum and 1 − 3(cos2 φ)
the nuclear spin angular momentum. The hyperfine tensor aniso = −gβg N β N I · S, (15)
r3
is a real matrix and can always be diagonalized. Thus a
general tensor referenced to an x–y–z axis system can be where r is the vector between the unpaired electron and
written as
the nucleus with which the interaction occurs and φ is the
Ax x Ax y Ax z angle between r and the electron spin angular momentum
vector S, which is in the direction of the external magnetic
A ≡ A yx A yy A yz . (12)
field. The A principal values are given by
A zx A zy A zz
−1 1 − 3(cos φ)
2
This general tensor can be transformed to another axis
A = −gβg N β N h , (16)
system in which the new tensor is diagonal. This requires r3
av
finding the proper transformation matrix L, which diago-
where av denotes a spatial average over the electronic or-
nalizes the A tensor as
bital of the unpaired electron. The three components of A
L · Ax yz · LT = dAαβγ , (13) are given by three different values of cos φ corresponding
where LT is the transpose of L. The diagonalized tensor d A to rotation in three mutually perpendicular planes of the
is diagonal in the α–β–γ axis system, and the components principal axis system.
of L are the three sets of direction cosines that relate the The dipolar function 1 − 3(cos2 φ)/r 3 )av can be evalu-
x–y–z axis system to the α–β–γ axis system. ated from known wave functions of electrons in s, p, d, and
The components of the diagonalized hyperfine tensor other orbitals on different atoms. For s orbitals the dipolar
consist of an isotropic part A0 and a purely anisotropic function is zero because of spherical symmetry. The cylin-
part A , whose orientational average is zero. This decom- drical symmetry of p orbitals gives three components, A ,
position is shown in Eq. (14): A⊥ , and A⊥ , which are related by A = −2A⊥ . Note that
the dipolar angular function changes sign at φ = 54◦ 44 .
Aαα 0 0 A0 0 0
Thus, the space around a nucleus can be divided into four
d
A= 0 Aββ 0 = 0 A0 0 regions alternating in sign. In the determination of the hy-
0 0 A 0 γγ 0 A0 perfine tensor, a set of signs of the components will be
(14) obtained so that the sum of the diagonal principal values
A 0 0 is zero.
αα
The A components from Eq. (16) have been evaluated
= 0 Aββ 0 .
theoretically for unit spin density in atomic radial wave
0 0 Aγ γ
functions for p orbitals. Thus the experimental anisotropic
Thus the sum of the diagonal elements of d A gives 3A0 , hyperfine components can be used to estimate the amount
and the sum of the diagonal elements of A is zero. The of spin density in p orbitals. This complements the use
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
of isotropic hyperfine values to obtain the amount of spin lated and compared with experimental data. The difficulty
density in s orbitals. is that information about the excited-state energy levels
needs to be known to properly calculate the g tensor and
this is generally known only for simple molecular systems.
VII. g ANISOTROPY In a few cases, such as for the CO− 2 radical ion, detailed
calculations have been carried out and the experimental
In the general spin Hamiltonian given by Eq. (2), the g fac- g anisotropy has given information about the molecular
tor given in the electron Zeeman energy term is written as wave function.
a tensor connecting the electron spin angular momentum Organic radicals generally have the unpaired electron
operator S and the magnetic field vector H. A free elec- in a p orbital, which has orbital angular momentum. How-
tron has only spin angular momentum, and its orientation ever, the weak molecular electrostatic field splits the M L
in a magnetic field is determined only by this physical components and gives M L = 0 as the lowest state. In this
property. However, in general, in atomic and molecular case g ge . Nevertheless, small deviations from ge do
systems there will be some contribution from orbital an- occur and can be readily measured, particularly in single
gular momentum to the total unpaired electron wave func- crystals, since at a typical 3300-G field, g = 0.0006 for a
tion. In this case the orbital and spin angular momentum 1-G shift.
vectors interact, and by convention this interaction is in- When an organic radical contains an atom with a large
corporated into an “effective” anisotropy in the g factor. In spin–orbit coupling constant, such as oxygen, sulfur, or
this representation the spin angular momentum vector S halogens, the g anisotropy becomes significantly greater
no longer represents “the true spin” because the true spin and the average g value generally shifts to larger values.
has only spin angular momentum and is associated with This serves as a diagnostic tool for radicals in solution and
an isotropic g value. Instead, when g is written as a tensor in solids. One particular example involves peroxy radicals
the spin angular momentum vector represents an effective in which the unpaired spin is localized largely on the oxy-
spin, which instead of being oriented along the magnetic gen and for which the average g equals 2.015 in a wide
field direction is oriented along the vector H · g. For most variety of environments. Sulfur-containing radicals also
purposes this nuance will not affect our utilization of the often show large g anisotropy. This can be used as a diag-
g tensor formulation. nostic test for the localization of the radical site in some
The experimental determination of the g tensor is car- biological molecules that contain sulfur.
ried out by a procedure completely analogous to that for The largest g anisotropy occurs for transition-metal
determination of the anisotropic hyperfine tensor. Mea- ions, where the g anisotropy is very useful for discrim-
surements are required as a function of angle in three inating between transition-metal ions in different types of
mutually perpendicular planes. From this data, a general environments. The range of g anisotropy can be rather
g tensor is obtained, which is diagonalized to find the prin- large. Typical values for axial g anisotropy range from
cipal values. The principal axes of the g tensor are often the g⊥ = 2.04 and g = 2.17 for copper complexes to g⊥ = 6
same as for the hyperfine tensor, but they do not have to be. and g = 2 for some ferric complexes.
The interpretation of the principal value of the g tensor
can be conveniently discussed by Eq. (17):
VIII. SPIN RELAXATION
Cλ
gobs = ge + . (17)
E The energy between the magnetic energy levels at 3000 G,
In this expression, ge is the g factor for an isolated spin gβH , is only 10−3 of kT at 300 K. At thermal equilibrium
(2.0023), λ is the spin–orbit coupling constant, C is a pro- the Boltzmann factor, exp(−gβH/kt), gives the popula-
portionality constant calculated from the electronic wave tion ratio of the two levels, so the levels are almost equally
functions, and
E is the energy difference between the populated. The application of microwave energy causes
ground state and the first excited state. Values of λ have transitions between the magnetic levels. The microwave
been obtained for a number of atoms and ions from atomic field stimulates transitions in both directions with a prob-
spectra, but the particular value to be used in a molecular ability that depends on the microwave power and on the
system can only be approximated by this. In general, λ number of spins in each level. Transitions from the lower
values increase with atomic number. The values of
E to upper levels absorb energy, while upper- to lower-level
can sometimes be deduced from electronic spectra. Thus transitions emit energy. Since the population is slightly
the g anisotropy is related to the electronic wave function, greater in the lower level there will be a net absorption of
and if sufficient information is known about the electronic microwave energy; this provides the observed ESR signal.
wave function the principal g components can be calcu- Under steady application of the microwave field with no
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
other interactions, the populations in the magnetic energy spin states are equalized, Mz = 0, and the resonance ab-
levels would soon become equal; there would then be no sorption disappears. After the pulse, the recovery of Mz
net absorption of microwave energy and no ESR signal. toward M0 with a time constant T1 can be observed by the
However, the spin system is subject to other interac- growth of the resonance line. The term T1 is also called the
tions, the very interactions that bring about thermal equi- longitudinal relaxation time, because it refers to relaxation
librium. These interactions can be collectively called spin– along the magnetic-field axis.
lattice interactions. They comprise radiationless interac- The Mx and M y components of M are not changed by
tions between the spin system and the thermal motion of a spin flip. The m x and m y components of each individual
the “lattice” or surroundings. The inverse of the rate of spin are randomly oriented before and after the magnetic
spin–lattice induced transitions is described by a charac- field Hz is applied. However, application of H1 in the x–y
teristic time called the spin–lattice relaxation time and is plane can produce a net phase alignment of the m x and m y
denoted by the symbol T1 . components to give Mx and M y . When H1 is removed, the
At sufficiently low microwave powers, the spin–lattice phase coherence of the spins decays by 63% in time T2 .
relaxation processes are fast enough to maintain a thermal The term T2 is also called the transverse relaxation time
equilibrium population between magnetic energy levels. because it refers to relaxation of magnetization compo-
As the microwave power is increased the net upward rate nents transverse to the external magnetic field.
of microwave-induced spin transitions from the lower to An ESR line is not infinitely sharp; it has a shape and
upper states is increased and eventually competes with width due to spin relaxation. The equations of motion
the spin–lattice induced net downward rate. The spin pop- for Mx , M y , and Mz in the presence of an applied field
ulations in the two magnetic states become more equal H0 and including the spin relaxation processes discussed
and the ESR signal intensity decreases; this is known as above are called the Bloch equations. The solution to these
power saturation. Normally, one wants to use low enough equations predicts a Lorentzian line with a halfwidth at
microwave power to avoid power saturation. halfheight of T2−1 . Lorentzian lineshapes are indeed of-
In addition to spin–lattice relaxation, in which energy ten found for free radicals in liquids. In this case T2 can
is transferred from the spin system to the lattice, there be determined from the linewidth. The Bloch equations
exist spin–spin relaxation mechanisms, in which energy is also predict how the ESR signal intensity will vary with
redistributed within the spin system. One may think of this increasing microwave power. The ESR signal increases,
redistribution as a modulation of the spin energy levels. In reaches a maximum, and then decreases with increasing
both fluid and solid phases, the net local magnetic fields are microwave power; this behavior is called power satura-
rapidly varying due to different types of molecular motion, tion. From an analysis of the power saturation curve of
and a given spin level at m S gβ H is therefore modulated. ESR intensity versus microwave power, it is possible to
At high spin concentrations, direct spin–spin exchange determine T1 .
and dipolar interaction can also occur. The characteristic In solids, typical ESR lineshapes are Gaussian in-
time for spin–spin relaxation within a single spin system stead of Lorentzian. One common interpretation of the
is symbolized by T2 . Gaussian lineshape is that it is composed of a distribution
In a single spin system the spin–lattice (T1 ) and spin– of Lorentzian lineshapes, each of which corresponds to a
spin (T2 ) relaxation times can be given a precise classi- group of spins forming a “spin packet” which “see” the
cal and quantum-mechanical description. A collection of same local magnetic environment. If these spin packets are
spins has a magnetic moment vector M, which can be randomly distributed in intensity they will superimpose to
resolved into three components, Mx , M y , and Mz . Be- give a Gaussian lineshape. Note that for Gaussian lines T2
fore a magnetic field is applied, the number of spins in cannot be determined from the linewidth. Gaussian lines
the two magnetic energy states is equal; after the field is still undergo microwave power saturation, but very care-
applied, some of the spins begin flipping to achieve a ther- ful and sometimes complex analysis is required to extract
mal equilibrium distribution between the two states. For values of T1 and T2 .
an applied magnetic field in the z-direction the spin flips A more direct method to obtain values of the spin–lattice
cause Mz to change toward a steady value M0 , which is and spin–spin relaxation times is to use time-domain ESR
proportional to the measured static magnetic susceptibil- methods, which are briefly described next.
ity. Mz approaches M0 with a time constant T1 such that
Mz = e−1 M0 = 63%M0 in time T1 . So that resonance can IX. DOUBLE-RESONANCE
be observed, the microwave magnetic field H1 is applied AND TIME-DOMAIN ESR
perpendicular to Hz . If the intensity of H1 is increased
greatly with a pulse of microwaves, the spin system satu- Double-resonance experiments are usually carried out in
rates. This means the populations in the upper and lower spectroscopy to increase spectral resolution. There are
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
several different kinds of double-resonance experiments the total number of lines is additive for different types
that can be done in ESR. The technique that appears to of coupled nuclei, whereas in ESR the number of lines
be the most useful currently and one that is becoming is multiplicative. A specific example can be considered
more widespread is electron nuclear double resonance, for the triphenylmethyl radical. In the ESR spectrum the
abbreviated ENDOR. Essentially, ENDOR is the obser- total number of lines is 196, spread over a 30-G range.
vation of the effect of applying a second frequency, which By ENDOR there are only 6 lines, which can be easily
induces nuclear spin flips, simultaneously with the mi- resolved because you see only one pair of ENDOR lines
crowave frequency that induces electron spin flips. The for each group of equivalent nuclei. In the triphenylmethyl
observation of ENDOR depends on partial saturation of radical there are three different types of protons—namely,
an ESR transition. In simple terms, the ENDOR effect is meta, ortho, and para protons—each of which have differ-
simply the change of the ESR intensity when a second ent hyperfine couplings. Thus this spectral simplification
radio frequency field is applied to the system. leads to effectively increased resolution for liquid-phase
Experimentally, ENDOR is carried out by fixing the spectra.
external magnetic field at an ESR hyperfine line and then 4. ENDOR is also very useful for studying details of
sweeping the radio frequency from perhaps 1 to 20 MHz. spin relaxation mechanisms. The effect of molecular mo-
ENDOR lines will be observed corresponding to differ- tions on relaxation processes can be studied. Solvation
ently coupled nuclei in the paramagnetic system. For and temperature effects on both electron and nuclear re-
weak hyperfine interactions, the ENDOR lines for each laxation mechanisms can also be investigated.
type of spin- 12 nucleus are given by ν N ± A/2 where
ν N = g N β N H/ h. For protons, νH ≈ 14 MHz at 3300 G. Time-domain ESR methods have become quite impor-
Thus a pair of lines is observed in the ENDOR spectrum tant for modern applications of ESR. In a time-domain
for each type of coupled protons. The hyperfine constant experiment one uses microwave pulses rather than steady-
A can be measured more accurately with ENDOR than state microwaves. It is possible to use pulses as short as
by ESR, because one is measuring megahertz instead of a few nanoseconds, which makes fast kinetic processes
gigahertz. Also, the hyperfine nucleus can be identified involving paramagnetic species observable and allows the
from the value of g N , since these values are characteristic direct measurement of relaxation times. In addition, time-
for different nuclei. domain ESR methods have been useful for more direct
The practical advantages of ENDOR experiments can determination and study of relaxation mechanisms and
be summarized as follows: for developing new methods for obtaining structural in-
formation in disordered systems.
1. Increased resolution can be obtained, which is par- Two methods of time-domain ESR will be briefly dis-
ticularly important for inhomogeneously broadened lines cussed. The oldest method is that of saturation recovery,
typically observed in solids where the hyperfine struc- which is a direct method to determine spin–lattice relax-
ture is unresolved. A typical ESR linewidth in solids ation times. The idea is to perturb the steady-state pop-
is 10 G, whereas a typical ENDOR linewidth is about ulation of spins with a partially saturating pulse of mi-
0.03 G. A classic example is that of a trapped electron crowaves and then to observe with a very weak microwave
or of trapped hydrogen atoms in potassium chloride crys- field the recovery of the perturbed spin population to equi-
tals. In these paramagnetic systems, ENDOR couplings librium. In the absence of complications, the recovery pro-
to different shells of magnetic nuclei are resolved, and a cess is exponential and can be related to the time constant
detailed picture of the electron and hydrogen-atom wave for spin–lattice relaxation. Exponential recoveries are gen-
functions can be obtained. By ordinary ESR, only a broad erally observed in liquids and in some cases in solids. If
line is observed, in which the hyperfine information for the spin–lattice relaxation time is not much longer than
the different magnetic nuclei in the solid is not resolved. the spin–spin relaxation time, which is atypical in param-
2. Hyperfine constants can be measured more accu- agnetic systems, the interpretation of saturation recovery
rately by ENDOR than by ESR. This is particularly impor- data becomes more complex.
tant for measuring very small hyperfine couplings. Also, The second type of time-domain method that will be
very small changes in the hyperfine couplings due to tem- discussed is electron spin-echo spectroscopy, abbreviated
perature and so forth are most accurately measured by ESE. Figure 9 gives an illustration of a two-pulse electron
ENDOR. spin-echo response. Two resonant microwave pulses are
3. ENDOR can lead to significant spectral simplifica- applied to the system with typical pulse lengths of 10–
tion, which is particularly important in looking at radicals 100 nsec and pulse powers corresponding to several hun-
in liquids. This is because the total number of spectral dred watts. The first pulse (1) essentially starts a clock and
lines is much less in ENDOR than in ESR. In ENDOR flips the spins from the external magnetic field direction
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
Steady-state electron spin resonance can be used to study so appropriate time-domain or trapping techniques must
kinetics over periods of milliseconds to hours. Thermo- be used to detect them before they decay into nonparam-
dynamic information can also be obtained by measuring agnetic species. In such studies, electron spin resonance
reaction intermediates as a function of temperature. plays a major role.
Another major area of application involves magnetic
energy transfer. By measuring magnetic relaxation times
of paramagnetic species, one can determine the degree of XI. NEW DEVELOPMENTS
interaction of paramagnetic species or free radicals with
the “lattice” in which they are embedded. This lattice may In the past few years electron spin resonance has seen
be a liquid or a solid. The mechanism of the magnetic impressive developments in the area of time-domain tech-
energy transfer is an important question and has been ad- niques. Improved solid-state microwave components, fast
dressed in many studies. Applications of this type relate digitizers, and computer systems have made possible new
to the diffusion of paramagnetic species in liquids and in types of pulsed electron spin resonance experiments. The
solids and to the coupling of the spin system to the elec- sensitivity of these new experiments also makes them
tronic system of the lattice. practical for many new chemical, physical, and biological
One simple application of magnetic energy transfer that problems. A second-generation general-purpose pulsed
is often neglected is to use microwave power saturation electron spin resonance spectrometer has been commer-
to distinguish overlapping radicals. Radicals of different cially introduced which makes these pulsed techniques,
chemical types often have quite different spin–lattice re- including pulsed ENDOR, widely available. Fourier trans-
laxation times. For example, alkyl radicals are typically form and two-dimensional electron spin-echo techniques
much more easily saturated than peroxy radicals. Thus, if are also included. A newer commercial development is
both alkyl and peroxy radical spectra are superimposed, pulsed ESR and pulsed ENDOR at the higher frequency
one can detect the presence of at least two different types of 95 GHz. This has particular advantages of better sen-
of radicals by carrying out selective power-saturation mea- sitivity for small samples and enhancement of ENDOR
surements. Similar distinctions can be made between rad- sensitivity for low gyromagnetic ratio nuclei.
ical cations and radical anions, where one may have a In continuous-wave electron spin resonance, extended
much shorter spin–lattice relaxation time than the other. multifrequency capabilities from 0.3 to over 100 GHz
An example involves the photoionization of chlorophyll in have been developed based on loop-gap and other types of
vesicle systems with an electron scavenger such as a halo- resonators. The lower frequencies seem particularly use-
genated quinone present. Typically one detects the chloro- ful for some biological applications. Very high frequency
phyll cation radical signal superimposed on the electron spectrometers have also been developed up to 700 GHz,
acceptor anion radical signal, but they can be distinguished with commercial instrumentation available at 95 GHz. The
by their different responses to microwave power satura- higher frequencies are based on Fabry-Perot resonators
tion. The radical anion involving a halogen atom will typ- and give superior g-anisotropy resolution, suppression
ically saturate with much more difficulty than will the of second-order effects, and better sensitivity for small
chlorophyll cation radical, which is more characteristic of samples.
an organic free radical.
Magnetic relaxation measurements can also be used to
determine the spatial distribution of paramagnetic species
SEE ALSO THE FOLLOWING ARTICLES
in a solid matrix. This is important if the spatial distribu-
tion is nonuniform. Often when radicals or paramagnetic
ATOMIC PHYSICS • ATOMIC SPECTROMETRY • CHEMI-
species are produced by photolysis or by radiolysis they
CAL KINETICS, EXPERIMENTATION • NUCLEAR MAG-
are trapped in a nonuniform manner, and this can be de-
NETIC RESONANCE • PERTURBATION THEORY • QUAN-
tected by a careful analysis of the magnetic relaxation
TUM MECHANICS
characteristics of the radical.
Potential and demonstrated applications of electron spin
resonance are ubiquitous. The technique is particularly
useful because it is sensitive only to those species that are BIBLIOGRAPHY
paramagnetic. If these are important reaction intermedi-
ates, one has a selective analytical technique to look only Atherton, N. M. (1993). “Principles of Electron Spin Resonance,” Ellis
Horwood, London.
at those specific types of reaction intermediates. Paramag- Berliner, L. J., and Rueben, J., eds. (1989). “Biological Magnetic Reso-
netic species are probably much more widespread than is nance. Spin Labeling: Theory and Applications,” Vol. 8, Plenum, New
generally believed. Radicals are typically reactive species, York.
P1: GLQ/LSK P2: FQP Final Pages
Encyclopedia of Physical Science and Technology EN005E-212 June 15, 2001 20:32
Dikanov, S., and Tsvetkov, Y. D. (1992). “Electron Spin Echo Envelope Kevan, L., and Schwartz, R. N., eds. (1979). “Time Domain Electron
Modulation Spectroscopy,” CRC Press, Boca Raton, Fla. Spin Resonance,” Wiley (Interscience), New York.
Hoff, A. J., ed. (1989). “Advanced EPR: Applications in Biology and Kurreck, H., Kirste, B., and Lubitz, W. (1988). “Electron Nuclear Double
Biochemistry,” Elsevier, Amsterdam. Resonance Spectroscopy of Radicals in Solution,” VCH, New York.
Keijzers, C. P., Reijerse, E. J., and Schmidt, J. (1989). “Pulsed EPR,” Pilbrow, J. R. (1990). “Transition Ion Electron Paramagnetic Magnetic
North-Holland, Amsterdam. Resonance,” Oxford Univ. Press, London.
Kevan, L., and Bowman, M. K., eds. (1990). “Modern Pulsed and Slichter, C. P. (1989). “Principles of Magnetic Resonance,” 3rd ed.,
Continuous-Wave Electron Spin Resonance,” Wiley, New York. Springer-Verlag, New York.
Kevan, L., and Kispert, L. (1976). “Electron Spin Double Resonance Weil, J. A., Bolton, J. R., and Wertz, J. E. (1994). “Electron Paramagnetic
Spectroscopy,” Wiley (Interscience), New York. Resonance,” Wiley, New York.
P1: GLM/GJK P2: GRB Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
Electrophoresis
S. P. Spragg
Birmingham University, England
I. Underlying Theory
II. Free-Solution Electrophoresis
III. Electrophoresis in a Stationary Matrix
IV. Electrophoresis in Oscillating Electric Fields
363
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
364 Electrophoresis
Solvent, solute A solvent is chemically identified as the solvating the free ion and the energies of interaction of the
substance present in considerable excess in a mixture, intact molecule with the solvent. It is clear in chemical
while the solute is that present in minor quantities. thermodynamics that these differences can be discussed
through chemical potentials of the two states, but ab initio
calculations of these processes are not possible.
ELECTROPHORESIS is the movement of charged par- The experimental basis for these laws of conduction of
ticles when suspended or dissolved in a polar liquid and solutions involved the measurement of conductivity of a
the mixture is placed in an electric field. This generally solution by applying Ohm’s law to the electrical measure-
excludes the movement of charged particles in gases, al- ments. From these experiments Kohlrausch showed that at
though with modifications the basic concepts could be infinite dilution each ion contributed a definite amount to
applied to these suspensions. Although the procedure is the conductivity irrespective of the nature of the other ion.
more widely known for studying biological molecules and In order to explain these phenomena it became necessary
colloidal particles, the underlying principles were discov- to introduce the concept of electrochemical equivalences
ered during early research on the electrical conduction of showing that the conductance of a solution is the product
solutions of inorganic salts. For most purposes the move- of the number of ion in the solution, the charge carried by
ment of the ion is translational and carried out under the each ion, and the velocity or their mobilities u± .
influence of a constant electric field. Movement in an os- Quantitative relationships were developed between the
cillating electric field could be included within the gen- current i carried through the solution by C+ gram moles of
eral term “electrophoresis”; at high frequencies, however, a univalent anion and the mobilities observed in an electric
the molecular motions of dipoles must be included, and field E:
here quantum treatments are more relevant, so these are
i+ = u+ C+ E. (1)
not considered to be electrophoretic phenomena. For this
reason little is mentioned in this article on the effects of The equivalent expression for the cation is the same as Eq.
oscillating fields on charged ions. It is also implicit in (1), and so the total current i is
the considerations of electrophoresis that the particles or
macroions are insulators and that the internal arrange- i = i+ + i− = (u+ C+ + u− C− )E, (2)
ments of ions in molecules are not affected by external which can be compared directly with Ohm’s law as used
fields. to define resistance,
E = 1/k, (3)
I. UNDERLYING THEORY where k is the conductance, hence, k = u+ C+ + u− C−
and the conductance is proportional to the velocities of the
A. Background ions. The quantity of positive electricity carried in one di-
rection by the cations is proportional to u+ and C+ , while
When a constant electric field is applied across a column of
that in the other direction by the anions is proportional
water, current flows and gases evolve at the electrodes—
to u− and C− . This makes it possible to define a coeffi-
hydrogen at the cathode and oxygen at the anode. This
cient that proportionates the current passed between the
phenomenon is known as electrolysis and was discovered
individual ions. This is called a transference or transport
by M. Faraday in 1832. He provided quantitative laws
number:
to relate the amount of chemical decomposition with the
number of coulombs passed. It is in recognition of his C+ u+
T+ =
work that the primary electrochemical constant is called C+ u+ + C− u−
the faraday. It was not until much later that the source of and
the decomposition was discovered, since the understand-
C− u−
ing of this phenomenon required the knowledge that salts T+ = . (4)
can dissociate into ions when dissolved in a polar sol- C+ u+ + C− u−
vent, a concept first appreciated by S. Arrhenius in 1887. Here, T± is dimensionless, but C± u± has dimensions
The dissociation of salts into ions is now universally ac- mol · A · sec2 · liters−3 , making this equivalent to the num-
cepted, and the fact that dissolving a salt in water causes ber of molar coulombs required to sweep out a unit volume
the spontaneous dissociation of a molecule into one or per second or number of faradays required to sweep out a
more cations (positively charged) and anions (negatively unit volume per second.
charged) is never questioned. The driving force for this In general, the parameter measured in electrophoresis
dissociation arises from the differences in energy between is the velocity of the ion, which is done by identifying
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
Electrophoresis 365
θ+ aa
∴ T+ = =
u+ + u− aa + bb
and
bb
T− = . (5)
aa
+ bb
366 Electrophoresis
circuit plus the time required for the randomly moving function; see Section I.A). If the conditions are such that
molecules to adopt a directional motion (this takes al- these factors do not distort the boundary and the starting
together <10−6 sec). This delay is not considered here condition was similar to that described for Fig. 1, where
because the electrophoretic mobility is slow relative to the concentration of the ion is uniform on one side of the
this transition. For most purposes it can be assumed that boundary and zero on the other, then the boundary shape
the current is constant throughout the experiment and the can be described by Eq. (8):
macroions move at a uniform velocity (d x/dt) determined (π/2)(Dt)1/2
by the field strength and total charge Q on the ion. In or- 1 2
C(x) = C0 1 − √
der to maintain a constant velocity, Newton’s laws show 2 π 0
that the movement in one direction must be opposed by an
equal but opposite force. This opposing force is ascribed x2
× exp − √ dx . (8)
to frictional forces f between the macroions and the sta- 2 Dt
tionary solvent. (In fact, the solvent is stationary only at
distances far removed from the surfaces of the particle.) In Eq. (8), C(x) is the concentration at x after t seconds of
The value of the frictional force is determined by the in- a substance having a diffusion coefficient D (the starting
trinsic size and shape of the ion, as well as the viscosity concentration was C0 ). If, however, a zone of width h con-
of the solvent, and increases with the velocity of the ion. taining the macroion is introduced between the electrode
Thus, we can equate the two opposing forces to give solution and the main bulk of the solvent, both sides of the
zone diffuse to produce a double sigmoid shape, which at
dx dx QE
QE = f or = . (6) its simplest can be described by Eq. (9):
dt dt f
If E is measured as volts per meter, then d x/dt is called C0 h−x h+x
C(x) = erf √ + erf √ , (9)
the mobility of the ion for the chosen experimental con- 2 2 Dt 2 Dt
ditions. The coefficient f has a theoretical foundation in √ a
where [erf (z) = (2/ π ) 0 exp (−a) da] and the distri-
hydrodynamics, and a functional relationship between f
bution is symmetrical about a plane at x = 0. With both
and the coordinates of the particle can be derived for a
types of boundary, the centroid [x̄, Eq. (7)] corresponds to
few regular shapes (for a sphere it is known as the Stokes’
the center of the boundary, but if the boundary is warped by
equation, but there are mathematical solutions for ellip-
electrical inhomogeneities or contains a mixture of unre-
soids and cylinders). It is also known that f is inversely
solved ions of slightly differing mobilities, then x̄ will still
proportional to the randomizing effects of diffusion of a
correspond to the required centroid of the boundary but
large number of ions (Einstein–Sutherland relationship).
not necessarily its geometrical center. This explains why
Essentially, Eq. (6) describes the movement of a single
it is important to use the centroid for calculating average
ion under the influence of an electrical field. Rarely, if
mobilities of electrophoresing boundaries.
ever, can one ion be studied experimentally, because at
The diffusion coefficient D and frictional coefficient f
finite concentrations of ions there are >1013 ions per liter
[Eqs. (6), (8), and (9)] of an ion are similar to that found for
(a 10−10 M solution of 0.1 µg liter−1 for an ion of relative
a neutral molecule. However, because the salt dissociates
mass 103 contains 1013 ions per liter). Diffusion of this
when dissolved, but electroneutrality must be maintained
population of ions spreads the boundary about an elec-
throughout the solution despite each ion having different
trophoretically transported point called the centroid, and
diffusion rates, it is necessary to modify Eq. (6) (and the
it is the velocity of this point that is described by Eq. (6)
other equations where a diffusion coefficient is employed
for experimental situations. The centroid or first moment
to replace frictional forces). This is done by replacing the
(x̄) can be evaluated from Eq. (7) using a set of rectangu-
frictional coefficient in Eq. (6) by F(u+ + u− )/2u+ u− (u+
lar coordinates determined experimentally over an elec-
and u− are the mobilities of the anion and cation, respec-
trophoresing boundary,
x2 tively) to give the Nernst equation.
xy dx In any solution of ions there always occurs electroneu-
x
x̄ = 1x2 . (7) trality, so every positive ion has a counter negative ion.
y 2d x For macroions these are called gegenions. They can be
x1 small ions such as Na+ or Cl− or larger organic ions. Each
The shape of the boundary approximates a Gaussian pro- ion transports a proportion of the current, and since small
file, but the exact description of the shape depends on the ions have greater mobilities than macroions, their trans-
starting conditions of the experiment, the mobilities of the port numbers dominate the system, which means that they
various ions, and whether sharpening of the boundary oc- carry most of the current on both sides of an interface.
curs as a result of electrical effects (Kohlrausch regulating The necessity of maintaining electroneutrality throughout
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
Electrophoresis 367
whether they are neutral or charged. The effect of these J = L 0E + L i (∂ui /∂ x), (11)
i
instabilities is to produce a rapid radjustment of concen-
trations of ions immediately after a boundary is formed, where E is the electric field strength, (∂ui /∂ x) is the force
making the sum of electrochemical and chemical poten- caused by concentration gradients of any ionic or neu-
tials equal across the boundary. It is to reduce these dis- tral species in the solution [for many purposes this mass
turbances in free-solution electrophoresis with macroions transport is expressed by the diffusion, Eq. (8)], and L 0
that the solvent is dialyzed to equilibrium against the so- and L i are phenomenological coefficients. The latter are
lution before an experiment is begun. defined by molecular parameters and composition of the
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
368 Electrophoresis
solution but do not depend on the magnitude of the forces, molecules caused by dispersion forces. In effect, the par-
while the flow J is directly proportional to the mobility u ticles are held apart in a low potential energy well similar
[Eq. (6)]. No existing theory of electrophoretic mobility is in shape to that found for all intermolecular forces (called
consistent with Eq. (11), since most assume that u or J is the Lennard–Jones potential). Thus, it is reasonable to ask
proportional only to the electric field strength. In any anal- how a particle formed from an apparently neutral molecule
ysis it is assumed that the fluid is stationary at the boundary can become charged in suspension. (An example of such
of the particle, and fluid movement is equal but opposite a suspension is a suspension of carbon particles to give
in direction to particulate movement when viewed at dis- printer’s ink.)
tances far removed from the particle. The result of this The source of the charges is adsorbed ions from the
definition is to consider the mobility as a vector. Using solvent collecting on the exposed surface of the particle.
these mathematical boundary conditions, it was possible The energy for this adsorption comes from the interfacial
to solve equations formulated using Eq. (11) and to include energy found when two phases share a common bound-
the close interaction of gegenions with the charged spher- ary and is called surface excess energy (or for aqueous
ical particle (Debye–Hückel relationship) and viscosity η, solutions with respect to air, surface tension). A thermo-
so giving the approximate relationship dynamic argument can be used to describe the energy of
adsorption of a solute. This includes the difference be-
Q 1 + κri
u= P(κa) (12a) tween the chemical potentials of the dissolved ions in the
6πη R 1 + κa occluded solvent around the particle and those in the bulk
phases. This provides the basis for the Gibbs adsorption
8π N ε2 1/2 1/2
κ= I . (12b) isotherm, which shows that the surface excess concentra-
1000 D̄kT
tion of the solute is proportional to the logarithm of the
In these equations ri is the radius of a typical gegenion, R chemical activity of the solute in the bulk solutions (this
is the radius of the macroion, a = R + ri , D̄ is the dielec- is equal to concentration for dilute solutions). The origin
tric constant of the medium, k is the Boltzman constant, of this energy is the differing interactions of the ions with
N is Avogadro’s number, ε is the charge on an electron, the solvent in the bulk of the two phases and that far re-
and I = 0.5 i Z i2 Ci (the ionic strength of the solution, moved from the interface. These interactions lead to either
where Z is the valency of the ith ion of concentration an accumulation or a deficit at the interface.
Ci ); P(κa) is a dimensionless function required to allow A natural outcome of these phenomena is that the charge
for the effect I has on the effective radius of the parti- measured by electrophoresis of macroions or particles is
cle and has values that vary between 1.0 and 1.5. The not necessarily equal to that found by algebraically sum-
inclusion of a in Eq. (12) is necessary in order to include ming the ionizations of the intrinsic side groupings of
the effect that the close approximation of the neutralizing the macroion. (This can be estimated for acidic and ba-
gegenions has on the particle. The effect is to increase sic groups by chemical titration.) The term “ζ potential”
the radius of the macroion beyond that expected from the has been used to describe this total electrophoretic charge.
neutral molecule. Several workers have changed Eq. (12) It is the charge at the plane of slip between the charged
to allow for other electrical phenomena, but despite these macroion or particle and the bulk solvent. The spatial de-
efforts it is reasonable to assume that no relationship be- pendence of the charges depends on many factors, but an
tween electrophoretic mobility and molecular parameters idealized example of the distribution is shown in Fig. 3
is rigorously applicable to macroions. for a positive ζ potential. It should be mentioned that the
This conclusion is disappointing because the major rea- distribution shown in Fig. 3 is an average because there is
son for the development of electrophoresis was to relate a general randomizing movement of the ions (Brownian
the charge of colloidal suspensions and biological macro- motion). This means that the instantaneous relationship
molecules to their known molecular parameters. The use for individual ions need not equal that shown in Fig. 3,
of electrophoresis in recent times has been restricted to and it is still possible for collisions to occur between the
empirical studies where electrical forces have been used particles despite the fact that on average they all carry
to separate mixtures of charged particles or macroions into the same charge. These collisions eventually precipitate
individual components. the suspensions.
The technique has many applications and has even A ζ potential is more generally used in discussions of
been used for separating suspensions containing appar- colloidal suspensions (e.g., pigments in paints) and is not
ently neutral molecules, because all particles in stable sus- often applied to the macroions found in biology. Here, the
pensions carry a net charge. The charging of a colloidal major part of the charge arises from ionization of the po-
particle is necessary if precipitation is to be reduced. This lar groups of the condensed subunits in the polymer. For
produces a balance between the repulsive forces produced proteins (an example of polyampholytes) there are ioniz-
by like charges and the ubiquitous attraction between all able carboxylic, amino, phenolic, and mercapto groups,
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
Electrophoresis 369
370 Electrophoresis
Electrophoresis 371
which describes charge and shape; (2) sedimentation co- venient to describe the action as though the middle were a
efficient s, which describes mass, density, and shape; and single section. The middle carried the optical windows on
(3) diffusion coefficient D, which describes shape only. two sides of each channel, and it was usual to monitor the
The common property for these three is shape, which in movement of the boundary through these windows by non-
turn is contained within the formulation of the frictional invasive optical methods. The experiment was performed
coefficient f [as in Eq. (6)]. Thus, each mobility can be by first introducing the solvent into the cell while all the
related to the other two, provided that the frictional coeffi- sections were connected. Then one would slide the center
cient is identical for the three conditions. Experience has section plus the top section across the bottom, so isolating
shown that this is a valid assumption, provided that exper- the bottom section. The solvent was removed from one
imental conditions are constant during the determinations of the limbs in the ∪-tube and replaced by the solution
(constant composition of solvent, pH, ionic strength, tem- (dialyzed if the experiment involved macroions). The top
perature, and concentration of macroion). Using the ideal section was then moved back, leaving all the sections iso-
relationships for s, D, and u± [Eqs. (22)–(24)], the rela- lated from the one limb and replaced by solvent before the
tionship for mobility and sedimentation is given in Eq. connection was made to the electrode vessels. The whole
(25): assembly was mounted on a mechanical support, and the
solvent was added in order to fill the electrode vessels and
M(1 − ρs /ρm )D
s= (22) top of the cell. Saturated KCl solution was added to the
RT bottom of each electrode vessel after the electrodes were
D = RT /F (23) inserted. The complete unit was placed in a thermostat-
u± = Q ± /F (24) ted bath having optical windows for examining the center
section from outside the bath. When it was equilibrated
M(1 − ρs /ρm )u±
s= , (25) for temperature, the middle section was moved across to
Q complete the channel through the ∪-tube. This operation
where M is the molar mass, ρs is the solvent density, ρm produced a boundary between solution and solvent at the
is the macroion weight density, and Q is the charge of two interfaces between top, middle, and bottom sections.
macromolecule calculated from the product of the num- During the experiment, the movement of the boundary was
ber of charges and the charge on an electron. Within the observed by a variety of methods of which the most popu-
maximum allowable experimental conditions that can be lar wer schlieren and interference optics. The patterns were
used in sedimentation, this procedure gives velocities that recorded on films, which were eventually measured to cal-
are about 10 times that produced by electrophoretic forces. culate velocities and boundary profiles. Considerable care
This fact explains in part why for many pruposes sedimen- was taken before the experiment began to equalize the col-
tation has proved a more usable preparative technique for umn heights, thus reducing hydrostatic distrubance. The
resolving mixtures than electrophoresis in free solution. electrodes were made from silver wire and coated with
AgCl, so when the electrodes were immersed in the satu-
rated KCl solution at the bottom of the electrode vessels
the major current was transported by Cl− . Electrolysis
II. FREE-SOLUTION ELECTROPHORESIS
did not occur, provided that the solution under investiga-
tion contained K+ and Cl− . The assembly was vertically
A. Conventional Procedures
mounted, and the densities of the solutions had to increase
Until the late 1950s electrophoretic experiments were car- from the top to the bottom in order to minimize mechanical
ried out in columns of aqueous solutions. The equipment mixing of the boundaries.
was, in principle, that given in Fig. 1, but, in practice, it Electrical heating of the solution occurred during the ex-
consisted of a glass ∪-tube having a square cross-section periment, and the resulting density changes could cause
and constructed from three sections (divided across the convection, which would modify the boundaries and in
channel of the ∪). Each section carried parts of both limbs extreme cases destroy them. To reduce this convection, ex-
of the ∪-tube, and they were designed so that each could periments were normally made at the temperature of maxi-
slide across the others in a plane set normal to the direc- mum density of water (∼277◦ C) while the current was kept
tion of the channels in the ∪-tube. The top section had low by employing organic solutes to buffer the pH when
outlets from the ∪-tube in order to connect the limbs to solutions of polyampholytes were examined. Despite all
separate electrode vessels, while the bottom section was these precautions, considerable experience was required
essentially a connector to complete the bottom part of the in setting up the experiments and interpreting the results.
∪-tube. The middle section was the optical component, This equipment (colloquially called the Tiselius appara-
and it could be divided into two sections, but it is more con- tus) was used extensively by biophysical chemists when
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
372 Electrophoresis
TABLE I Examples of Mobilities u of Several Proteins Mea- quency is the same as the incident light, but in solutions
sured in Solutions Having Ionic Strengths of ∼0.1 M the molecules become stationary only when the temper-
Isoelectric u × 10−9 ature is absolute zero. At higher temperatures molecules
Protein point pH (m2 sec−1 V−1 ) show the random motion known as Brownian motion, and
light scattered from these is shifted to higher or lower fre-
Ovalbumin 4.58 6.8 −6.1
quencies depending on the relative direction the molecules
α-Lactoglobulin A 5.09 5.3–6.0 −0.63
were moving at the time of scatter. This is called quasi-
α-Lactoglobulin B 5.23 5.3–6.0 −0.12
elastic light scattering (QEL) in order to contrast it with
α-Casein 4.1 8.6 −6.7
the situation where no change in frequency of the light
Serum albumin 4.7 8.6 −6.7
occurs during scattering, which is called elastic scatter-
ing. The frequency shift is less than 1 MHz for macroions
and becomes smaller as the size of the molecule increases.
studying mixtrues of proteins. Information from these ex- Experimentally, the problem was to determine this small
periments was obtained on the role of charges in stabilizing shift relative to the incident light frequency of about
macroions in solutions. Some selected results of mobili- 1015 Hz.
ties and isoelectric points are given in Table I in order to Measuring small shifts in frequency of light can best
gauge the magnitudes of the mobilities as well as illus- be achieved by interferometric methods in which the scat-
trate the values of typical isoelectric points (where zero tered light is mixed with the incident light at the surface of
mobility would be observed) for these proteins. a detecter (e.g., a photomultiplier cathode). The resulting
It should be pointed out that results from free-solution signal has a frequency that equals the difference between
electrophoresis were tested against models used to gener- the two frequencies. Two basic procedures have been de-
ate Eqs. (6)–(12), and these comparisons showed that no veloped for recording these beat frequencies: heterodyne
general model could be proposed to describe adequately and homodyne detection. The heterodyne method requires
the properties of charged macroions in solution. When either a direct mixing of the incident light with the scat-
this conclusion is combined with the technical problems tered light (after reduction of the incident intensity) on
of interpreting incompletely separated boundaries and the the detecter surface or the positioning of a stationary scat-
difficulties of covering wide-ranging conditions for the ex- terer in the solution (e.g., a captive polystyrene sphere of
periments, it is clear why the use of the Tiselius apparatus considerably larger dimensions than the macroions being
declined during the 1960s and hydrodynamic properties investigated) and recording the resultant signal. In homo-
were studied by ultracentrifugal analyses. The ultracen- dyne experiments the scattered light at time t is autocorre-
trifuge had a built-in stabilizing force that reduced con- lated with that recorded a short time later (t + τ ). Unlike
vection, and it could be used to study neutral molecules conventional light scattering, where the light intensity is
as well as charged molecules in a wide variety of solvents recorded and which requires high incident light levels, the
and temperatures. Possibly because of its unique features, intensity of scattered light for QEL must be sufficiently
electrophoresis did not disappear but evolved into a qual- low that the photons arriving at the detecter can be counted.
itative tool through the use of stabilizing gels. This made When counting is employed it is possible to sample the flux
the technique one of the most widely applied procedures in of photons temporally, and using statistical procedures of
biochemical studies (see Section III). As for small ions, the autocorrelation the counts are processed to produce a re-
understanding of the conductance of ionic solutions has laxation curve with an exponential decay of the first-order
not been seriously pursued in recent times, so the need to correlation function with time. The time constant of the
develop new experimental procedures that use boundaries curve is related to the diffusion constant of the scattering
has not been present. macroions and therefore provides a measure of the diffu-
sion of the ions. These measurements are obtained from
a solution that is at true equilibrium, and no separation
B. Electrophoretic Light Scattering
occurs during the experiment.
A development in light scattering in the 1970s made it pos- Electrophoretic applications of this method depend on
sible to study the electrophoretic movement of macroions the known inverse relationship between the first-order cor-
without forming boundaries. The physical principle be- relation function and the power spectrum. (They are a
hind the technique was to measure molecular motions Fourier transform pair.) With random motion the frequen-
through the Doppler shift in the frequency of scattered cies of the scattered light spread about the incident light,
light relative to the incident beam. The measurements were producing a Lorentzian distribution (the center being at
made using a single-mode laser as the source of light. the frequency of the incident light; the spectrum is known
When light is scattered from a stationary object, its fre- as a Rayleigh line to distinguish it from other spectral
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
Electrophoresis 373
374 Electrophoresis
polymerization by one of a variety of procedures, all of ported with the ion (or remain stationary on the gel), there-
which produce free radicals in solution. These radicals fore making their radii different from that expected from
are produced by hydrolysis of water, either photochem- crystallographic models. In general the penetrating ions
ically using riboflavin phosphate as a catalyst or chem- move through the pores with mobilities commensurate
ically using chemical catalysts. Once the initiation has with those measured in free solution. It is usually diffi-
occurred, the acrylamide crosslinks with itself, with the cult to measure these mobilities because the path length
evolution of heat to produce linear chains that are termi- taken by the ion is unknown, although it is certainly greater
nated when the activated chain ends are neutralized by than the simple linear distance measured macroscopically
collision with other free radicals from the solvent. The along the gel. This means that PAGE cannot yield physico-
product is a stable gel containing only a few percentage chemical data on absolute charges of diffusible ions; only
points (by weight) of polyacrylamide. Gels formed from relative mobilities can be obtained and estimated by ref-
only acrylamide have many undesirable properties (e.g., erence to standards electrophoresed in the same gel.
they are glutinous and stick to glass), and the pore sizes In the case of macroions the relationship between rela-
are illdefined. A more satisfactory gel was formed by in- tive mobilities and charge is complicated by the possibility
cluding a small proportion of the bifunctional acrylamide that selective retardation of the ions occurs as a result of
(Fig. 4b) in the mixture. This forms random cross-links physical impedence by the gel matrix and adsorption onto
between the linear chains during polymerization. The av- the stationary polyacrylamide. Contributions from both
erage size of the pores formed by this mixture is deter- occur with all ions, but for proteins the adsorption is less
mined by the proportion of bisacrylamide (Fig. 4b) to the important than the filtering. At the simplest level this fil-
normal acrylamide in the original solution. The mechan- tering would be described through frictional coefficients
ical stability of the gel decreases with increased propor- [Eq. (6)], which for a fixed pore size increase in proportion
tion of crosslinking, and for some purposes the bisacry- to the cube root of the mass. Thus, the velocity is propor-
lamide is replaced by N,N
-diallyltartardiamide. The tional to M −1/3 , and the relative position of macroions on
latter produces a gel that is more restrictive to macroions a gel at the end of an experiment are a function of charge
than bisacrylamide at high proportions of cross-linker, but and mass (that is, proportional to Q/M 1/3 ), which means
in turn it is more manageable. The solvent is held in these that the charge cannot be measured independently of the
pores, and the whole gel can be handled without mechan- mass. Because of this proportional dependence of relative
ical supports, despite the fact that it consists of up to 95% positions on the two dependent parameters, one must be
(by weight) of solvent. fixed before the other can be estimated using relative mo-
Polyacrylamide gel electrophoresis (PAGE) is carried bilities. Thus, if Q were constant for the macroions in a
out using simple equipment. The gels are cast either in mixture, the effective van der Waals radii could be esti-
glass tubes or as slabs supported on nonconducting plates, mated for each from its relative mobility, and if the shape
which can themselves be thermostatted. Because the gels of the ions were constant (that is, all spheres), this would
are generally thin, they have low electrical conductance, so give relative masses M.
relatively high voltages can be applied without excessive These criteria have been used extensively for measuring
heating (say, 1000 V across a 15-cm tube, 2-mm diameter). relative masses of proteins and nucleic acids by including
Along with the developments of PAGE were the pro- in a separate channel a mixture of standard proteins or
duction of a variety of organic ions that would buffer pH nucleic acids of known relative masses; the masses are es-
without producing solutions of high conductivity. High timated in advance using absolute procedures such as that
conductivity arises because of high mobilities of ions, and involving the analytical ultracentrifuge. It is usually not
this is inversely proportional to the van der Waals radius of possible to measure charge this way, because a range of
an ion [Eq. (6); the frictional coefficient is proportional to macroions are not usually available having a fixed mass
this radius]. Organic ions have larger radii than inorganic but carrying a range of charges. It is possible to generate
ions such as phosphates, and so their conductivity is less. such a standard by progressive carbamylation of amino
groups (easily achieved by cautions warming of a protein
with urea solutions) to produce secondary amines that are
B. Molecular Properties Important for
not charged at neutral pH. If the carbamylation is not al-
Polyacrylamide Gel Electrophoresis
lowed to go to completion for all the side chains, this
When ions are transported through the pores of the gel, will produce a mixture of molecules of relatively constant
their relative mobilities can be used to determine their mass (the carbamyl group has a relative mass of 60, so
sizes, provided that they are not excluded from the pores even adding 10 groups to one macroion will not signifi-
of the gel. The radii of the transporting ions are not simple cantly increase the mass). This standard mixture can be
to define because for both the stationary matrix and the used to determine relative charges of the unknowns; then
diffusible ions there are layers of solvent that are trans- if the original absolute charge and amino acid composition
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
Electrophoresis 375
are known (required to give the number of amines substi- peptides, making it possible to study otherwise insoluble
tuted), it is possible to calculate the number of charges mixtures.
on the unknown protein. Inaccuracies are introduced into When a highly charged macroion is produced, the ag-
these calculations because the amount of adsorption of the gregate moves rapidly in the gel matrix under a moderate
charged macroions to the polyacrylamide is dependent on potential gradient. Their relative velocities are related to
the charge, so the position of the polyampholyte after car- size, but since these macroions still carry some residual
bamylation does not remain a simple function of charge. shape from the native molecule, the frictional coefficient
This brief definition of the molecular parameters that is not always equal to the expected sphere. Hence, it is
affect rates of transport in PAGE illustrates the impor- important to use reference standards whose overall origi-
tant assumptions made in converting relative positions to nal shape is similar to the unknown proteins if reasonable
molecular parameters for the procedures discussed in the estimates are to be obtained. (The absolute accuracy is sel-
following sections. dom better than ±10%, although reproducibility is much
higher.) To reduce the contribution from variable shapes
it is usual to perform several experiments in gels formed
C. Estimation of Relative Masses Using
from various concentrations of acrylamide and bisacry-
Polyacrylamide Gel Electrophoresis
lamide. Plotting the logarithm of the relative mobility
Most of the charge of a biological macroion comes from against the concentration of acrylamide gives a straight
dissociation of the intrinsic chemical groups. In the case line whose slope can be related to the molecular size,
of many of the nonparticulate and soluble proteins, these while the intercept on the ordinate (infinite dilution of
groups have isoelectric points in the pH range 4–5, which acrylamide and bisacrylamide) is a measure of the mobil-
means that at neutral pH they are negatively charged. The ity of the SDS–protein in free solution (Ferguson plot).
absolute charge is not independent of mass, because the These plots can be used to determine relative masses of
capacity to carry more amino acids bearing charged side native proteins, because the slope is a measure of the ef-
chains is greater the larger the mass, while the compo- fective ratio of charge to mass at unit charge. Proteins that
sition is determined by genetic factors. This means the contain a significant amount of covalently linked carbohy-
ratio Q/M 1/3 is not constant for all proteins, so that sep- drate can still give anomalous results in this plot because
arations between individual proteins can be obtained ex- the randomly arranged carbohydrate chains change the
perimentally. In the case of nucleic acids the total charge overall shape of the ellipsoid from that given by standards
is generally related to mass for a given type of nucleic using purer proteins. Another factor to be considered is
acid, because here each nucleoside (the effective monomer the dependence on the amount of detergent bound per
of nucleic acids) carries free phosphates that are equally unit weight of peptide. Although this is generally con-
ionized at neutral pH. Thus, the relative positions after stant, there are notable exceptions where the equilibrium
PAGE can be related to size. As a result, masses deter- between free SDS and that bound does not follow the ex-
mined by a single PAGE experiment with native proteins pected relationship. To overcome this problem, high con-
are less readily interpreted in terms of van der Waals radii centrations of SDS (say, 10% solutions) may be required
than those made with nucleic acids (but see the later dis- in some cases, and this has its own limitations.
cussion of the Ferguson plot). It has been found, how- The bands or spots produced by SDS–PAGE are
ever, that when a protein is mixed with certain charged widened by diffusion of the micelles within the pores of
detergents [the most popular being sodium dodecyl sul- the polyacrylamide, but since the electrophoretic mobil-
fate (SDS)] the quantity of detergent associated with a ities of the bands are unidirectional and are greater than
gram of protein is relatively constant. The result of this those produced by diffusion, the leading edge of the band
association is a spheroidal micelle having a charge and is sharper than that expected from a simple diffusional
frictional coefficient proportional to the relative mass of model [Eq. (9)]. The concentration of the micelles at the
the protein [see Eq. (6)]. The addition of SDS dissociates leading edge is an advantage when small amounts of a
multisubunit proteins into their respective components, so macroion are being studied. Some experimental proce-
although adding SDS produces a macroion whose mass dures enhance this sharpening by enlisting the Kohlrausch
can be estimated from PAGE (the intrinsic charge of the regulating function [T± /C, Eq. (4)]. To produce sharp
protein is swamped by the added charge from the SDS), bands this ratio must be unequal on the two sides of an
the native biologically active units cannot be examined interface, and to achieve this the salt concentrations (and
in the detergent. Despite this deficiency, SDS–PAGE has pH for polyampholytes) must be different across the inter-
become the most popular method of determining relative face. Practically, this is achieved by layering a thin band
masses of protein subunits and has displaced the ultracen- of gel containing different buffers on top of the main gel
trifuge in routine investigations. Another attraction is that and electrophoresing the protein through this band before
the detergent solubilizes otherwise insoluble proteins and entering the main gel. More elaborate arrangements of
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
376 Electrophoresis
different salt concentrations and pH have been used to trophorsing the slab for a period before adding the pro-
produce stacking of native proteins by varying their rela- tein. This pre-electrophoresing settles the ampholines at
tive charges in a plane in the gel. This changes the relative their varied isoelectric points. Because ampholines con-
mobilities, so that T± /C for each protein is not equal to tain many functional groups,1 they possess greater buffer-
that in the next plane (called isotachophoresis). ing capacity than a similar mass of peptides, and so the
Procedures similar to those used for proteins can be gradient is not disturbed by the presence of the protein.
used for nucleic acids. Here, the gels not only act as sta- In these experiments the current is very low after the
ble supports for mechanical handling, but also as separate initial removal of excess charged diffusible ions because
mixtures according to mass. No additional detergents are the only remaining transportable ions in the system are H+
required because the structure of nucleic acids in solution and OH− . The focused boundaries follow the approximate
is essentially a random coil formed from highly charged shape described by Eq. (21) and are stationary within the
polyelectrolytes. gel for long periods. True equilibrium, where the position
For technical reasons gels containing less than 2% (by is independent of very long times, is rarely achieved be-
weight) of acrylamide are unmanageable, and yet even cause the transport of the H+ and OH− continues, and this
at that concentration the pore sizes are too small to admit slowly drags the ampholines with them, eventually de-
large macroions (radii >10 nm). To handle these ions, gels stroying the gradient. Ampholines have been developed
must be formed from polysaccharides. The most common that can be copoymerized into the polyacrylamide gel to
is agarose, a polymer of galactose. Agarose is a fraction prevent their movement, and when these are used the pH
from agar, a seaweed polysaccharide, which is partly sul- gradient is formed mechanically before the acrylamide is
fonated so the charged parts must be removed for use in polymerized.
electrophoresis. These gels are mechanically fragile but With this technique it is possible to focus individual pro-
can be partly stabilized if a few covalent bonds are formed teins from a mixture into bands that are fractions of a mil-
between some of the galactose units. limeter wide and have effective concentrations exceeding
the solubility of the protein. At these high concentrations
the protein “steals” the water from the polyacrylamide
D. Isoelectric Focusing and Isotachophoresis chains, and this weakens the matrix, making the columns
The procedures described for PAGE employ the charges fragile at this point. The technique can be applied to the
only as a means of electrically driving the macroions along study of native or denatured proteins. In the latter case a
the gel in a fixed direction in order to separate mixtures charged detergent (e.g., SDS) is displaced from the protein
and estimate relative masses. These procedures yield lit- during electrophoresis because the unassociated detergent
tle information on the charge of the protein or utilize moves to the anode, therefore forcing the micelles to dis-
the unique pH where polyampholytes have no net charge sociate in order to maintain chemical equilibrium between
(see Section II.D) in order to separate a complex mix- associated and free detergent. Eventually, all the adsorbed
ture. A stationary boundary forms at this pH, but since SDS is stripped away from the peptide, leaving it with its
the object of the experiment is to both separate and con- native intrinsic charge (and possible insolubility).
centrate proteins into narrow bands, a stable pH gradient For preparative procedures where relatively high con-
must be generated. To do this the solution containing the centrations of a mixture of polyampholytes are applied
monomeric acrylamide must contain ampholyte buffers initially to the pH gradient, it is often uneconomic to em-
that when electrophoresed move more rapidly than the ploy supporting ampholines in the gel. In these cases a pH
polyampholytes and settle at their respective isoelectric gradient is generated during preparation of the gel column
points where they buffer the pH. If a wide-ranging mixture using conventional buffers. The object of an experiment
of ampholytes, themselves having many different isoelec- is to isolate one polyampholyte selectively by stacking it
tric points, is used, the result is a stable pH gradient, which at an interface between two zones using the Kohlrausch
can be made approximately linear with distance between regulating function as an underlying theoretical guide (see
the electrodes. Section I.A). In order to apply the technique it is neces-
A pH gradient can be formed if water is electrolyzed— sary to have studied the protein using analytical PAGE in
acid at the anode and alkaline at the cathode—but the order to ascertain its relative charge and mass. The latter is
buffering capacity of water is negligible and the gradient required in order to adjust the properties of the supporting
is easily swamped by the protein when they are included. polyacrylamide gel in isotachophoresis so that it does not
To make a stable gradient a variety of organic ampholines
1 They can be likened to polyacrylic acid, where each monomer ex-
have been synthesized, with various proportions of acidic
poses a charged carboxylic group, but in the ampholines each monomer
and basic groups in the heterogeneous mixture of ampho- exposes a basic and an acidic group, so by producing a mixture of
lines. When these are included in the polyacrylamide gel oligomers from the ampholines a wide range of isoelectric points can
as free solutes, a pH gradient can be generated by elec- be produced.
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
Electrophoresis 377
restrict the movement of the polyampholyte. Then a se- ing complex biological mixtures of macroions. In this evo-
ries of buffer zones must be inserted, and both the type lution, attempts were made to expand to two-dimensional
and the concentrations of these buffers must be controlled separations, but only recently has this procedure been suc-
so that a pH gradient is generated yet the ionic strength in cessful. The method illustrates the resolving power pro-
the stacking regions is low. Then after electrophoresing of duced by combining isoelectric focusing in one direction
the polyampholytes, the selected few that produce equality with separations according to van der Waals radii in the
for the regulating function across the phases produce sharp second. In order to carry out the experiment the protein
boundaries at these interfaces. The selection depends on mixture is first separated according to the isoelectric points
the charge of the polyampholyte (which is dependent on of the components (see Section III.D). At the end of this
pH in the zone) as well as the buffer concentration. Ideally, experiment these gels are placed on top of a slab of poly-
the polyampholyte should be the major transporter of the acrylamide, usually formed as a gradient in concentration
current rather than the buffer ions. Computer programs ex- of acrylamide and increasing in concentration away from
ist that optimize the buffering conditions for these experi- the isoelectrically focused gel. This combination is then
ments. The procedure is often known as steady-state stack- electrophoresed by SDS–PAGE (Section II.C) before the
ing and multiphasic zone electrophoresis. Although these separated peptides in the gel are visualized by selective
procedures have much in common with isotachophore- staining or autoradiogaphy. Combining the two orthogo-
sis, there are experimental variations between them which nal properties of proteins produces the typical result shown
might distinguish them for some applications. in Fig. 5 for a mixture taken from a rat’s liver. The pro-
cedure does not lend itself to inclusion of reference stan-
dards in the gel, while prior denaturation of the proteins
E. Two-Dimensional Polyacrylamide
with SDS yields a valuable map of the gene products, as in
Gel Electrophoresis
Fig. 5. It is helpful in genetic studies if the disulfide cross-
Present applications of electrophoresis are far removed links between the subunits have been destroyed by adding
from the early physical work where the technique provided mercapto compunds to the original preparations. These
the only means of measuring charges on particles. Now maps are reproducible provided that the experimental pro-
electrophoresis has become an empirical tool for separat- cedures are rigorously controlled, making it possible to
FIGURE 5 Two-dimensional separation of a mixture of denatured peptides from a rat’s liver. The separation hor-
izontally is by isoelectric focusing (Section III.D) with the low pH to the left. The vertical separation was made by
SDS–PAGE according to size, the largest at the top. (The upper limit of relative mass is ∼105 and the lower limit is
∼15,000.)
P1: GLM/GJK P2: GRB Final Pages
Encyclopedia of Physical Science and Technology En005H-218 June 15, 2001 20:33
378 Electrophoresis
locate unique peptides affected by biological experiments studied by spectroscopic procedures. At lower frequen-
through direct superposition of many maps on top of one cies procedures have been devised to measure the rota-
another. As many as 2000 peptides have been resolved in tional diffusion of polar molecules; among these may be
one map from a single mixture, and individual variations mentioned dielectrophoresis, which unlike electrophore-
can be noted by suitable computer analyses of the digitized sis is concerned with the motion of both charged and un-
maps. charged particles in a nonuniform electric field. The di-
Two-dimensional maps of nucleic acid fragments are electric losses are measured as the frequency of the field
also made during sequencing procedure in order to read is increased, and when the frequency exceeds that where
gene codes. These maps are different in format from those the rotating dipolar molecules can follow the oscillating
prepared from protein mixtures because the radioactively field there is a marked change in the dielectric properties of
labeled and partially hydrolyzed nucleic acids are sepa- the solution. This frequency can be related to the distance
rated in one direction according to size, while the second between the poles in the dipole, provided that assump-
dimension contains parallel “ladders” formed from repli- tions are made about the shape of the molecule and values
cate experiments in which the nucleic acids are hydrolyzed for the local viscosity of the solvent can be obtained. In
by different enzymes. The result is a series of bands that general these observations have not been applied widely
can be correlated with other bands in order to produce because in solutions containing mixture of small and large
the correct sequence of nucleotides in the original nu- ions difficulties arise in interpreting the uncertain contri-
cleic acid. These gels are usually much longer than those butions from relaxations of gegenions or counterions with
used for protein separations, and the visualization is gen- respect to the macroion. In other words, the intention of
erally carried out after labeling with radioactive isotopes the experiments is frustrated by the lack of reliable inter-
or covalently linked fluorescent dyes. pretation. The work must be supplemented by data from
other sources in order to describe the electrical properties
F. Other Techniques of shielded dipoles.
In general the matrix methods utilize only two properties
of ions in their separation, namely, charge and/or size. It is SEE ALSO THE FOLLOWING ARTICLES
in the visualization of the separated products that the var-
ious methods differ. Some procedures employ immuno- CAPILLARY ZONE ELECTROPHORESIS • CHEMICAL THER-
logical reactions to visualize the products. Either this is MODYNAMICS • ELECTROCHEMISTRY • ELECTROLYTE
done after the separation by transferring the proteins to SOLUTIONS, TRANSPORT PROPERTIES • GAS CHRO-
other media prior to reacting with antibodies, or the anti- MATOGRAPHY • MICELLES
bodies are included in the acrylamide gel so the reactions
occur during the separation. Other methods rely on spe-
cific enzyme reactions that produce colored bands in order BIBLIOGRAPHY
to locate the enzymes; they cannot be performed on de-
natured products. Finally, fluorescent labeling before and Foret, F., Krivanková, L., and Boc, P. (1993). “Capillary Zone Elec-
after electrophoresis is superceding radioactive labeling trophoresis,” Wiley, New York.
for simple detection, which requires the development of Jandik, P., and Bonn, G. (1993). “Capillary Electrophoresis of Small
Molecules and Ions,” VCH Publishers, Weinheim/New York.
new physical detecters for scanning the gels. Khaledi, M. G., ed. (1998). “High-Performance Capillary Electrophore-
sis: Theory, Techniques, and Applications,” Wiley, New York.
Lunn, G. (2000). “Capillary Electrophoresis Methods for Pharmaceutical
IV. ELECTROPHORESIS IN OSCILLATING Analysis,” Wiley, New York.
ELECTRIC FIELDS Rabilloud, T. (1999). “Proteome Research: Two-Dimensional Gel Elec-
trophoresis and Detection Methods,” Principles and Practice Series,
Electrophoresis is generally associated with transport in a Springer-Verlag, Berlin/New York.
Rothe, G. M. (1994). “Electrophoresis of Enzymes Laboratory Methods,”
stationary electric gradient, but the velocity of small ions
Springer Laboratory Series, Springer-Verlag, Berlin/New York.
is sufficient for them to have been trasported during half Tietz, D., ed. (1998). “Nucleic Acid Electrophoresis,” Springer Labora-
a cycle of an alternating field (see Section II.B). As the tory Series, Springer-Verlag, Berlin/New York.
frequency of the field is increased, this directional trans- Wehr, T., and Rodriguez-Diaz, R. (1998). “Capillary Electrophoresis of
port of the whole molecule in the solution is replaced Proteins,” Dekker, New York.
by rotation of the polarized molecule until eventually at Weinberger, R. (2000). “Practical Capillary Electrophoresis,” Academic
Press, San Diego.
frequencies above 1 GHz all the mass transport ceases. Westermeier, R. (1997). “Electrophoresis in Practice: A Guide to Meth-
Above this frequency the motions of groups and individ- ods and Applications of DNA and Protein Separations,” 2nd ed., Wiley,
ual atoms become increasingly important, and these are New York.
P1: LLL/LPB P2: FJU Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN005F220 June 15, 2001 20:44
Elemental Analysis,
Organic Compounds
T. S. Ma
City University of New York
I. Introduction
II. Determination of Carbon,
Hydrogen, and Nitrogen
III. Determination of Oxygen
IV. Determination of Sulfur
V. Determination of Chlorine, Bromine, and Iodine
VI. Determination of Fluorine
VII. Determination of Arsenic and Phosphorus
VIII. Determination of Metallic Elements
IX. Other Methods for Determining Atomic Ratios
393
P1: LLL/LPB P2: FJU Final Pages
Encyclopedia of Physical Science and Technology EN005F220 June 15, 2001 20:44
CHN analyzer, an apparatus that determines all three el- CO2 , H2 O, and N2 is performed by three pairs of thermal
ements simultaneously on the basis of the chemical reac- conductivity cells connected in series. A trap between the
tions discussed above. Several models of this apparatus are first pair of cells absorbs H2 O from the gas mixture before
available commercially. An example is shown in Fig. 1, it enters the second cell so that the signal is proportional
in which the CHN analyzer is connected to a microcom- to the amount of H2 O removed; another trap between the
puter. With this equipment, after the accurately weighed second pair of cells removes CO2 so that the signal is pro-
sample has been placed in the combustion tube and the portional to the CO2 removed; and the last pair determines
sample weight entered on the keyboard, the analysis is N2 by comparing the remaining sample gas plus helium
started and all operations of combustion: measurement of with pure helium. The instrument is calibrated with a pure
CO2 , and H2 O and N2 ; calculations; and printout of results known organic nitrogen compound such as acetanilide be-
are carried out automatically. Figure 2 shows a schematic fore samples are run.
diagram of the combustion train. The boat that contains Adaptation of the CHN analyzer for total automatic
the organic sample is put in the ladle and introduced into operation when a large number of samples have to be ana-
the combustion tube by opening the sample fitting plug. lyzed within a short period involves changing the combus-
By means of the magnet, the boat is pushed to the high tion train from horizontal to the vertical arrangement. The
heat coil area. The automated process now commences. head of the combustion tube is connected to a turntable
The organic compound is combusted in the pyrolysis tube device known as the autosampler. Figure 3 depicts the
in an oxygen atmosphere under static conditions. The car- schematic flow diagram of the PE 2400 CHN Analyzer
bon dioxide, water vapor, and nitrogen oxides produced equipped with an autosampler which holds 60 samples in
are then swept by a stream of helium into the reduction the carousel sample tray. Figure 4 shows a complete pic-
tube, where nitrogen oxides are converted to nitrogen gas ture of this instrument connected to the microprocessor
and the excess oxygen is removed. The gaseous mixture and microbalance. Instead of being placed in an open mi-
is conducted into the analyzer, where the measurement of croboat, each sample is weighed in a tin or an aluminum
P1: LLL/LPB P2: FJU Final Pages
Encyclopedia of Physical Science and Technology EN005F220 June 15, 2001 20:44
FIGURE 2 Schematic diagram of the combustion train in the Perkin-Elmer 240C Elemental Analyzer. [Courtesy of
Perkin-Elmer Corporation.]
capsule. The capsule is then sealed and placed in its cor- the gaseous mixture containing nitrogen, carbon dioxide,
responding location in the 60-sample tray. Sample weight and water vapor is conducted through two infrared absorp-
and identification are stored in memory. As the automatic tion cells where H2 O and CO2 are measured, respectively.
analysis mechanism is activated, the capsules drop sequen- Helium serves as the carrier gas.
tially from the turntable into the combustion tube. The re-
action products N2 , CO2 , and H2 O are separated by frontal
D. Kjeldahl Method for Nitrogen Determination
chromatography and measured by thermal conductivity,
respectively. Oxygen is used during the combustion, while Kjeldahl discovered in 1883 that when agricultural ma-
helium serves as the carrier gas for chromatography. At the terials were heated with concentrated sulfuric acid, the
completion of each sample analysis, results are printed out nitrogen originally present in the organic matter was
to record the identification number, sample weight, and C, transformed into ammonium bisulfate. He accomplished
H, and N percentages. the determination of nitrogen by liberation of ammonia
In the Leco CHN Analyzer, infrared spectroscopy is uti- through the action of a strong alkali and titration of the am-
lized to determine carbon and nitrogen. After combustion, monia. The chemistry involved can be depicted as follows:
FIGURE 3 Schematic flow diagram of the PE 2400 CHN Analyzer. [Courtesy of Perkin-Elmer Corporation.]
P1: LLL/LPB P2: FJU Final Pages
Encyclopedia of Physical Science and Technology EN005F220 June 15, 2001 20:44
FIGURE 4 The complete assembly of the PE 2400 CHN Analyzer. [Courtesy of Perkin-Elmer Corporation.]
FIGURE 9 Flow diagram of the Leco Analyzer for CHNS/O. The pyrolysis furnace is for oxygen determination.
[Courtesy of Leco Corporation.]
the oxidation furnace containing copper oxide and silver A special sulfur test apparatus is used for the determina-
tungstate. Heating at 1000◦ C in an oxygen atmosphere tion of sulfur in petroleum products by lamp combustion.
produces CO2 , H2 O, N2 , nitrogen oxides, and sulfur ox- The apparatus consists of the burner flask, chimney, spray,
ides. The gaseous mixture then passes through the reduc- and absorption tube. The sulfate that is collected can be
tion furnace containing copper metal at 600◦ C in a helium determined by a turbidimetric method.
atmosphere, whereupon all nitrogen oxides are converted
to N2 and sulfure oxides to SO2 . With helium as the carrier
gas, the mixture is conducted through three infrared (IR)
V. DETERMINATION OF CHLORINE,
cells to measure H2 O, SO2 , and CO2 , sequentially. Finally,
BROMINE, AND IODINE
these three components are removed by Anhydrone (mag-
nesium perchlorate) and Lecosorb (sodium hydroxide),
A. Decomposition Methods
which leaves N2 to be measured by thermal conductivity.
The Perkin-Elmer Analyzer uses gas chromatography to The nature of the organic substance containing chlorine,
separate all four components. Neither CHNS/O method is bromine, or iodine determines what is the best method
applicable to the analysis of materials containing metallic for decomposition. On the one hand, organic compounds
elements which form thermally stable sulfates. that have ionizable halogens, such as the alkaloid halides,
P1: LLL/LPB P2: FJU Final Pages
Encyclopedia of Physical Science and Technology EN005F220 June 15, 2001 20:44
are soluble in aqueous solution, which liberates the halide For the determination of iodine, one mode of finish con-
ions, which can be titrated directly. Similarly, some com- sists of titration with standardized mercuric nitrate solu-
pounds can be dissolved in a nonaqueous solvent like tion, with diphenylcarbazone as the indicator:
ethyl alcohol or liquid ammonia; then addition of metallic
sodium dislodges the halogen from the organic molecule 2I− + Hg(NO3 )2 → HgI2 + 2NO−
3. (21)
and produces sodium halide. On the other hand, certain Another method suitable for the determination of small
compounds, especially the polyhalogenated ones, require amounts of iodine in organic materials involves the am-
drastic reactions at high temperatures to destroy the or- plification technique. The iodide ions obtained after de-
ganic molecule in order to convert the halogens into ionic composition of the sample are oxidized to iodate by the
forms. The closed-flask combustion technique (see Fig. 7) addition of bromine in an acetate buffer. Excess bromine is
is commonly used for this purpose. The absorption liquid removed with formic acid. Then the iodate is determined
for chlorine or bromine contains sodium hydroxide and by liberation of iodine on addition of iodide in sulfuric
hydrogen peroxide and that for iodine contains hydrazine acid solution, followed by titration of the liberated iodine
sulfate, so that chloride, bromide, and iodide, respectively, with standardized sodium thiosulfate solution with starch
are obtained as the final products. as the indicator. The sequence of reactions can be depicted
Another technique for decomposing organic halogen as follows:
compounds utilizes a metal bomb constructed of nickel.
acetate
Commercial metal bombs are available in two sizes: a buffer
I− + 3Br2 + 3H2 O −−−→ HIO3 + Br− + 5HBr,
2.5-mL bomb for decomposing up to 50 mg of organic
material, and a 22-mL bomb that can handle as much as (22)
0.5 g of sample mixed with 15 g of solid reagents. The
Excess Br2 + HCOOH −−−→ 2HBr + CO2 , (23)
reagents are sodium peroxide and sucrose or potassium
nitrate. After being locked tightly, the bomb is heated. HIO3 + 5KI + 5H2 SO4 −−−→ 3I2 + 3H2 O + 5KHSO4 ,
Vigorous oxidation reactions take place, resulting in the
(24)
transformation of the chlorine, bromine, and iodine origi- and
nally present in the organic substance to chloride, bromide,
and iodate, respectively. 3I2 + 6Na2 S2 O3 −−−→ 6NaI + 3Na2 S4 O6 . (25)
experimental procedure should be strictly controlled in is conducted to a tube containing gold foil, which traps the
order to get reproducible results. mercury to produce amalgam. The increase in weight of
Arsenate can be determined by an iodimetric method in the gold tube gives the content of mercury in the sample.
which it is reduced by iodide in acidic solution to produce Amalgamation in a gold trap is also employed in envi-
iodine: ronmental analysis of organic mercury compounds such
+ − as methyl mercury. The mercury is commonly determined
4 + 2H + 2I → I2 + AsO3 + H2 O.
AsO3− 3−
(29)
by cold vapor atomic absorption spectroscopy.
The iodine liberated is then titrated with a standardized When an organic mercury compound is decomposed by
sodium thiosulfate solution [see Eq. (25)]. digestion in a mixture containing nitric acid, sulfuric acid,
If the organic sample contains a low percentage of ar- and potassium permanganate, the resultant mercuric ions
senic, arsenate in the digestion mixture is preferably re- are preferably determined by titration with standardized
duced by means of sodium borohydride to liberate arsine, potassium thiocyanate solution:
which is measured by atomic absorption spectrometry. H2 SO4 ,HNO3 ,KMnO4
Organic mercury compound −−−−−−−−−−−−→ Hg2+
(33)
VIII. DETERMINATION OF
METALLIC ELEMENTS and
analysis of total phosphorus and Kjeldahl nitrogen,” Anal. Chem. 68, Ma, T. S. (1997). “Organic elemental analysis,” In “Analytical Instru-
2610. mentation Handbook” (G. W. Ewing, ed.), Chap. 3, Marcel Dekker,
Hara, T. et al. (1977, 1978, 1980, 1981, 1982, 1983). “Simultaneous New York.
determination of the atomic ratio between C, H, O, and N by the Ma, T. S., Gutterson, M., and Wang, C. Y. (1982, 1984, 1986, 1988,
pyrolytic sulfurization method,” Bull. Chem. Soc. Jpn. 50, 2292; 51, 1990). “Fundamental reviews—Organic elemental analysis,” Anal.
1110, 2951, 3079; 53, 951, 1308; 54, 2956; 55, 329, 2127, 3450, Chem. 54, 87R; 56, 88R; 58, 144R; 60, 175R; 62, 78R.
3800; 56, 1378, 3615. Ma, T. S., and Hassan, S. S. M. (1982). “Organic Analysis Using
Hughes, S. K., Brown, R. M., Jr., and Fry, R. C. (1981). “Photodiode array Ion-Selective Electrodes,” Vol. 2, Academic Press, London.
studies of near infrared and red atomic emissions of C, H, N, and O Yu, W. L. et al. (1981). “Development and application of a prototype
in the argon inductively coupled plasma,” Appl. Spectrosc. 35, 396. (GC)2 -MFS hyphenated apparatus,” In “Proceedings of the 4th
Kirsten, W. (1983). “Organic Elemental Analysis,” Academic Press, International Symposium on Capillary Chromatography” (R. E.
New York. Kaiser, ed.), p. 445, Hüthig, Heidelberg.
P1: GQT Final Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
Gas Chromatography
Milos Novotny
Indiana University
.455
P1: GQT Final
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
a Gas–gas and solid–solid equilibria do occur in nature; however, they are impractical for chromato-
graphic separations.
b C = Chromatography.
chromatography were discovered by a Russian botanist (a decade later, recognized by a Nobel Prize in Chem-
M. S. Tswett (1872–1919) but hardly developed into use- istry). Several investigations pertaining to the use of gas
ful chemical separation procedures until the 1930s. The as the mobile phase in gas/adsorption systems were re-
name chromatography was originated by Tswett who ported in Austria, Czechoslovakia, Russia, and Sweden
primarily investigated plant pigments (chromatos is the during the 1940s. However, the development of gas–liquid
Greek name for color). However, any method that utilizes chromatography, reported in 1952 by A. T. James and
a distribution of the molecules to be separated between the A. J. P. Martin, is widely considered the beginning of GC
mobile phase (a gas or a liquid) and the stationary phase (a as a powerful analytical method.
solid or a liquid that is immiscible with the mobile phase) Today, GC is complementary to other separation meth-
now qualifies as chromatography. The physical state of ods. It can be practiced on either a small (analytical) scale
the mobile phase determines whether we deal with gas or or a large (preparative or industrial) scale. The prepara-
liquid chromatography. tive uses of GC are relatively uncommon. While typical
Variation in the type of stationary phase is important as amounts of chemical substances analyzed by the modern
well: if a solid is used as the stationary phase, the inter- GC are between the microgram (10−6 g) and nanogram
action of the molecules under separation with it is due to (10−9 g), samples as small as a femtogram (10−15 g) can
adsorption forces; if a liquid is used in the same capacity, be measured in special circumstances. Importantly, con-
the molecules under separation interact with it based on temporary GC can often simultaneously recognize up to
their solubilities. According to this type of interaction, we several hundred chemical substances.
distinguish between adsorption chromatography and parti-
tion chromatography. This classification is further evident
in Table I. II. PHYSICAL PRINCIPLES
Tswett’s original work pertained to liquid adsorption
chromatography, while the first experiments on liquid par- The apparatus designated to separate compounds by GC
tition chromatography were described in the early 1940s is called the gas chromatograph. Its essential parts are
by A. J. P. Martin and his co-workers in Great Britain shown in Fig. 1. At the heart of the system is the separation
column, at which the crucial physicochemical processes stationary phase, the individual sample components even-
of the actual compound separation occur. tually form their own concentration bands, which reach
The separation column contains the stationary phase, the column’s end at different times. A detector is situated
while the mobile phase (frequently referred to as the car- at the column’s end to sense and quantitatively measure
rier gas) is permitted to flow through this column from a the relative amounts of these sample components.
pressurized gas cylinder (source of the mobile phase). The The detector, together with auxiliary electronic and
rate of mobile-phase delivery is controlled by a pressure recording devices, is instrumental in generating the chro-
and/or flow-regulating unit. An exclusive separation mode matogram, shown in Fig. 2. Such a chromatogram is, ba-
for the analytical GC is elution chromatography, in which sically, a plot of the sample concentration versus time. It
the sample (a mixture of chemicals to be separated) is in- represents the individual component bands, separated by
troduced at once, as a sharp concentration impulse (band), the chromatographic column and modified by a variety
into the mobile-phase stream. The unit where sample in- of physical processes into a peak shape. The position of
troduction is performed is called the injector. The unfrac- a peak on the time scale of the total chromatogram bears
tionated sample is transferred from the injector into the some qualitative information, since each chromatographic
chromatographic column, where it is subjected to a con- peak represents at least one chemical substance. The ar-
tinuous redistribution between the mobile phase and the eas under the peaks are, however, related to the amounts
stationary phase. Due to their different affinities for the of individual substances separated in time and space.
FIGURE 2 Chromatogram.
P1: GQT Final
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
A typical gas chromatograph has three independently Unlike some other chromatographic processes, the
controlled thermal zones: proper temperature of the in- physical interactions between the mobile phase and solute
jector zone ensures rapid volatilization of the introduced molecules in GC are, for all practical purposes, negligible.
sample; the column temperature is controlled to optimize Thus, the carrier gas serves only as means of molecular
the actual separation process; and the detector must also (solute) transport from the beginning to the end of a chro-
be at temperatures where the individual sample compo- matographic column. The component separation is then
nents are measured in the vapor phase. For certain GC primarily due to the interaction of solute molecules with
separations, it is advisable to program the temperature of those of the stationary phase. Since a variety of column
the chromatographic column. materials are available, various molecular intertactions
As shown in Fig. 2, different sample components appear can now be utilized to enhance the component separation.
at the column’s end at different times. The retention time Moreover, these interactions are temperature-dependent.
tR is the time elapsed between injection and the maximum For the mixture component with no affinity for the sta-
of a chromatographic peak. It is defined as tionary phase, the retention time t0 serves merely as the
marker of gas linear velocity u (in cm/s) and is actually
tR = t0 (1 + k), (1)
defined as
where t0 is the retention time of a mixture component that L
has no interaction with the stationary phase (occasionally t0 = , (3)
u
referred to as dead time), and k is the capacity factor. The where L is the column length. The gas velocity is, in turn,
capacity factor is further defined as related to the volumetric flow rate F since
Vs F
k=K , (2) u= , (4)
VM s
where K is the solute’s distribution coefficient (pertaining where s is the column cross-sectional area. The gas-flow
to a distribution between the stationary phase and the mo- rate is chiefly regulated by the inlet pressure value; the
bile phase), Vs is the volume of the stationary phase, and higher the inlet pressure the greater the gas-flow rate (and
VM is the volume of the mobile phase in a chromatographic linear velocity) becomes, and consequently, the shorter t0
column. The distribution coefficient K = Cs /CM (where is. The retention time tR of a retained solute is also mod-
Cs is the solute concentration in the stationary phase and ified accordingly. Correspondingly, fast GC separations
CM is the solute concentration in the mobile phase) is a are performed at high gas-inlet pressures. The so-called
thermodynamic quantity that depends on temperature as retention volume VR is a product of the retention time and
do all equilibrium constants. The molecular interactions volumetric gas-flow rate:
between the phases and the solutes under separation are
VR = tR F. (5)
strongly temperature-dependent. If, for example, a solid
adsorbent (column material) is brought into contact with Since the retention times are somewhat indicative of
a permanent (inorganic) gas and a defined concentration the solute’s nature, a means of their comparison must be
of organic (solute) molecules in the gas phase at a certain available. Within a given chemical laboratory, the relative
temperature, some solute molecules become adsorbed on retention times (the values relative to an arbitrarily chosen
the solid, and others remain in the permanent gas. When chromatographic peak) are frequently used:
we elevate the system temperature, less solute molecules
tR 2 VR K2
are adsorbed, and more of them join the permanent gas; α2,1 = = 2 = . (6)
the distribution (adsorption) coefficient, as defined above, tR1 VR1 K1
changes correspondingly. Likewise, if the stationary phase This equation is also a straightforward consequence of
happens to be a liquid, the solute’s solubility in it decreases Eqs. (1) and (2). Because the relative retention represents
with increasing temperature, according to Henry’s law, re- the ratio of distribution coefficients for two different so-
sulting in a decrease of the distribution (partition) coeffi- lutes, it is frequently utilized (for the solutes of selected
cient. chemical structures) as a means to judge selectivity of the
According to Eqs. (1) and (2), the retention time in GC solute–column interactions.
depends on several variables: (a) the chemical nature of For interlaboratory comparisons, the retention index ap-
the phase system and its temperature, as reflected by the pears to provide the best method for documenting the GC
distribution coefficient; (b) the ratio of the phase volumes properties of any compound. The retention index system
in the column Vs /VM ; and (c) the value of t0 . In the practice compares retention of a given solute (on a logarithmic
of chromatography, these variables are used to maximize scale) with the retention characteristics of a set of stan-
the component separation and the speed of analysis. dard solutes that are the members of a homologous series:
P1: GQT Final
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
chromatographic peak. This number is simply calculated The column efficiency N can be dependent on a number
from the measured retention distance tR (in length units) of variables. Most importantly, the plate height is shown
and the peak width at the peak half-height W1/2 : to be a function of the linear gas velocity u according to
the van Deemter equation:
tR 2
N = 5.54 . (8) B
w1/2 H = A+ + Cu, (10)
u
The length of a chromatogrphic column L is viewed as
divided into imaginary volume units (plates) in which a where the constant A describes the chromatographic band
complete equilibrium of the solute between the two phases dispersion caused by the gas-flow irregularities in the col-
is attained. Obviously, for a given value of tR , narrower umn. The B-term represents the peak dispersion due to
peaks provide greater numbers of theoretical plates than the diffusion processes occurring longitudinally inside the
broader peaks. Turning once again to Fig. 3, we see that column, and the C-term is due to a flow-dependent lack
cases (a) and (b) represent low column efficiencies (plate of the instantaneous equilibrium of solute molecules be-
numbers), while case (c) demonstrates a high-efficiency tween the gas and the stationary phase. The mass transfer
separation. between the two phases occurs due to a radial diffusion of
Equation (8), used to determine the number of theoret- the solute molecules.
ical plates, relates to a perfectly symmetrical peak (Gaus- Equation (10) is represented graphically by a hyperbolic
sian distribution). While good GC practice results in peaks plot, the van Deemter curve, in Fig. 6. The curve shows the
that are nearly Gaussian, departures from peak symmetry existence of an optimum velocity at which a given column
occasionally occur. In Fig. 5, (a) is usually caused by a exhibits its highest number of theoretical plates. Shapes of
slow desorption process and undesirable interactions of the van Deemter curves are further dependent on a number
the solute molecules with the column material, and (b) is of variables: solute diffusion rates in both phases, column
associated with the phenomenon of column overloading dimensions and various geometrical constants, the phase
(if the amount of solute is too large, exceeding saturation ratio, and retention times. Highly effective GC separations
of the stationary phase, a fraction of the solute molecules often depend on thorough understanding and optimization
is eluted with a shorter retention time than the average). of such variables.
When feasible, GC should be carried out at the solute con-
centrations that give a linear distribution between the two
phases. III. SEPARATION COLUMNS
The length element of a chromatographic column oc-
cupied by a theoretical plate is the plate height (H ): Since the introduction of GC in the early 1950s, many
different column types have been developed, as is widely
L
H= . (9) documented by numerous column technology studies re-
N ported in the chemical literature. The column design is ex-
tremely important to the analytical performance and util-
ity for different sample types and applications. The most
important features include (a) type of column sorption ma-
terial (in both physical and chemical terms), (b) column
diameter, (c) column length, and (d) surface characteris-
tics of a column tubing material. A proper combination
of these column design features can often be crucial to a
particular chemical separation.
Based on their constructional features, GC columns can
be divided into three main groups: packed columns, capil-
lary (open tubular) columns, and porous-layer open tubu-
lar columns. Their basic geometrical characteristics are
shown in Fig. 7.
A packed column is basically a tube, made from glass
or metal, that is filled with a granular column material.
The material is usually held in place by small plugs of a
glass wool situated at each column’s end. During a GC
FIGURE 5 Departures from peak symmetry: (a) slow desorption run, such a column is attached to the instrument through a
process and (b) column overloading. (c) Gaussian distribution. gas-tight connection; the carrier gas is forced through the
P1: GQT Final
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
FIGURE 6 Relationship of the plate height and linear gas velocity (van Deemter curve).
free channels between the individual particles, while the reversible, preserving the chemical integrity of the so-
sample molecules are allowed to interact with the parti- lutes (unlike in some forms of contact catalysis, where
cles. Typical column inner diameters are 1–4 mm, and the a strong compound adsorption precedes chemical conver-
lengths are 1–3 m, although departures from these dimen- sion). Consequently, not all adsorbing solids qualify as
sions may exist for special applications. The inner diam- suitable column packings in GC. Examples of suitable
eters of preparative columns can be considerably larger. GC adsorbents are silica gel, alumina, zeolites, carbona-
The granular packing can be either an adsorbent (if the ceous adsorbents, and certain porous organic polymers.
method of choice is gas–solid chromatography) or an inert Surface porosity and a relatively large surface area are
solid support that is impregnated with a defined amount among the characteristic features of GC adsorbents. For
of a liquid stationary phase (for gas–liquid chromatog- example, certain synthetic zeolites, molecular sieves, may
raphy). In either case, packing materials with uniformly have a specific surface area as high as 700–800 m2 /g.
small particles are sought, as the column performance is Specificity of certain solute-adsorbent interactions is
strongly dependent on the particle size. In fact, a distinct a major advantage of gas–solid chromatography. Various
advantage of small particles is their closer contact with adsorbents readily discriminate between different molec-
diffusing sample molecules and a greater number of the ular geometries of otherwise similar solutes (e.g., geo-
mentioned equilibrium units (i.e., theoretical plates). Be- metrical isomers). At present, however, major difficulties
cause extremely small particles present a great hindrance exist as well: (1) large distribution coefficients (com-
to gas flow, materials with a particle size between 100 and pared with partitioning liquids) result in long retention
150 µm are typically used as a sensible compromise be- times; (2) the separaton process can often be strongly de-
tween the column efficiency and technological limitations pendent on sample size, which is a serious problem for
of high gas pressure at the column inlet. analytical determinations; (3) the physical processes in
In gas–solid chromatography, the solute molecules in- adsorption chromatography are less amenable to a rig-
teract with the surface of solid adsorbents through rela- orous theoretical description compared with gas-partition
tively weak physical adsorption forces. Such weak forces chromatography; and (4) the current adsorbent technology
are desirable, because the adsorption process must be does not permit an effective suppression of minor catalytic
P1: GQT Final
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
Si OH Si O CH3
H3C Cl
+ Si O Si
O -2HCl
H3C Cl
Si OH Si O CH3
FIGURE 8 The resolution advantage of a capillary column (a) over a packed column (b), in Calmus oil analysis.
[From Grob, K., and Grob, G. (1979). J. High Resolut. Chromatogr. 3, 109.]
to theoretical description. An example is a description of packed columns, various (less accurate) empirical con-
the physical processes that occur inside such capillary stants must be used.
columns. Equation (10), the van Deemter equation, can Equation (11) is strictly valid only for the cases where
be translated, for the capillary column into the stationary phase film thickness amounts to no more
than a few tenths of a micrometer (thin-film columns).
2DG (1 + 6k + 11k 2 ) r 2 Columns with film thicknesses up to several microme-
H= + u, (11)
u (1 + k)2 24DG ters can also be prepared. Although their efficiencies are
lower than those obtained for the thin-film columns, due to
where DG is the solute diffusion coefficient in the mobile the impaired solute mass transfer, they can tolerate larger
(gas) phase and r the capillary inner radius. The equa- sample amounts without signs of overloading.
tion shows explicitly how the plate height is dependent on Refined aspects of column technology have been cru-
the diffusion processes and the column radius. At low gas cial to the success of GC capillary columns. Early in the
velocities, molecular diffusion significantly increases the development of such columns, metal or plastic tubes were
plate height. At higher velocities, the opposite is true (DG used exclusively. Highly efficient glass capillary columns
is in the denominator), reflecting the fact that the solute were developed at a later stage, and the problems of glass
mass transfer from one phase to another is primarily diffu- fragility were successfully overcome through the technol-
sion controlled. Reducing the column radius is a powerful ogy of fused-silica flexible tubes. Production of fused-
way to increase the column performance. silica capillaries is reminiscent of the fabrication of optical
Note that Eq. (11) is an accurate description of the col- fibers: thin-walled sillica tubes, drawn from a hot zone are
umn processes because the column geometry is well de- immediately protected by an overcoat of a stable organic
fined. Although somewhat similar equations exist for the polymer (Fig. 9).
P1: GQT Final
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
mechanism has been utilized. Synthesized optically active While most GC determinations are performed with so-
polymers are highly effective in resolving various racemic lute quantities between 10−6 and 10−9 , certain selective
mixtures. Certain metal chelates, used as additives to the detectors can reach down to the 10−15 -g levels, represent-
common stationary phases, can retain selected solutes ing some of the most sensitive measurement techniques
through the formation of reversible complexes. Finally, available to the chemist. Some GC detection principles
highly organized liquids (such as various liquid crystals) are based on the measurement of certain transport prop-
tend to retain more strongly the molecules of elongated erties of the solutes (e.g., thermal conductivity or optical
rather than bulky structures. properties), while other detectors are transducers, mea-
Through advances in synthetic chemistry and polymer suring ultimately some product of a solute molecule (e.g.,
research, new GC stationary phases will become available. gas-phase ionization products). The latter detectors are
Additional column selectivities can also be achieved by destructive to the solutes.
mixing the existing stationary phases with each other, in
suitable proportions.
1. The Thermal-Conductivity Detector
This detector, occasionally referred to as the hot-wire de-
IV. DETECTORS AND ANCILLARY tector or katharometer, operates on the basis of measuring
TECHNIQUES the difference in thermal conductivity of pure carrier gas
and the carrier gas plus a solute. Typically, the column ef-
The detector has an extremely important role in the overall fluent is passed through a thermostatted cavity (measuring
process of GC analysis. The current popularity and success cell) that contains a resistor element heated by passage of
of GC as an analytical method is attributable in great part a constant current. Various changes in the thermal conduc-
to the early development of highly sensitive and reliable tivity of the surrounding gas causes the element tempera-
means of detection. In sensing the vapor concentration ture (and, consequently, its electric resistance) to decrease
at the column outlet, the detector provides information on or increase. Pure carrier gas is passed, under the same
the distribution of individual peaks within a chromatogram conditions, through a reference cell of identical design.
(which compound?) as well as their relative amounts (how The resistor elements of both cells are parts of a Wheat-
much?). The area measured under a chromatographic peak stone bridge circuit that records any imbalance caused by
is generally related to the quality of the compound. the passage of individual solutes.
Many detection principles in GC were investigated over In thermal-conductivity measurements, it is advisable
the years, but only a few pass the criteria of reliability to choose a carrier gas that differs maximally from the
needed for precise analytical measurements. Detectors organic solutes (e.g., hydrogen or helium). The detector
can broadly be classified as universal or selective. Uni- is a truly universal and simple device, but its sensitivity
versal detectors measure all (or nearly all) components is marginal; at best, submicrogram amounts are detected.
of a mixture, although their response to the same quan- The thermal-conductivity detector is most typically em-
tities of different compounds is seldom similar. Selective ployed for the analysis of permanent gases and light hy-
detectors respond only to mixture components that pos- drocarbons.
sess a unique structural feature in their molecules. For
example, a typical gasoline sample contains a number of
2. The Flame-Ionization Detector
organic components which, after being separated by an
appropriate chromatographic column, are all detected by This detector is the workhorse of GC. It operates on the
a universal detector. However, if a lead-selective detector basis of decomposing the solute-neutral molecules in a
is used instead, only a few peaks are recorded, those due flame into charged species and electrically measuring the
to the lead-containing additives in gasoline, while the re- resultant changes of conductivity. A cross-sectional view
maining mixture constituents are ignored. The so-called of a flame-ionization detector is shown in Fig. 10. A small
ancillary techniques go a step further as highly selective flame is sustained at the jet tip by a steady stream of pure
detectors, because they actually characterize the individ- hydrogen, while the necessary air (oxidant) is supplied
ual GC peaks qualitatively. through the diffuser. At the detector base, the column
effluent is continuously introduced, mixed with hydro-
gen, and passed into the flame. Conductivity changes be-
A. DETECTORS
tween the electrodes are monitored, electronically ampli-
The most important analytical properties of a GC de- fied, and recorded. A conventional carrier gas contributes
tector are sensitivity, linearity over an extensive concen- little to the flame conductivity; however, when organic so-
tration range, long-term stability, and ease of operation. lute molecules enter the flame, they are rapidly ionized,
P1: GQT Final
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
TABLE II Properties of Some Gas Chromatography tral thermal fragments. Most typically, the samples under
Selective Detectors investigation are large and nonvolatile compounds, such
Approximate as synthetic or natural polymers. A reproducible pyroly-
Detector Selectivity mode sensitivity (g) sis/GC process results in the formation of pyrograms that
are often highly indicative of some structural details of
Electron-capture Affinity to low-energy 10−13 –10−14
electrons
the original substance; both the presence of certain chro-
Thermionic Nitrogen 10−12
matographic peaks and their areas are judged. As small as
Phosphorus 10−13
submicrogram samples have been successfully analyzed
Flame-photometric Sulfur 10−9
by this combination.
Phosphorus 10−11
Design of a precolumn pyrolysis unit and the method
Electrolytic-conductivity Halogen compounds 10−10
of thermal degradation are crucial to the acquisition of
Ultraviolet Aromatics 10−9
diagnostically useful pyrograms. Sample size and the py-
Photoionization Partially enhanced 10−11 –10−12
rolysis temperature must also be carefully controlled. The
response to certain three most common pyrolysis techniques use (a) filament
organic molecules as (ohmical) heating, (b) rapid warp-up of a ferromagnetic
compared with flame conductor in a high-frequency field (Curie-point pyroly-
ionization (not sis), and (c) direct thermal degradation in a heated quartz
truly selective)
tube. In each case, the sample is deposited from its so-
lution onto a suitable matrix, and the solvent is dried off
prior to pyrolysis. Alternatively, small pieces of solids are
The most common GC selective detectors are listed in directly pyrolyzed.
Table II together with their analytically important features. Pyrolysis/GC is used extensively in the analysis of poly-
These selective detectors have been finding an increasing mers, paints, textile fibers, and even whole microorgan-
utilization in the analysis of environmental and biological isms. Certain materials of forensic interest have been char-
mixtures. As seen from Table II, sensitivities at the low acterized by this approach. A unique pyrolysis/GC system
nanogram level are very common, while some detectors was aboard the Viking 1975 Mission spacecraft to investi-
reach levels even below picogram amounts. Parallel uses gate the possible occurrence of organic compounds in the
of a nonselective and a selective detector are quite popular martian soil.
in chemical identification efforts.
While GC is a powerful separation method, it provides Mass spectrometers are sophisticated instruments that
only limited information on the chemical nature of the sub- work on the principles of compound ionization and frag-
stances it so effectively separates. Consequently, it has to mentation (typically through the bombardment by elec-
be combined with ancillary techniques. These are certain trons or selected ions), the physical separation of the
sample manipulative techniques that are coupled in either charged fragments, and their detection. The information
a precolumn or a postcolumn arrangement to GC. Their obtained by mass spectrometry is a mass spectrum (ion
purpose is to enhance qualitative information about the intensity versus mass) that is highly indicative of the
sample, to characterize it chemically, or ideally, to deter- sample’s original structure, virtually a fingerprint of a
mine unequivocally its structure. Some of these ancillary molecule. Consequently, the method provides a power-
techniques chemically alter the sample during the process; ful means to identify various organic compounds but
others measure only its physical parameters, such as op- works more effectively with pure substances than with
tical spectra. On occasion, ancillary tools may represent substance mixtures. The combination of GC with mass
instruments that are considerably more sophisticated and spectrometry provides an ideal analytical system, in which
expensive than the GC instrumentation itself. The three the complex mixtures are first separated, and the mass
GC ancillary techniques discussed below are among the spectrometer is permitted to analyze the substances, one
most powerful and illustrative of this direction. at a time.
Commercial instruments that combine the two tech-
niques vary in several respects. The low-resolution in-
1. Pyrolysis/GC
struments provide the designation of nominal molecular
This combination is an example of the precolumn arrange- weights, while the high-resolution instruments can work
ment. Pyrolysis/GC combines a controlled thermal degra- up to the precision of a small fraction of such nomi-
dation of a sample with the subsequent separation of neu- nal masses. For example, a low-resolution masses. mass
P1: GQT Final
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
spectrometer “sees” the proton (1 H) as the mass 1 and graphs quite sophisticated instruments with precise elec-
oxygen (16 O) as 16; a high-resolution instrument can mea- tronic and pneumatic controls.
sure the same species as 1.0078 and 15.9949, respectively. The carrier gas and the auxiliary gases for detectors are
Consequently, the high-resolution instruments are capable controlled by a set of pneumatic devices (pressure regula-
of providing measurements of exact elemental composi- tors, flow-controllers, and restrictors) to assure (a) repro-
tion for various compounds. Different physical principles ducibility of the column flow rate, and thus retention times,
of mass separation are involved with these instruments. in multiple analyses; (b) adjustment of the gas linear veloc-
Importantly, both the low- and high-resolution mass spec- ity for optimal column efficiencies; and (c) reproducibil-
trometers can be combined with GC. The methods also ity of detector response for reliable quantitative measure-
strongly overlap with respect to the amounts necessary ments. In addition, filtering devices are inserted in the gas
for analysis. lines to purify all gases mechanically and chemically.
At first, a coupling of GC and mass spectrometry en- Type and design of the injection port are crucial to
countered technological difficulties because the gas chro- performing separations with different types of chromato-
matograph operates at gas pressures above atmospheric graphic columns. Different physical dimensions of the
pressure, while most mass spectrometers operate at high packed and capillary columns cause substantial differ-
vacuum. To overcome these difficulties, molecule separa- ences in the optimum volumetric flow rates. While typical
tors were developed. These devices, working on princi- values for conventional capillary columns range around 1
ples such as molecular effusion, the jet separation effect, ml/min, various packed columns pass one to two orders
and preferential adsorption on a membrane, selectively of magnitude greater gas flows. The volumes of injected
remove most carrier gas, reduce pressure in the interface, samples must be adjusted accordingly. In a typical sam-
and allow most sample molecules to pass into an evacuated pling procedure with a packed column, liquid samples of
mass spectrometer. The process of coupling GC to mass up to a few microliters are injected by a miniature syringe,
spectrometry is further aided by modern pumping tech- through a rubber septum, into the hot zone of the injection
nology. In fact, modern combination instruments need no port. Rapid sample evaporation and transfer into the first
molecule separators for capillary columns (typical flow section of the column are feasible because of a sufficiently
rates around 1 ml/min). high flow rate of the carrier gas.
Contemporary GC/mass spectrometry instruments are Considerably smaller samples are necessary for the
greatly aided by computers, which can control various in- much narrower capillary columns. Since small fractions
strumental parameters, provide data reduction, and com- of a microliter can be neither reproducibly measured nor
pare acquired mass spectra with the extensive libraries of easily introduced into the capillary GC system, indirect
many thousands of previously recorded spectra. sampling techniques are employed. In a commonly used
sampling method, a sample volume of approximately 1µl,
3. GC/Infrared Spectroscopy or slightly less, is injected into a heated T-piece, where an
uneven separation of the vaporized sample stream occurs.
Infrared (IR) spectra of organic compounds are charac- While the major part of the sample is allowed to escape
teristic of various functional groups in the molecules. IR from the system, a small fraction (typically, less than 1%)
spectral information is somewhat complementary to mass enters the first section of a capillary column. Sampling
spectral information. Therefore, the combination of GC devices based on this principle are called splitting injec-
with IR spectroscopy is, after GC/mass spectrometry, the tors or splitters. They are generally adequate in situations
second most important structural identification tool. Since where samples with high concentrations of the analyzed
conventional IR spectroscopy is less sensitive than most substances are encountered.
GC detectors, the necessary sensitivity enhancement is Other ways of indirect sampling onto a capillary column
achieved through the use of Fourier transform techniques. involve the injections of (relatively nonvolatile) samples
With the advent of refined optical systems and fast compu- diluted in a sufficiently large (measurable) volume of a
tational techniques, the combination of GC with Fourier- volatile solvent (which serves as a sample “vehicle”). With
transform IR spectrometry is becoming widely used, al- the column inlet kept at a sufficiently low temperature, the
though its sensitivity is currently less than that of mass nonvolatile sample trace is trapped at the inlet and focused
spectrometry. Special optical cells were designed for the into a narrow zone, while the volatile solvent is allowed to
purposes of this combination. pass through the column and widely separate from the
sample. A subsequent increase of temperature permits
V. INSTRUMENTATION the sample zone to desorb from its inlet position and enter
the usual separation process.
The variety of GC analytical applications, columns, and Most sample introduction techniques in GC have now
specialized techniques make the modern gas chromato- been automated. Process automation permits repeatable
P1: GQT Final
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
analysis and unattended operation of the instrument. to use appropriate standard compounds. Some automated
Moreover, reproducibility of the sample injection is im- analyzers based on GC principles are also used in the pro-
proved considerably. cess control and continuous analysis of industrial streams.
A precise column temperature control is now required Specialized GC techniques find their place in scientific
for all commercial gas chromatographs. In practice, the research. The extremely high sensitivities of some GC de-
GC ovens are designed to have low thermal mass. Resis- tectors are unparalleled.
tance spirals situated inside the oven are proportionally The GC method is employed for a variety of mixtures,
heated, while the air circulation throughout the oven is ranging from permanent gases up to molecules that are al-
provided by a fan. For adequate analytical work, the col- most as large as 1000 Da of molecular weight. The variety
umn temperature should be reproducible within at least of chromatographic columns and detectors available to GC
±0.1◦ C. continues to expand its applicability to various analytical
Retention times in GC are affected by temperature. In problems. Several representative examples will now be de-
accordance with Eqs. (1) and (2), the retention time de- scribed to demonstrate the method’s versatility, resolving
creases with increasing temperature because the partition power, selectivity, and sensitivity. These examples have
coefficient is decreased. Various solutes, depending on been chosen from the areas of industrial analysis, occu-
their structures and the chemical nature of a particular pational hygiene, and biochemical research. Other major
phase system, have different dependencies on tempera- areas, not covered here, are geochemistry, food and aroma
ture. Consequently, temperature optimization is necessary analysis, various agricultural and environmental analyses,
for the maximum resolution of the analyzed components. atmospheric measurements, and forensic investigations.
For the mixtures of components with very different val- The analysis of light gases (permanent gases, gaseous
ues of partition coefficient, column temperature program- oxides, and C1 –C5 hydrocarbons) has been traditionally
ming is often employed. Commercial gas chromatographs performed in gas–solid chromatographic systems. Vari-
are equipped with convective heating systems that facili- ous porous adsorbents possess the capability to adsorb and
tate linear temperature programs. As the column temper- separate these relatively small molecules. An example is
ature is being gradually raised from a certain initial value shown in Fig. 12, where the carbon molecular sieve col-
to the maximum permissible temperature for an analy- umn (6 ft × 1/8 in. i.d.) rapidly resolved a mixture consist-
sis, the sample components with increasingly higher boil- ing of air, methane, carbon dioxide, acetylene, ethylene,
ing points are eluted from the column. According to the
needs of analysis, programming rates are adjustable from
as slow as 0.5◦ C/min up to 30◦ C/min. Nonlinear and mul-
tistep temperature programs are also feasible for special
applications.
Modern GC utilizes sensitive detectors. As the mea-
sured detector signals (changes in current, voltage, etc.)
are quite small, electronic signal amplification is neces-
sary. Since the gas chromatographs are further provided
with integrating devices and small computers to calculate
exact retention times and peak areas for quantitation, the
signals are converted to their digital forms. In addition to
displaying a chromatogram on a recorder, modern GC in-
struments are capable of performing some advanced tasks,
such as computing relative retention values, adjusting the
detector baseline, and performing certain forms of data
reduction.
VI. APPLICATIONS
FIGURE 13 Capillary chromatogram of a gasoline sample. [From Adlard, E.R., Bowen, A.W., and Salmon, D. G.
(1979). J. Chromatogr. 186, 207.]
and ethane at 150◦ C (thermal conductivity detection was petrochemical interest are very complex, so the highly
used). In particular, the separation of acetylene and ethy- efficient capillary columns are frequently utilized. An ex-
lene is industrially important. While air (a mixture of two ample of major-component analysis is shown in Fig. 13,
major components, oxygen and nitrogen) is eluted here where a full-range gasoline sample has been resolved into
as a single peak, there exist other GC adsorbents that can a substantial number of components. A 70-m long capil-
separate oxygen from nitrogen. lary column was employed, the column temperature was
The petrochemical industries have long utilized GC as programmed from 0 to 95◦ C, and the flame-ionization
the analytical method for characterization of various fos- principle was used in the peak detection. A neat gasoline
sil fuels, in monitoring the efficiency of distillation proce- sample was injected (using a sample-splitting technique).
dures, cracking processes, various chemical conversions, Environmental pollution is among the chief concerns of
identification of oil spills, and so on. Most samples of our industrial society. Highly sensitive analytical methods
FIGURE 14 Chromatogram of trace aromatic amines (after preconcentration) from the atmosphere of a film-
processing laboratory. [From Becher, G. (1981). J. Chromatogr. 211, 103.]
P1: GQT Final
Encyclopedia of Physical Science and Technology EN006F-275 June 29, 2001 21:12
FIGURE 16 Capillary GC separators of a racemic mixture of 19 amino acids on an optically active stationary phase.
[From Frank, H., Nicholson, G. J., and Bayer, E. (1978). J. Chromatogr. 167, 187.]
Infrared Spectroscopy
Norman B. Colthup
American Cyanamid Company
I. Basic Theory
II. Instrumentation
III. Sample Handling Techniques
IV. The Near Infrared Region
V. Quantitative Analysis
VI. Group Frequencies
793
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
THE INFRARED (IR) spectrum results from the interac- vibration absorption in the IR. The low wave number limit
tion of radiation with molecular vibrations and, in gases, is more variable since it is more or less an instrumental
with molecular rotations. The spectrum itself is a plot of limitation. The region between the visible and the mid-IR
sample transmission of IR radiation as a function of wave- regions is called the near-IR region. This region of the IR
length or related units. Infrared spectroscopy is the physics has been used for many applications, especially for quan-
that deals with the theory and interpretation of this spec- titative analysis. The region beyond ∼50 µm (200 cm−1 )
trum and is one of the most popular techniques for identi- is called the far-IR region. This region is used for studying
fying molecules. The IR spectrum can be used as a type of low-frequency vibrations and some molecular rotations.
“fingerprint” unique to a molecule. In addition, the pres- Electronic transitions give rise to absorption in the ul-
ence or absence of many chemical functional groups such traviolet and visible regions of the spectrum, and pure
as phenyls and carbonyls usually can be established from rotations of gaseous molecules give rise to absorption in
the spectrum. Quantitative analyses of mixtures can be the far-IR and microwave regions of the spectrum. In-
obtained. Infrared spectra can be run for liquids, solids, tramolecular vibrations of molecules give rise to absorp-
or gases without special difficulties. Different types of tion throughout most of the IR region.
spectrometers can be used, and a wide variety of sam-
ple handling techniques are available, many of which are B. Infrared Spectra Coordinates
described in this article.
Some examples of IR spectra are given in Figs. 1 through 6.
Chemical group vibrations associated with spectral bands
are indicated. These are all liquids run in a 0.01-mm-thick
I. BASIC THEORY NaCl cell. The horizontal coordinates for IR spectra are
usually either linear with wavelength (µm) or linear with
A. Electromagnetic Spectrum
wave number (cm−1 ) with generally a factor of 2 scale
Electromagnetic radiation can be characterized by its change at 2000 cm−1 . An advantage of the wave number
wavelength λ, its frequency ν, or its wave number ν̄. In the scale is that the wave number of the radiation is propor-
IR region the unit used for wavelength is the micrometer tional to its frequency and to photon energy, and these
(µm). The frequency unit is cycles per second or hertz properties are related to the frequencies and energies of
(Hz). The wave number unit is cycles per centimeter or molecular vibrations.
reciprocal centimeters (cm−1 ). The wave number (cm−1 ) The vertical coordinate in a single-beam spectrum is a
is the number of waves in a continuous wave sequence 1 measure of the intensity of the radiation of a given wave
cm long. The relationship between the units is given in number that has passed through a sample and reached the
Eq. (1): detector. Usually, this spectrum is ratioed with another
single-beam reference spectrum without a sample to give
1 104
ν̄ (cm−1 ) = ν̄ (cm−1 ) = a ratioed or double-beam spectrum. The transmittance T is
λ (cm) λ (µm) the intensity of the sample single-beam spectrum divided
ν (Hz) (1) by the intensity of the reference single-beam spectrum
ν̄ (cm−1 ) = . at the same wave number. The vertical coordinate most
c (cm/sec)
commonly seen in an IR spectrum is linear with percent
From this, one can see that wave number (cm−1 ) is equal transmittance (%T ), which is transmittance T multiplied
to the reciprocal of the wavelength (cm) or is equal to by 100.
104 times the reciprocal of the wavelength (µm). The wave
number (cm−1 ) in a vacuum is also equal to the frequency
(Hz) divided by c, the velocity of light in a vacuum given
in centimeters per second. This makes the wave number
proportional to the frequency.
The visible region of electromagnetic radiation extends
from about 0.38 to 0.78 µm. The IR region extends from
the end of the visible region at 0.78 µm to the microwave
region with a wavelength of ∼1 mm. The IR region is
usually divided into three sections. The section used most
by chemists is the mid-IR region extending from 2.5 µm,
or 4000 cm−1 , to ∼50 µm, or 200 cm−1 . The division at FIGURE 1 Infrared spectrum of chloroform in a 0.01-mm NaCl
4000 cm−1 is the high wave numberlimit for fundamental cell.
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
FIGURE 2 Infrared spectrum of 1-decene in a 0.01-mm NaCl FIGURE 4 Infrared spectrum of n-amyl acetate in a 0.01-mm
cell. NaCl cell.
Another vertical coordinate scale used increasingly is to restore equilibrium. In the diatomic molecule, there is
the linear absorbance scale, where absorbance A is given only one bond and only one vibration, that which peri-
by the log10 of the transmittance reciprocal: odically changes the length of the bond. In the harmonic
A = log10 (1/T ). (2) oscillator approximation, the restoring force on each mass
is assumed to be linearly proportional to the bond length
The advantage of an absorbance scale is that the ab- change from the equilibrium bond length. The proportion-
sorbance is proportional to the product of sample thickness ality constant is called the force constant k. The vibrational
and concentration. This is discussed in section V.A. frequency v of the diatomic molecule can be calculated by
the methods of classical mechanics, as in Eq. (3):
C. Diatomic Vibrations 1/2
1 1 1
When molecular vibrations are studied, it is useful to con- ν= k + . (3)
2π m1 m2
sider the nucleus of any atom in the molecule as a mass
concentrated at a single point, held in place by chemical Here ν is the vibrational frequency, m 1 and m 2 the two
bonds that act much like massless coil springs. The di- atomic masses, and k the force constant, namely, the
atomic molecule is the simplest molecule type, consisting restoring force on either mass divided by the bond length
of two nuclei connected by a chemical bond formed by change from equilibrium at any time. Note that the fre-
electrons. The electrons in a molecule move much more quency is independent of the vibrational amplitude. The
rapidily than the nuclei, so the electrons can quickly equi- masses move in a manner that keeps the center of mass
librate into new electronic structures as the slowly moving stationary. If a1 /a2 is the relative amplitude of the two
nuclei change their spacing. This means that there is a def- masses m 1 and m 2 during the vibration, then
inite potential energy for each nuclear configuration. For a1 m2
=− . (4)
the diatomic molecule this is something like the poten- a2 m1
tial energy of an ordinary coil spring. There is a certain
Stretching force constants are usually expressed in
equilibrium bond length where the energy is a minimum.
millidynes per angstrom (mdyne/Å); 1 mdyne/Å equals
If this length is increased or decreased, the potential en-
100 N/m, the equivalent SI unit. Masses are usually ex-
ergy increases and a restoring force is generated, tending
pressed in unified atomic mass units (carbon = 12). If these
FIGURE 3 Infrared spectrum of 2-heptanone in a 0.01-mm NaCl FIGURE 5 Infrared spectrum of n-decyl alcohol in a 0.01-mm
cell. NaCl cell.
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
units are used, the wave number ν̄ of the radiation that has FIGURE 7 Normal modes of vibration of carbon dioxide.
the same frequency as the molecular vibration is given by
1/2
1 1 internal degrees of freedom change the size or shape of
ν̄ = 1303 k + . (5) the molecule without rotating it or translating its center of
m1 m2
mass.
Some examples of diatomic molecules that absorb in the It can be shown by the methods of classical mechanics
IR are listed in Table I. Equation (5) relates the wave num- that the 3n − 6 (or 3n − 5) internal degrees of freedom of
ber of the absorption band given to the force constant and motion correspond to 3n − 6 (or 3n − 5) different normal
atomic masses of the molecule. modes of vibration. In a normal mode of vibration the
Cartesian displacement coordinates of every atom change
periodically, each oscillating with the same frequency and
D. Polyatomic Vibrations
passing through the equilibrium configuration at the same
Polyatomic molecules have more than one vibrational fre- time. The molecule does not translate its center of mass
quency. The number can be calculated from the follow- or rotate.
ing. One atom in the molecule can move independently in The vibrational form can be described by specifying
three directions, the x, y, and z directions in a Cartesian the relative amplitudes of the Cartesian displacements of
coordinate system. Therefore, in a molecule with n atoms, each mass (Fig. 7). The vibration can also be described in
the n atoms have 3n independent ways they can move. terms of the relative changes in the internal coordinates
The center of mass of the molecule can move in three in- of the molecule, namely, changes in the bond lengths and
dependent directions, x, y, and z. A nonlinear molecule bond angles. For example, in Fig. 7 the CO2 molecule
can rotate in three independent ways about the x, y, and has two bonds; in one vibration, both bonds stretch at the
z axes, which pass through the center of mass. A linear same time (inphase stretch), whereas in another vibration
molecule has one less degree of rotational freedom since one bond stretches while the other bond contracts (out-
rotation about its own axis does not displace any atoms. of-phase stretch). There are two mutually perpendicular
These translations of the center of mass and rotations can bending vibrations that have the same frequency.
be performed with a rigid molecule and do not change its For each normal mode of vibration, a single coordinate
shape or size. Substracting these motions, there remain can be defined called a normal coordinate. When one nor-
3n − 6 degrees of freedom of internal motion for non- mal mode of vibration is activated, one normal coordinate
linear molecules and 3n − 5 for linear molecules. These periodically changes in value. At the same time that the
normal coordinate changes, each Cartesian displacement
coordinate changes in a specified proportion (positive or
TABLE I Diatomic Molecules negative) to the change in the normal coordinate, so that
Wave number Force constant the resulting motion is a normal mode of vibration. Nor-
Molecule (cm−1 ) (mdyne/Å) mal coordinates are very useful for theoretical studies of
molecular vibrations.
CO 2143 18.6
NO 1876 15.5
HF 3962 8.9 E. Infrared Absorption
HCl 2886 4.8
In a spectrometer, a source of IR radiation sends all IR
HBr 2559 3.8
wavelengths of interest through a sample. The IR radiation
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
causes some of the molecules to vibrate with increased oscillating dipole moment in the molecule. If the radiation-
amplitude, which increases the vibrational energy. The induced dipole moment oscillation has the same frequency
increase in vibrational energy is at the expense of radi- as the dipole moment oscillation resulting from a molecu-
ation energy, resulting in the absorption of IR radiaition lar vibration, then the radiation may induce the molecule
at certain frequencies. The absorption frequency of a fun- to vibrate with increased amplitude. If a vibration causes
damental absorption band in the IR is the same as the no change in dipole moment, then there is no way the radi-
frequency of the molecular vibration that caused the ab- ation can excite that vibration. In a homonuclear diatomic
sorption. This provides a means of observing molecular molecule such as H2 or Br2 , the two atoms have identical
vibrational frequencies that can provide a molecular “fin- excess charges (namely, zero), and the dipole moment does
gerprint” for identifying molecules. The frequencies can not change during the vibration as it is always zero. There
also be used to characterize internal features within the is no way the electric field of the radiation can induce the
molecule that can provide information about the molecu- two similar atoms to move in opposite directions as re-
lar structure. quired in a vibration. The selection rule for IR absorption
requires that in order to be IR-active, a molecular vibration
must cause a change in dipole moment. The IR absorp-
F. Dipole Moment Change
tion intensity is proportional to the square of the change
There must be some means by which the radiation energy in dipole moment with respect to the change in the normal
can be transferred to the molecule when the molecule ab- coordinate. This reflects the fact that the more the dipole
sorbs radiation energy. This involves the dipole moment moment changes during a vibration, the greater the proba-
of the molecule. A dipole consists of a positive and a neg- bility that the radiation of the proper frequency can excite
ative charge of equal magnitude separated by a distance. that vibration.
The dipole moment is the magnitude of either charge mul-
tiplied by the spacing. Within a molecule, we can picture
G. Symmetry and Infrared Activity
atoms as particles with small excess negative or positive
charges, since chemical forces act to make some atoms If a molecule has some symmetry, a particular vibration
have a slight excess or deficiency of electrons. We can of that molecule may be IR-inactive; that is, the vibration
picture the negative charge of the dipole as the total ex- will not give rise to any IR absorption. This is because the
cess negative charge of the negative atoms concentrated dipole moment change can be zero as a direct consequence
at the center of the excess negative charge and can pic- of the symmetry.
ture the positive charge of the dipole as the total excess One symmetry element is the center of symmetry. If a
positive charge on the positive atoms concentrated at the molecule in the equilibrium configuration has a center of
center of the excess positive charge. In CO2 , for exam- symmetry, one can start at any atom and go in a straight line
ple, the center of excess negative charge is between the through the center and an equal distance beyond, where
two electronegative oxygens and the center of excess pos- one will find another atom of identical type. An example
itive charge is at the relatively electropositive carbon. At is carbon dioxide O C O (see Fig. 7). Such a molecule at
equilibrium, these two charge centers coincide with zero equilibrium would have a dipole moment of zero. The vi-
spacing, so the dipole moment is zero. brationally distorted molecule where each atom has moved
Many molecular vibrations cause the dipole moment to to the end of its displacement vector may have less sym-
change. For example, in HBr, the bromine is more elec- metry than the molecule at equilibrium. One can repeat
tronegative than the hydrogen, so the bromine has a slight this procedure for the vibrationally distorted molecule.
excess negative charge and the hydrogen a slight excess If it still has a center of symmetry, the vibration is said
positive charge. During the vibration, the H–Br spacing to be symmetric with respect to the center of symmetry.
changes and also the amount of excess charge on each An example in Fig. 7 is the in-phase stretch of O C O,
atom changes, causing the dipole moment to change. where both CO bond lengths are always equal. Such a
The electromagnetic radiation imposes an electric field vibration is IR-inactive since the dipole moment (which
on the molecule. This electric field exerts forces on is always zero) does not change. If the vibrationally dis-
charges, and by definition the forces on positive and neg- torted molecule no longer has a center of symmetry, then
ative charges are oppositely directed. The atoms with ex- the vibration is said to be antisymmetric with respect to
cess negative charge are pulled in one direction, while the the center of symmetry. An example in Fig. 7 is the out-
atoms with excess positive charge are pulled in the op- of-phase stretch vibration of O C O, where the positive
posite direction. These forces tend to induce a change in carbon is not a longer midway between the two negative
dipole moment. The electric field of the radiation oscil- oxygens. The dipole moment changes during this vibra-
lates at the radiation frequency, and this tends to induce an tion, which is IR active.
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
Symmetry elements other than the center of symmetry able, but has discrete values given from quantum mechan-
include planes of symmetry and two-fold or higher axes ics as
of symmetry. When a plane of symmetry is present, the
E = v + 12 hν◦ , v = 0, 1, 2, . . . (8)
plane can be thought of as a mirror. When each atom in
the molecule is moved to the position of its mirror im- Here E is the vibrational energy, h Planck’s constant, ν◦ the
age, the resulting configuration is indistinguishable from classical vibrational frequency of the oscillator, and v the
the original. When a twofold axis of symmetry is present, quantum number, which can have only integer values. In
the molecule can be rotated by half a full circle to give a the classical oscillator, the lowest possible energy is zero
configuration indistinguishable from the original. A full when there is no vibration. In the quantum mechanical
discussion of symmetry and group theory cannot be given oscillator, the lowest possible energy is 12 hνo , which is not
here. However, molecules that do not have a center of zero, so the molecule can never stop vibrating entirely.
symmetry may have IR-inactive vibrations as a conse- This state where v = 0 is called the ground vibrational
quence of these other symmetry elements. For example, state.
the tetrahedral sulfate ion SO2−
4 does not have a center of
If the vibrational energy is to be increased, the quantum
symmetry, but the in-phase stretch of the four SO bonds number v must be increased. When the quantum number
is IR-inactive. The four negative oxygens move radially at is increased by 1, the energy change
E from the previous
the same time, but the symmetry requires that the center equation is
of their excess negative charge does not move relative to
E = hνo . (9)
the more positive sulfur.
A photon has an energy E given by
H. Quantum Mechanical Harmonic Oscillator E = hνp , (10)
The simplest classical harmonic oscillator is a single mass where νp is the frequency of the photon. When the photon
m suspended from the ceiling by a spring that obeys electric field frequency νp is equal to the classical dipole
Hooke’s law. If the mass is pulled down a distance x from moment oscillation frequency νo for this vibration, the
its equilibrium point, the spring length minus its length at photon will have exactly the right energy (
E) needed to
equilibrium is x. A restoring force on the mass is gener- increase the vibrational quantum number by 1.
ated that is proportional to the spring length change. The The transition when the quantum number changes by
magnitude of the restoring force equals kx, where k is the 1 is called an allowed transition in a harmonic oscilla-
force constant. If the mass is held stationary at this point, tor. The most important of these is the transition where
the potential energy PE is the oscillator goes from the v = 0 level to the v = 1 level.
This is called the fundamental transition and is responsi-
PE = 12 kx 2 . (6) ble for most of the strong bands in the IR spectrum. The
This is also the total energy for this condition since the ki- (v = 0 → v = 1) transition is much more probable than
netic energy is zero. If the mass is released, it moves toward the (v = 1 → v = 2) transition because at room tempera-
the equilibrium point and the kinetic energy increases as ture many more oscillators exist in the low-energy v = 0
the potential energy decreases. At equilibrium, the energy state than in the v = 1 state (or higher states).
is entirely kinetic; the mass overshoots the equilibrium In a polyatomic molecule with 3n − 6 different nor-
point and continues on until the energy is again entirely mal modes of vibration, each normal mode of vibra-
potential at maximum amplitude, and the cycle is repeated tion can be treated separately. In the harmonic oscillator
again. Throughout the vibration the total classical energy approximation,
is unchanged and is E = v1 + 12 hν1 + ν2 + 12 hν2 + · · · (11)
E = 12 kxmax
2
, (7) where each mode has its own quantum number v and fre-
quency ν. In the harmonic oscillator only one vibration
where xmax is the maximum amplitude. In a classical vi- may be excited at one time and the quantum number may
bration the maximum amplitude is continuously variable change only by 1.
since one is free to pull out the spring to any length before
it is released to vibrate. This means that the energy of the
I. Effect of Anharmonicity
classical harmonic oscillator is continuously variable and
can have any value. In the single-mass harmonic oscillator discussed, the
Oscillators the size of molecules obey the laws of quan- restoring force is a linear function of the mass displace-
tum mechanics. The vibrational energy of the quantum ment and the potential energy is a squared function of
mechanical harmonic oscillator is not continuously vari- the mass displacement 12 kx 2 . Mechanical anharmonicity
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
results if the restoring force is not a linear function of a fundamental energy level (v2 = 1) of a completely dif-
the mass displacement, in which case the potential energy ferent vibrational mode. This means that, if no pertur-
will have higher-order terms such as cubic and quartic bation occurred, an overtone absorption band would have
terms. Electrical anharmonicity results if the dipole mo- nearly the same frequency as that of the fundamental band
ment change is not a linear function of the mass displace- of a different vibration. If anharmonicity is present, the
ment. If either mechanical or electrical anharmonicity is higher-order terms in the potential energy may cause per-
present, transitions where the quantum number changes turbations between the fundamental and overtone types
by 2 or more will no longer be forbidden in the IR spec- involved, generating new mixed energy levels. The vibra-
trum. This allows overtones to appear in the spectrum. In tion types involved should be those that can be coupled by
a fundamental transition, the quantum number changes by the anharmonic potential function, which requires them
1 and the photon causing the transition has the same fre- to be of the same symmetry type. The perturbation can
quency as the classical dipole moment oscillation. In an become significant when the unperturbed energy level dif-
overtone transition, the quantum number changes by 2 or ference is small. Combination bands, as well as overtones,
more. The photon that has the right energy to change the can be involved in this interaction, which is called Fermi
quantum number by 2 has a frequency twice that of the resonance.
molecular dipole moment oscillation, and in a harmonic- Consider the case where the unperturbed overtone and
type vibration there will be no dipole moment component fundamental nearly coincide. When interaction occurs,
changing at this frequency. In an anharmonic vibration, two strong bands appear in the spectrum, above and below
the dipole moment change is complicated by the anhar- the expected position of the overtone and the fundamen-
monicity, and overtones are allowed in the spectrum. The tal before interaction. Both bands involve the fundamental
overtone intensity depends on the amount of anharmonic- and both involve the overtone. The strong intensity of both
ity. Overtones are usually fairly weak. bands comes from the fact that the fundamental is involved
In a harmonic oscillator, the spacing
E between the in both bands. The frequency spacing is a function of the
energy levels for v = 0, 1, 2 . . . has a constant value hv. perturbation (Table II).
If mechanical anharmonicity is present, the spacing is If the expected frequencies of the unperturbed overtone
no longer exactly constant, which means that overtone and the interacting fundamental are not identical but are
frequencies will not be exactly 2, 3, or more times the still close to one another in frequency, interaction will not
frequency of the fundamental. For example, CHCl3 has be as strong as before. Two bands of unequal intensity
a CH bending fundamental band at 1216 cm−1 and a will be seen at again somewhat wider spacing than that
much weaker CH bending overtone band at 2400 cm−1 . for the two unperturbed bands. The stronger band will
A ketone has a carbonyl stretching fundamental band be nearer the unperturbed fundamental and will involve
near 1715 cm−1 and a much weaker overtone band near more of the fundamental vibration. The weaker band will
3410 cm−1 . be nearer the unperturbed overtone and will involve more
In polyatomic molecules, combination and difference of the overtone vibration. The weaker band will still in-
bands are allowed when anharmonicity is present. In a volve some fundamental vibration, however, which will
combination-type transition one photon excites two differ- cause this “overtone” band to be more intense than an
ent vibrations at the same time to a new excited state where unperturbed overtone.
both vibrational modes have nonzero quantum numbers
(say, v1 = 1 and v2 = 1). If both quantum numbers are 1,
the combination band will appear in the spectrum near the TABLE II Examples of Fermi Resonance
frequency sum of the two fundamentals. In a difference-
Wave number
type transition, the molecule that is already vibrating in an Molecule (cm−1 ) Assignment
excited state for one vibration (say, v1 = 1) absorbs a pho-
ton of the proper energy and changes to an excited state NaNCO 620 NCO bend
of a different vibration (say, v2 = 1). The difference band 1216 a NCO in-phase stretch plus
appears at exactly the frequency difference of the two fun- 1305 overtone of NCO bend
damentals in this case. Combination and difference bands, 2220 NCO out-of-phase stretch
like overtones, are usually fairly weak. C6 H5 CHO 1392 Aldehyde CH in-plane bend
1700 Aldehyde C O stretch
2740 a Aldehyde CH stretch plus
J. Fermi Resonance 2825 overtone of CH bend
In a polyatomic molecule it may happen that an over- a The two bands in parentheses have nearly equal intensities, and both
tone energy level (v1 = 2) has nearly the same energy as involve a fundamental mixed with an overtone of another vibration.
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
K. Molecular Rotation
Pure rotation of molecules in the gaseous state causes ab-
sorption of radiation in the microwave region and to some
extent in the far-IR region. In order for a pure rotation to
absorb radiation, the rotating molecule must have a per-
manent dipole moment. Imagine that the dipole moment
is oriented perpendicularly to the radiation electric field FIGURE 8 Infrared spectra of gas-phase bands for linear
direction. The field exerts forces in opposite directions on molecules and asymmetric top molecules. Unresolved contours
the negative and positive ends of the dipole. This gener- are shown for different types of bands. Asymmetric top molecules
ates a torque, which tends to rotate the dipole moment have different contours for different ratios of the moments of iner-
tia. The parallel band of linear molecules and the B-type band of
and therefore tends to increase the rotational frequency
the asymmetric top have no central peak.
of the molecule. If the rotational frequency increases, the
rotational energy increases at the expense of the radiation
energy. The rotational energy, like the vibrational energy,
caused by the vibration. In polyatomic linear molecules
is not continuously variable but is quantized. Imagine a
such as CO2 and acetylene, IR-active stretching and bend-
rotating linear molecule such as HCl in a certain rota-
ing vibrations cause dipole moment changes parallel and
tional energy state with a quantum number J , which has
perpendicular to the molecular axis, respectively. For par-
an integer value (1, 2, 3, . . . ). The selection rule for pure
allel vibrations, the gas-phase band contour is a broad
rotation states that photon energy absorption can increase
doublet (Fig. 8). In the low-frequency and high-frequency
the quantum number by only 1 to the state J + 1. The
wings of the band, the rotational quantum numbers for
photon that has the right energy to cause this transition
each of the various energy states have decreased and in-
has a frequency intermediate between the classical rota-
creased by 1, respectively, during the vibrational transi-
tional frequencies for the initial (J ) and the final (J + 1)
tion. Perpendicular bands have the same broad doublet
states. Unlike the classical vibrational frequency, the clas-
seen in the parallel bands, but a perpendicular band has
sical rotational frequency of the molecule increases dur-
an additional central peak not seen in the parallel band,
ing the transition, and the oscillating electric field of the
where the rotational energy remains unchanged during the
photon with this intermediate frequency can stay nearly
vibrational transition.
synchronized with the rotating dipole moment throughout
Tetrahedral or octahedral molecules such as CH4 and
the transition. An analysis of the rotational fine structure
SF4 are called spherical tops and have three equal moments
in the spectrum may yield information about the moments
of inertia for rotation about three mutually perpendicular
of inertia of the molecule.
axes. The gas-phase contour is similar to the perpendicular
band of the linear molecule, with two broad wings and a
central peak for all the IR-active vibrations.
L. Gas-Phase Band Contours
Molecules with one threefold or higher axis of symme-
In the vibrational spectrum the molecule usually changes try such as CHCl3 , BF3 , and C2 H6 are called symmetric
from the ground vibrational state (v = 0) to the first excited tops. Two moments of inertia are equal and differ from the
vibrational state (v = 1). When the sample is in the gaseous third unique moment of inertia for rotation about the axis
state, the molecule may change its rotational state at the of threefold or higher symmetry. When a vibration causes
same time it changes its vibrational state. The molecule a dipole moment change parallel to the major symme-
in the ground vibrational state is rotating with a certain try axis, the unresolved gas-phase contour is similar to the
angular momentum. When the molecule ends up at the perpendicular band of the linear molecule, a broad doublet
first excited vibrational state, it may be rotating with an with a central peak. When the dipole moment change is
increased or decreased angular momentum. As a result perpendicular to the major symmetry axis, the band struc-
of the rotational energy changes, rotational structure is ture is more complex and the unresolved contour depends
superimposed on the vibrational band, which is referred on the relative magnitudes of the moments of inertia.
to as a vibration–rotation band. In molecules with less symmetry, the three moments
If the molecular moments of inertia are sufficiently low of inertia are different. These are called asymmetric tops.
and the spectrometer has adequate resolution, rotational The axes with minimum and maximum moments of in-
fine structure can be resolved in the vibration–rotation ertia are called the a and c axes, respectively, and the
band. For larger molecules, the fine structure is usually axis with intermediate moment of inertia is called the b
unresolved, resulting in a broad band. The contour of the axis. Vibrations with dipole moment changes parallel to
band may reveal the direction of the dipole moment change these axes are called A, B, and C bands. The unresolved
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
the monochromator toward the detector. When the grating ically at 45◦ . Ideally the beam splitter transmits half the
is rotated to a slightly different angle, the path length dif- radiation striking it and reflects the other half. One type of
ference for beams from adjacent grooves will be slightly beam splitter is a thin layer of germanium coated on an IR-
different, so radiation with a slightly different wavelength transmitting support. The transmitted and reflected beams
will pass through the monochromator. leave the beam splitter at right angles, and both strike mir-
When the spectrometer is set at a given wavelength, the rors, which return the two beams to the beam splitter. The
sample beam and reference beam alternately pass through two beams recombine at the beam splitter and show in-
the monochromator and activate the detector. If the two terference. The radiation leaving the beam splitter may
beams do not have the same intensity because of sam- go back to the source or may go at right angles, passing
ple absorption, an alternating signal is generated and is through the sample and going on to the detector.
used to measure the percent transmission of the sample One of the two mirrors is movable, so its distance from
at that wavelength. The grating angle is changed, and the the beam splitter can be varied. The path length differ-
whole spectrum is generated wavelength by wavelength. ence for the two beams in the interferometer is called the
Usually, several gratings are used for the whole spectral retardation and is two times the displacement of the mov-
range, and a grating may be used in more than one order. able mirror from the equidistant point. If a monochromatic
As the wavelength increases, the slit is widened to allow source such as a laser is used, the radiant energy reaching
more energy through to compensate for decreased source the detector will vary as a cosine function of the retarda-
emission at long wavelengths. tion. The detector response will reach a maximum every
time the retardation is an integral number of wavelengths
of the radiation. At this time the beams from the two mir-
C. Fourier Transform Infrared Spectrometers rors combine at the beam splitter in phase for the beam
In a Fourier transform infrared (FT-IR) spectrometer, there going to the detector and show constructive interference.
is no monochromator to disperse or separate the radiation If the movable mirror is then moved one-quarter of a wave-
by wavelength. Instead, a whole single-beam spectrum is length of radiation, the retardation is changed by one-half
generated all at once. The intensities of all of the wave- of a wavelength. The beams from the two mirrors com-
length elements are analyzed simultaneously. Since all the bine at the beam splitter one-half of a wavelenth out of
radiation frequencies reach the detector at the same time, phase for the beam going to the detector and show de-
there is a large signal-to-noise ratio. This is called the structive interference. The detector response as a function
multiplex or Fellgett advantage and is one of the principal of the retardation is called the interferogram. The spec-
advantages that an FT-IR spectrometer has over a disper- trum can be generated from the interferogram by a Fourier
sive instrument. This advantage is particularly noticeable transform. The Fourier transform of a single cosine wave-
for low-energy conditions or where scale expansion is re- type interferogram is a single wavelength, in this case that
quired to bring out very weak bands. There is also an of the laser source.
advantage in that the spectrum can be recorded in less If a polychromatic source is used, its spectrum can be
time. thought of as a series of closely spaced laserlike emission
The unique part of an FT-IR spectrometer is the lines, each with its own wavelength and intensity. Each
Michelson-type interferometer (Fig. 10). Radiation from of these generates a cosine function-type interferogram.
the source is made parallel and strikes a beam splitter, typ- The interferogram of a polychromatic source of radiation
is a summation of all the cosine functions for each of the
laserlike resolution elements. An interferometer does not
produce a spectrum but produces an interferogram. A com-
puter must be used to perform the Fourier transform, which
generates the spectrum from the interferogram. Once this
is done, the computer is available for further processing
of the spectrum.
Just one scan of the movable mirror produces a whole
single-beam spectrum. However, a spectrum produced
from one scan has a relatively high noise level. Usually,
a number of scans are taken and signal-averaged by the
computer. The noise is reduced by the square root of the
FIGURE 10 Michelson-type interferometer. Left: Source radia-
tion is transmitted and also reflected by a beam splitter to two mir- number of scans. A single-beam spectrum with the sam-
rors. Right: Both mirrors reflect radiation back to the beam splitter, ple in place is stored in the computer memory. A reference
where interference occurs. single-beam spectrum is taken without the sample and is
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
also stored in the memory. These two single-beam spectra can be polished the same way, but with alcohol substi-
are ratioed by the computer to give a percent transmittance tuted for water. The best polish comes when the lap is
spectrum. nearly dry.
The computer can be used to modify the spectrum fur-
ther. For example, the vertical or horizontal scale can be C. Liquid Samples
expanded, the background can be straightened, or a linear
absorbance scale can be generated. A useful procedure is The easiest samples to run on IR instrumentation are those
spectral subtraction whereby, for example, a solvent spec- in the liquid state. Slightly viscous samples can be simply
trum can be subtracted from a solution spectrum to yield squeezed between two polished IR-transmitting plates and
the pure solute spectrum. run as a thin film. A typical film thickness is ∼0.01 mm.
If the liquid is not viscous, usually a spacer is added be-
tween the plates to keep the plates apart at the appropriate
III. SAMPLE HANDLING TECHNIQUES spacing. Spacer material can be metal foil or an insolu-
ble polymeric film. Two strips roughly 10 by 2 mm can
A. Infrared-Transmitting Materials be used, for example, one on each side of the area the
One of the features of IR spectroscopy is that solids, liq- IR beam will pass through. These are called temporary
uids, and gases can be run without special difficulties. cells and are disassembled and cleaned after each use.
Usually, some sort of IR-transmitting material is needed The thickness cannot be accurately reproduced.
to support or enclose the sample. Materials such as glass Fixed cells are not disassembled after use but instead
and quartz are useful as windows in the near-IR but even are filled, emptied, and cleaned with solvent through ports
thin windows do not transmit much below 3000 cm−1 . The on the cell assembly. The liquid enters the leak-proof sam-
low wave number transmission limits of IR-transmitting ple area between the plates through holes in the cell win-
materials are not sharply defined but depend on the win- dow. These are used for volatile liquids or when the thick-
dow thickness. Four commonly used materials and their ness needs to be accurately known or held constant as
approximate low wave number limits are NaCl, 600 cm−1 ; in quantitative analysis. Many commercially available de-
KBr, 350 cm−1 ; CsBr, 250 cm−1 ; and CsI, 200 cm−1 . signs are used, and cells come in thickness from 0.01 to
These materials are all water-soluble. Water-insoluble ma- 4 mm.
terials and their low wave number limits include CaF2 , If the cell windows are sufficiently flat, the cell thickness
1200 cm−1 ; BaF2 , 850 cm−1 ; Irtran-2, 700 cm−1 ; AgCl, can be measured by running the IR spectrum of the empty
350 cm−1 ; and KRS-5 250 cm−1 . Irtran-2 is made of zinc cell and observing interference fringes in the form of per-
sulfide an is often used for water solutions or for making cent transmission undulations. Wave number ν̄1 (cm−1 ) is
films from water solution. Silver chloride is useful but is read at one transmission maximum, and wave number ν̄2
soft, deforms easily, and darkens with exposure to light. is read at another transmission maximum that is 1, 2, 3, or
KRS-5 is thallium bromide iodide and is often used in the more generally n maximum away from the first. The cell
internal reflection technique to be discussed later. In the thickness t is
far-IR, high-density polyethylene transmits to as low as 5n
t (mm) = . (12)
30 cm−1 but cannot be used above 600 cm−1 because of ν̄1 − ν̄2
its absorption. Interference results because part of the beam is twice re-
flected inside the cell and is retarded by twice the cell
B. Salt Polishing thickness relative to the transmitted beam with which it
interferes.
Sodium chloride can be easily polished between use. The
crystal is sanded flat with a finegrade sandpaper if it is
D. Gas Samples
freshly cleaved, scratched, or damaged by water. Two pol-
ishing laps are prepared. There are different types, but the Gas cells used for IR spectroscopy come in a variety of
wet lap can be simply two thicknesses of fine nylon cloth types. The simplest is a basic cylinder 10 cm long with
stretched over a flat surface. The wet lap is wetted with IR-transmitting windows on each end. These may be ce-
water and sprinkled with a little fine polishing powder mented on or clamped in place, with vacuum-tight gaskets
such as aluminum oxide or cerium oxide. This is rubbed providing the seal. Entrance and exit tubes are provided
smooth and all excess water is wiped off. The flat salt plate and fitted with stopcocks. The cell is filled and emptied
is rubbed about 20 strokes on the wet lap and then, without with a gas handling system.
delay, is buffed about 7 strokes on the dry lap, which can be The sampling chamber of most IR spectrometers is not
simply a layer of diaper cloth held flat. Cesium bromide large enough to accommodate longer cell lengths directly.
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
However, cells of much longer path length are available can be put onto a plate and the solvent evaporated to form
that use mirrors to deflect the IR beam and to reflect it back a film. This is a good technique for running water-soluble
and forth many times in the cell chamber before it leaves polymers, for example. Sometimes a film can be prepared
the cell and reenters the spectrometer. These long-path- on a substrate and stripped off and run as an unsupported
length cells are used for detecting very small quantities of film. If a film is too uniform in thickness, interference
gas in pollution studies, for example. fringes similar to those from an empty cell may be seen,
A technique often used with FT-IR instruments is gas as discussed earlier. If a film is too irregular in thickness,
chromatography, or GC–FT-IR. Here the effluent from a a spectrum with a false percent transmittance will result
gas chromatography column is fed through a heated light from the fact that different parts of the beam go through
pipe with IR-transmitting windows on the ends. Source ra- sample areas with different thicknesses.
diation passes through the cell into the FT-IR spectrometer.
The gas chromatography column separates the gas-phase G. Mulls
components and ideally sends them one by one through
One of the best techniques for running crystal-line solids
the light pipe, where the high speed of the FT-IR instru-
is the use of a mineral oil or Nujol mull. Here a few mil-
ment is utilized to get the spectrum of each component
ligrams of sample are finely ground with a small amount
“on the fly,” so to speak.
of mineral oil to make a thick paste like cold cream, for
example. The paste can be prepared with a mortar and
E. Solution Spectra pestle and spread between two IR-transmitting windows.
A well-ground sample has a brownish color like smoke
The techniques for running solids in IR are quite varied.
when one looks through it. Most beginners do not grind
In the first case a solid can be dissolved in a suitable sol-
the sample well enough and use too much oil. Mineral oil
vent and run as a liquid. Unfortunately, no solvent is free
has only a few bands in narrow regions. The CH stretch
of absorption in the IR region and, usually, the better the
region between 3000 and 2800 cm−1 and the CH bend re-
solvent, the greater its absorption. This means that more
gion at about 1460 and 1375 cm−1 are obscured, however.
than one solvent must be used to get the whole IR solution
If information is needed in these regions a second mull
spectrum in all regions. A commonly used pair of solvents
must be prepared using a halogenated oil such as Halocar-
are CCl4 above 1330 cm−1 and CS2 below 1330 cm−1 .
bon or Fluorolube, which contain CF2 and CFCl groups
These can be used in cells 0.1–1 mm thick, for example,
but no CH. These have no bands from 4000 to 1300 cm−1
with solute concentrations in the range 10–1%. In double-
but have strong bands below 1300 cm−1 . Some people
beam grating spectrophotometers a cell of matching thick-
use the halogenated oil spectrum above 1300 cm−1 and
ness containing solvent only can be put into the reference
the mineral oil spectrum below. In this case care must be
beam to compensate for the solvent bands. In FT-IR in-
taken to ensure that the sample thickness is the same in
struments, a reference solvent spectrum can be subtracted
both preparations.
from the solution spectrum to remove solvent bands. An-
other common solvent for solution spectra is CHCl3 often
H. Potassium Bromide Disks
used in 0.1-mm cells with 5 to 10% solute. CHCl3 has
strong bands at 1216 and 757 cm−1 , where solute infor- A very popular technique for running solids is the KBr
mation is often lost or inadequately presented. Even water disk technique. Here a few milligrams of sample are very
has been used as a solvent for some applications. Here finely ground and then mixed with 50 to 100 parts of dry
the cell thickness must be kept small, as water is a very KBr powder. The mixture is placed in a special device and
strong IR absorber. The internal reflection technique de- compressed into a disk at high pressure. If all goes well, a
scribed in Sec. III.I has been successfully used for water transparent disk results, which is put into the spectrome-
solutions. ter and run. Commercial KBr disk makers are available in
many forms. Some are activated with wrenches or levers,
while others are used with a hydraulic press. Some can be
F. Films
evacuated, which gives the disk transparency a longer life-
Solid-state films of suitable thickness can be prepared time, but this is not necessary if the disk is used promptly.
from melts or solution. Such films are most suitable Advantages of the disk over the mull include the fact
for amorphous materials, especially polymers. Crystalline that KBr, unlike mineral oil, has no bands above 400 cm−1 .
films may scatter light and show nonreproducible orienta- Also, many polymers are more easily ground in KBr.
tion effects from special orientations of the crystal on the Microsamples are easier to prepare with the KBr disk.
IR window surface. A sample can be heated between two The KBr disk has disadvantages compared with the mull,
salt plates until molten and allowed to solidify. Solutions however. The biggest problem is that KBr is hygroscopic,
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
I. Internal Reflectance
FIGURE 11 Plate used for internal reflection spectroscopy. The
Internal reflectance results when a beam of radiation inside lower drawing shows the sample in contact with the plate and
a material of relatively high index of refraction is reflected radiation being multiply internally reflected within the plate.
from the surface interface between this and a material of
lower index of refraction. The angle of incidence α is the a few micrometers into the sample on the plate surface
angle between the beam and a line perpendicular to the at each reflection. The sample thickness is immaterial as
surface interface. If the angle of incidence is small, much long as it exceeds a few micrometers. The sample con-
of the radiation is transmitted through the surface interface tact area should go all the way across the plate so none of
and a little is internally reflected. As the angle of incidence the beam can bypass the sample. The lengthwise coverage
gets larger, a certain critical angle αc , is exceeded, after only affects the attenuation intensity.
which all the radiation is internally reflected from the in- When the internally reflected beam is introduced into a
terface and none is transmitted. The sine of the critical spectrometer, the resulting spectrum is similar to a trans-
angle, sin αc is equal to the ratio n2 /n1 , where n1 is the mission spectrum. There is one major difference. Since
higher index of refraction and n2 is the lower index of the radiation penetration is wavelength dependent in the
refraction on the two sides of the interface. penetration equation, longer wavelengths penetrate more.
When the angle of incidence is larger than the critical The internal reflectance spectrum resembles a transmis-
angle, then the beam in the material with the higher index sion spectrum where the sample thickness gets larger in
of refraction penetrates a little into the material with the direct proportion to the radiation wavelength.
lower index of refraction in the form of an exponentially There is another effect on the penetration, and that is that
decaying wave. It is then reflected back out. When the the index of refraction of the sample (n2 ) is not constant.
amplitude of the wave passing through the interface has It changes in the region of an absorption band, becom-
decayed by a factor of (1/e) or about 37%, the reflective ing smaller than average on the high-wave number side of
penetration (d), into the material with the lower index of the band center and larger on the low-wave number side.
refraction is given by From the penetration equation, the radiation penetration,
λ and therefore the band intensity, will be decreased on the
d(1/e) = 1/2 , (13) high-wave number side of the band center and increased
2πn1 sin α1 − (n2 /n1 )2
2
on the low-wave number side. This distorts the band shape.
where e is the natural log base, λ is the wavelength of the To avoid this, the denominator in the penetration formula
radiation, and α1 is the angle of incidence in the material should not get too small. This means that the angle of inci-
with the higher index of refraction. If the material with the dence should not be too small and the index of refraction
lower index of refraction should absorb part of the radia- of the crystal n1 should be relatively high. In Fig. 11, the
tion penetrating into it, then the internally reflected beam angle of incidence can be kept large enough to reduce
leaving the interface will be attenuated by this absorp- the band distortion, but this also reduces the band inten-
tion. Hence we have the name attenuated total reflectance sity. This is compensated for by using multiple internal
(ATR) for this effect. reflections.
As used in infrared spectroscopy, one type of internal There have been many variations in the design of in-
reflectance plate (Fig. 11) is made of a high index of re- ternal reflection accessories. In one design, the plate in
fraction material, such as thallium bromide–iodide. The Fig. 11 is mounted horizontally at the bottom of a shell
plate is usually a few millimeters thick, and the ends are container so that a liquid can be simply spread over the
beveled to allow radiation entry into one end at an angle top of the plate and run without further changes.
inside the plate. The beam is multiply internally reflected In another design, the internal reflectance crystal has
and zigzags between the surfaces until it leaves at the a hemispherical shape. In this arrangement, the radiation
other end. A sample with a lower index of refraction than enters into the curved surface, and is internally reflected off
the plate is pressed into intimate contact with the plate the flat surface, and then exits through the curved surface.
on one or both sides. The zigzagging beam penetrates In one version of this, the sample is in optical contact
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
with only a small raised area of the flat surface that allows The sampling region can be viewed visually through the
spectra to be taken of quite small areas. Since there is only microscope, and selected areas can be isolated by mask-
one internal reflection, the crystal used, such as silicon, has ing off the unwanted parts of the field. Variable aperture
a very high index of refraction. masks are located in remote image planes of the sample
There are many applications for internal reflectance area, located above and below the sample to reduce diffrac-
spectroscopy, and only a few will be mentioned here. In- tion effects. Then, the optical path is changed so that the
ternal reflectance spectroscopy can be used to obtain the source radiation goes through the unmasked areas and the
spectra of rubbery materials that are hard to grind. The infrared spectrum of the sample is recorded. This can be
rubbery material is simply pressed against the internal re- divided by the spectrum of a similarly masked blank for
flectance plate, and it is ready to run. Carbon-filled rubber example, to give a percent transmission spectrum of the
or other polymers may be run using a high index of refrac- sample, or from this, an absorbance-type spectrum of the
tion germanium as the internal reflectance element. Inter- sample.
nal reflectance is used to obtain selectively the top few The types of applications are basically similar to those
micrometers of a sample surface where the composition handled by transmission spectroscopy, but with a signif-
may be different than that further down. It is also good for icant difference. The microscope can yield good spec-
water solutions because the controlled penetration keeps tra on much smaller sample areas. This means that many
the effective sample thickness small. new types of problems can be handled that were difficult
or impossible to solve with standard instruments. Sam-
ple areas that are heterogeneous are now easily measured
J. Diffuse Reflectance
in the chemical industry, forensic work, and biological
Diffuse reflectance is a technique usually used with FT- studies.
IR instruments. A powdered sample is placed in a small In the chemical industry for example, some polymer
container, where source radiation strikes it and is diffusely products may show some very small impurity areas whose
reflected in various directions. This radiation is collected chemical composition may be characterized with little
and measured by the spectrometer. Usually in the mid-IR difficulty by the spectra. Polymer films weathered by
region the finely powdered sample is diluted to 5 to 10% outdoor exposure may show chemical modification as a
with finely powdered KBr or KCl. The spectrum is ra- function of the layer depth below the surface. This may
tioed against a reference spectrum of pure powdered KBr be characterized by the infrared spectra of the various
or KCl. The ratioed spectrum is processed by a computer layers.
using a function f (R∞ ) derived by Kubelka and Munk, In forensic work, heterogeneous fields can be ex-
which changes the reflectance spectrum into one resem- amined selectively for hair strands, polymer fibers, or
bling a linear absorbance spectrum: lint particles, for example, and their chemical com-
position can be characterized by their infrared spec-
(1 − R∞ )2 k R∞ (sample)
f (R∞ ) = = R∞ = . tra. Cross-sections of paint chips can be examined and
2R∞ s R∞ (reference) the various layers can be characterized, which may be
(14) distinctive.
Here (R∞ ) is the reflectance of a thick scattering layer, k In the biological field, heterogeneous areas of various
the molar extinction coefficient, and s a scattering coeffi- biological specimens can be examined, and the infrared
cient, which is a function of particle size. The spectrum is spectrum of different microareas of the field can be taken.
quite sensitive to particle size, which affects the radiation The spectrum of a single red blood corpuscle can be eas-
scattering. Spectral distortion (compared with a transmis- ily measured. One difference that can be observed in these
sion spectrum) may occur if the particle size is not uni- samples is the ratio of protein to fat. Proteins have bands
formly fine. Black, strongly scattering materials such as near 1650 and 1550 cm−1 for the O C-NH group, and fats
coal can be run by this technique. have an ester C O band near 1740 cm−1 and an alkane
chain doublet near 2925 and 2855 cm−1 , with an internal
unsaturation band near 3015 cm−1 for the CH group.
K. Infrared Microspectroscopy
Some studies may reveal chemical differences between
In this method of sampling, a special type of microscope adjacent tissues of various types. The spectral character-
is used to select very small sample areas for examination istics of two-dimensional areas may be mapped by running
by the infrared spectrometer. As in the macroscopic case, a series of spectra of a grid pattern. An array consisting
samples can be prepared with thicknesses on the order of a number of detectors can be used, each one of which
of 0.01 mm. Since infrared radiation must pass through generates the infrared spectrum of its own small area of
the microscope, all the optics are front surface mirrors. the grid pattern.
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
IV. THE NEAR INFRARED REGION TABLE III Selected Near Infrared Characteristic Bands
Group nm cm−1 Assignmenta
A. Qualitative and Quantitative Analyses
CH3 1680–1760 5950–5680 2 x CH3 str.
The region of the near infrared spectrum extends from CH3 2240–2360 4465–4235 CH3 str. + bend
about 700 nm (14,000 cm−1 ) near the end of the visible re- >C CH2 1625–1675 6155–5970 2 x CH2 asym. str.
gion, to 2500 nm (4,000 cm−1 ), which is the beginning of cis–CH CH 2120–2170 4715–4610 CH str. + C C str.
the infrared fundamental region. The HF molecule, for ex- C C H 1520–1565 6580–6390 2 x C–H str.
ample, has its fundamental absorption band at 3962 cm−1 . R OH, H2 Ob 1385–1460 7220–6850 2 x OH str.
The near infrared spectrum of water is shown in Fig. 12. H2 Ob 1880–1950 5290–5130 H2 O str. + bend
The near infrared region is populated with overtone and R2 NH 1520–1560 6580–6410 2 x NH str.
combination bands of lower frequency fundamentals be- R SH 1960–2020 5100–4950 2 x SH str.
low 4,000 cm−1 . These bands are considerably weaker in >C O 1920–1970 5810–5075 3 x C O str.
intensity than fundamental bands. The intensity weakness
a str, stretch.
is an advantage for quantitative analyses, particularly in b Varies with changes in hydrogen bonding.
production processing, because the larger cell thicknesses
used can be better controlled, yielding more reproducia-
ble intensity measurements. The cells can be made with B. Derivative Spectra
water-insoluable quartz or saphire windows for example,
which if they are thin enough, can transmit down to about In the near infrared region, the bands are not only weaker,
3000 cm−1 . On the negative side, the qualitative informa- they can be broader and more overlapping. For example,
tion is considerably less extensive than in the fundamental the near infrared spectrum of oleic acid
region, as most of the near infrared bands involve summa- CH3 (CH2 )7 CH CH(CH2 )7 COOH
tion bands (overtones and combination bands) of CH, OH,
and NH stretching vibrations. Binary summation bands are can be seen in Fig. 13. The overtones of the CH2 group
usually the most intense, with tertiary and higher summa- out-of-phase and in-phase stretch bands show clearly at
tion bands being progressively weaker. Even so, there is 1723 and 1758 nm. However, the expected weaker bands
still some useful information to be had about functional for the CH3 out-of-phase stretch overtone and the CH
groups. Characteristic group frequency charts and tables stretch overtone, which are expected below 1723 nm, are
of these group summation bands can be found in the litera- so broad and weak that they are hard to detect.
ture. A very brief selection of these are shown in Table III For this reason, second derivatives of the spectra are
to serve as some examples. often used, as seen in Fig. 14. Now the strong CH2 overtone
bands at 1726 and 1761 nm are much narrower and better
FIGURE 12 The near infrared spectrum of water run in a 0.1-mm FIGURE 13 The near infrared spectrum of oleic acid run in a
cell with bands at 1450 nm (2 × H2 O stretch) and 1950 nm (H2 O 1-mm cell with bands at 1723 nm (2 x CH2 out-of-phase stretch)
stretch + bend). and 1758 nm (2 x CH2 in-phase stretch).
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
A = abc, (17)
for the “zero sample” condition and use that for the back- Another ratio method is the internal standard method. In
ground point. This can be used but is somewhat subjective this method a known amount of an internal standard mate-
and may not be very reproducible. An alternative is to use rial is added to the sample mixture. The internal standard is
a line tangent to the spectrum at the band wings or even chosen so that it has no absorption at the analytical wave-
connecting some more distant points. The more closely length for the sample and vice versa. Again from Beer’s
the line approaches the true background, the better, but it Law, the absorbance ratio is
is not necessary that the tangent line and the true back- A ac as A
ground coincide exactly. A consistent base line construc- = and c = cs (22)
As a s cs a As
tion should yield a reproducible measured absorbance for
a given concentration, and the absorbance should vary where the subscript s denotes the internal standard. Here
linearly with concentration. Again, this is calibrated with the internal standard concentration cs is known, and the
standards. One of the advantages of the method is that it constant ratio as /a can be determined from one sample of
can reduce or eliminate a correction for the finite back- known concentration, after which the concentration can
ground absorption of the solvent or major component. be calculated from the absorbance ratio.
Some care should be taken in choosing the tangent points
for the base line. These points should not be too sensitive G. Gas State Analysis
to concentration variations of other components that may
be present. In a quantitative analysis involving gases, the concentra-
tion term c in Beer’s law is replaced by the pressure or
partial pressure p of the gas being measured. Beer’s law
F. Ratio Methods reads
In most solid-state spectra, the sample film thickness and A = abp. (23)
the amount of sample in a KBr disk or a Nujol mull are
not known. It is still possible to do quantitative analysis by In this case, the measurement of the background ab-
the ratio method. This is based on the fact that, in a given sorbance is usually that of the evacuated cell or the cell
mixture, the absorbance ratio of any two bands in one filled with nonabsorbing gas such as nitrogen. There is a
spectrum should be independent of the sample thickness. complication in the use of Beer’s law in the vapor phase
Consider a two-component mixture, each component of that is called pressure broadening. Not only is the ab-
which has an analytical band with no interference from sorbance of a gaseous component a function of its partial
the other component. The absorbance ratio for the two pressure; it is also a function of the total pressure. This
analytical bands is, from Beer’s law, means that the absorbance of a gaseous component with a
fixed partial pressure can be changed by introducing an-
A1 a 1 b1 c 1 A1 a 1 c1 other completely nonabsorbing gas such as nitrogen. The
= or = . (19)
A2 a 2 b2 c 2 A2 a 2 c2 rotational fine structures of gas-phase bands are broadened
by collisions between the molecules of the component be-
The b values are identical and cancel since the absorbances ing measured, and other gas molecules and collisions vary
are measured for the same sample preparation. Since a1 /a2 in frequency and severity as the pressure increases. For this
is a constant, the absorbance ratio is proportional to the reason the total pressure is often kept constant in a quanti-
concentration ratio. One can also write tative analysis of gases. The desired partial pressure of the
100 gas to be analyzed is introduced into an evacuated cell.
c1 = (20) Then the total pressure is increased to a standard value
1 + c2 /c1
such as 760 mm Hg by introducing nitrogen, for example,
which, when multiplied out simply states that the sum of before measuring the absorbance.
the two concentrations c1 and c2 is 100%. In this equa-
tion the ratio c2 /c1 is replaced by its equivalent from the
previous equation to give H. Multicomponent Analysis
spectrum, the total absorbance is equal to the sum of the This full-spectrum method improves the precision over
abc values for each component. One can specify that all those that use only a few wave numbers. Corrections can
the measurements will be done in the same cell, so the be added for Beer’s law deviations or fitting spectral base-
thickness is constant for the whole analysis. This means lines. However, all components present must be included
that b, the cell thickness, can be combined with a, the in the calibration mixtures.
absorbtivity, to give a new constant k, which replaces ab In a variation of this type of analysis, the concentra-
in Beer’s law. If there are three components, 1, 2, and 3, tions are expressed as functions of the various absorbances
absorbing at a specific wave number, then Beer’s law is rather that vice-versa as before. This is called the inverse
A = k 1 c1 + k 2 c2 + k 3 c3 . (24) least squares (ILS) (or the Pmatrix method). An advantage
of this method is that a quantitative analysis can be per-
In order to measure these three concentrations in an un- formed on some components using calibrated standards,
known, three absorbances are needed at three different even if some other components with unknown concentra-
wave numbers. tions are present in the standards in amounts bracketing
A1 = k11 c1 + k12 c2 + k13 c3 (25a) those in the samples. A disadvantage is that it is not a full-
spectrum method. In the analysis, there must be at least
A2 = k21 c1 + k22 c2 + k23 c3 (25b) as many standards for calibration as there are analytical
A3 = k31 c1 + k32 c2 + k33 c3 . (25c) wave numbers used.
Here, A1 , A2 and A3 absorbances at wave numbers 1, 2, Two factor analysis methods that are used are the prin-
and 3, where concentrations of components 1, 2, and 3 cipal components regression (PCR) and the partial least
are best measured. In k12 for example, the first subscript squares (PLS). In the PRC method, the concentrations are
is for wave number 1 and the second subscript is for the expressed as functions of the principal components (PC)
concentration of component 2. Three standards (std.) are instead of absorbances as in ILS. The PC are orthogonal
prepared with known concentrations with suitable ranges vectors that are linear combinations of the original spectral
for the analysis. These are all run at the first analytical data of the standards. Here, PC1 accounts for the maxi-
wave number to give three A1 equations. mum variability in the data, and PC2 accounts for the
maximum variability not accounted for by PC1, etc. The
(std. 1) A11 = k11 c11 + k12 c21 + k13 c31 (26a) other method PLS, is similar to the PCR method except
(std. 2) A12 = k11 c12 + k12 c22 + k13 c32 (26b) that the PCs are weighted. The weighting is based on the
correlation of the PCs with concentration. These are full-
(std. 3) A13 = k11 c13 + k12 c23 + k13 c33 . (26c)
spectrum methods like CLS, but like ILS, one can analyze
Here, the second subscript on A and c is for the standard one component at a time. These methods are most often
number. For these three equations, where the A values used for quantitative analysis in the near infrared region
and the nine concentrations (C11 etc.) in the standards because of the broadness and overlapping nature of the
are known, the three unknown k values can be evaluated. bands here.
This same procedure is used for the A2 and A3 equations
(25b and c) to evaluate all nine of the unknown k values.
VI. GROUP FREQUENCIES
Once these are known, the three equations (25a, b, and c)
can be used to calculate all the unknown concentrations
A. Concept of Group Frequencies
(c1 , c2 , and c3 ) in a sample from the measured ab-
sorbencies. This is called the method of simultaneous Bands at certain frequencies in the IR spectra have been
equations. While this method is straightforward, the dis- related to the presence of certain functional groups in
advantages are that Beer’s law nonlinearities are difficult the chemical structures. For example, in the spectra of
to handle and all the components in the mixture must be a series of unconjugated ketones, a band common to
accounted for. Also, only a limited number of analytical all is a strong band near 1715 cm−1 . This has been as-
wave numbers can be used. signed to the stretching of the carbony1 bond and is
For use in repetitive analyses, these quantitative meth- a group frequency for unconjugated ketones. There are
ods have been computerized. In the example discussed, many other bands in these ketone spectra, which differ
the absorbances are expressed as functions of the various from molecule to molecule, especially below 1300 cm−1 .
concentrations. A computerized version of this is called These are fingerprint-type bands that can be used to distin-
the classical least squares (CLS) (or the Kmatrix method), guish one ketone from another. A large body of empirical
as it gives the least squares prediction for the concentra- knowledge has been built up about the characteristic group
tions. This method can use many more than the minimum frequencies, which has proved to be very useful to the
number of analytical wave numbers, or even all of them. chemist. The vibrations that give rise to group frequencies
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
are those whose vibrational forms are nearly the same in or C S bond. This means that the rest of the molecule
a series of related molecules. does not affect the OH, C N, or SH vibrations mechani-
cally very much since the nearly stationary attached car-
B. Vibrational Interaction bon atom localizes the vibration. In this manner, one can
If all the bonds in a molecule vibrated separately, the di- see that such group vibrations as OH, C N, and SH are
atomic vibrational frequency formula given earlier could group frequencies; that is, their presence in a molecule
be used to predict the whole spectrum. The fact is that gives rise to absorption in a predictable frequency range.
bonds do not usually vibrate separately and much interac- While the OH vibration is mechanically unaffected by
tion occurs. However, interactions occur only if vibrations the rest of the molecule, the OH group has a force constant
have the same type of symmetry. For example, in a pla- that can be changed by hydrogen-bonding effects. An al-
nar molecule such as vinyl fluoride, planar vibrations do cohol in dilute CCI4 solution has a free OH band near
not interact with non-planar vibrations since these have 3640 cm−1 . Hydrogen bonding lowers the OH frequency
different symmetries with respect to the plane. and increases the bandwidth and intensity. A pure alcohol
One type of interaction occurs when two identical bonds is hydrogen bonded (OH · · · O) and absorbs broadly and
share a common atom. An example is the H2 O molecule, strongly near 3300 cm−1 .
which has two OH bonds with a common oxygen atom. In In a ketone, the high-frequency C O bond is connected
the H2 O molecule, one OH bond cannot be vibrationally to the rest of the molecule by two low-frequency C C
excited without also exciting the other identical OH bond bonds. In the carbonyl vibration the two attached car-
at the same time. The second bond vibrates either in phase bons hardly move, making the carbonyl a good group
or out of phase with the first. In one case, both bonds stretch frequency, mechanically nearly independent of the rest
at the same time and contract at the same time, and in the of the molecule. The carbonyl frequency can be varied by
other, one bond stretches while the other bond contracts. electron donation or withdrawal effects from the attached
In the H2 O gas-phase spectra, these are observed at 3652 groups, and these can shift the frequency. These effects
and 3756 cm−1 , respectively. Thus, the in-phase and out- are fairly well understood, which means that the shifts are
of-phase vibrations do not have the same frequencies. The predictable.
main reason for this is that, when the two vibrating bonds The CH2 , NH2 , and SO2 groups discussed earlier are
are not at the equilibrium length, both bonds exert restor- attached to the molecule by low-frequency C C, C N, or
ing forces on the common oxygen atom. The force resul- C S bonds, so the attached atoms hardly move, isolating
tant is different for the in-phase and out-of-phase stretch the CH2 , NH2 , and SO2 stretching vibrations. This means
vibrations, and this affects the frequencies. This is called that both the in-phase and out-of-phase stretching vibra-
vibrational interaction. tions for these groups are good group frequencies, nearly
There are several XY2 groups that have correlatable independent mechanically of the rest of the molecule.
stretching frequencies. For example, alkane CH2 groups In a group such as the C S group, the vibration is not
have stretching vibrations that absorb near 2930 and isolated like a C O vibration. The C S is attached to the
2850 cm−1 , amino NH2 groups have stretching vibrations rest of the molecule by C C or C N bonds, which have
that absorb near 3370 and 3300 cm−1 , and sulfone SO2 nearly the same frequency as the C S bond. As a result,
groups have stretching vibrations that absorb near 1300 interaction will take place and more than one vibration
and 1130 cm−1 . In each case the higher wave number in- will involve C S stretching.
volves the out-of-phase stretch and the lower wave number In such groups as CH3 and SO3 , there are three identical
involves the inphase stretch. bonds. These interact, so the group vibrates as a whole in
Consider the XYZ group where the X Y bond and the three different modes: an in-phase stretch and two different
Y Z bond have quite different frequencies when uncon- out-of-phase stretch vibrations. These are good group fre-
nected. Examples include the C O H group in alcohols, quencies because the group is connected to the molecule
the C C N group in nitriles, and the C S H group in with a low-frequency C C or C S bond. In a group such
mercaptans. Here the other groups on the carbon are ig- as a benzene ring there are six identical or nearly identical
nored. The XYZ group as a whole has two stretching fre- C C bonds, which interact to give six different stretching
quencies, but the interaction is different from that for the modes. Some of these are group frequencies.
XY2 case. In the high-frequency vibration of the XYZ
C. Survey of Group Frequencies
group, only the atoms of the high-frequency bond move
appreciably. In the examples given, this means that when Table IV contains some selected group frequencies used in
the high-frequency OH, C N, or SH bonds vibrate, the qualitative analysis. Some discussion of the data follows.
attached carbon hardly moves since it is attached to the In the region from 4000 to 2000 cm−1 , various XH groups
high-frequency bond with a low-frequency C O, C C, absorb. In the region from 3700 to 3100 cm−1 , OH groups
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
Alkanes
R CH2 R 2936–2916 s Out-of-phase str.
R CH2 R 2863–2843 m In-phase str.
R CH2 R 1475–1450 m CH2 def.
CH2 CH2 CH2 CH2 726–722 w In-phase rockd
R CH3 2972–2952 s Out-of-phase str.
R CH3 2882–2862 m In-phase str.
R CH3 1475–1450 m Out-of-phase def.
R CH3 1383–1377 m In-phase def.
C(CH3 )2 1389–1381 m Sym. in-phase def.
C(CH3 )2 1372–1368 m Antisym. in-phase def.
C(CH3 )3 1401–1393 m Sym. in-phase def.
C(CH3 )3 1374–1366 s Antisym. in-phase def.
Olefins
C C 1680–1630 mw C C str.
C CH2 3100–3075 w CH2 out-of-phase str.
R CH CH2 995–985 s Trans CH wage
R CH CH2 910–905 s CH2 wag
R2 C CH2 895–885 s CH2 wag
RCH CHR (trans) 980–965 s Trans CH wag
RCH CHR (cis) 730–650 m Cis CH wag
CH2 CH CO OR 970–960 m CH2 wag
CH2 CH O R 820–810 s CH2 wag
X Y and X Y Z
C C C H 2140–2100 w C C str.
C C C H 3340–3267 s C H str.
C C N 2260–2240 m C N str.
C N (conjugated) 2240–2220 v C N str.
S C N 2170–2135 m C N str.
C C CH2 2000–1900 s Out-of-phase str.
N C O 2275–2263 s Out-of-phase str.
N C S 2150–2050 s Out-of-phase str.
Aromatics
Aromatic CH 3100–3000 w CH str.
Aromatic ring 1620–1585 v Ring str.
Aromatic ring 1590–1565 v Ring str.
Aromatic ring 1525–1470 v Ring str.
Mono and meta 710–665 s Ring bend out of plane
Five adjacent ring H’s 800–730 s In-phase CH wag
Four adjacent ring H’s 805–735 s In-phase CH wag
Three adjacent ring H’s 825–750 s In-phase CH wag
Two adjacent ring H’s 860–795 s In-phase CH wag
Isolated ring H 935–810 s CH wag
Carbonyls
R CO R 1725–1705 s C O str.
Conjugated ketones 1700–1640 s C O str.
R CO H 1740–1720 s C O str.
Aryl CO H 1710–1685 s C O str.
R CO O R 1750–1735 s C O str.
continues
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
TABLE IV (continued )
Groupa Wave number region (cm−1 )b Assignmentc
TABLE IV (continued )
Groupa Wave number region (cm−1 )b Assignmentc
in alcohols and phenols and water have bands. Various olefins. In aromatics, these bands can usually distinguish
types of NH also absorb here, as well as acetylenic CH the number of adjacent aromatic hydrogens between sub-
groups. In the region from 3100 to 3000 cm−1 , absorption stituents: five for monosubstituted, four for ortho, three
occurs for aryl CH and olefinic CH. In the region from and one for meta, and two for para.
3000 to 2800 cm−1 , various types of CH3 and CH2 groups In the range from 1400 to 900 cm−1 vibrations from
absorb. Strongly hydrogen bonded hydrogens in acidic SO2 , SO3 , P O, and many types of C O groups have
compounds absorb broadly in the region from 3100 to strong bands. Usually below 1350 cm−1 only strong bands
2400 cm−1 . Finally, SH, BH, PH, and SiH groups absorb are interpreted since there are many bands in this region
in the region from 2600 to 2100 cm−1 . that are not group frequencies. This, of course, is only a
Various types of triple bonds such as nitriles and cumu- very small sample of the large body of group frequency
lated double bonds such as isocyanate groups absorb in the correlations available for qualitative analysis that can be
region from 2300 to 1900 cm−1 . Various types of double found in IR texts.
bonds, including C O, C N, and C C absorb from 1900
to 1550 cm−1 . Aromatic ring vibrations absorb near 1600
and 1500 cm−1 .
Hydrogen bending vibrations absorb from 1600 to SEE ALSO THE FOLLOWING ARTICLES
700 cm−1 . Bending vibrations for CH2 and CH3 absorb
from 1500 to 1350 cm−1 , and very useful out-of-plane CH INFRARED ASTRONOMY • INFRARED TECHNOLOGY •
wag vibrations in olefins and aromatics absorb from 1000 MACROMOLECULES, STRUCTURE • MICROWAVE MOLEC-
to 700 cm−1 . In olefins these can be used to distinguish ULAR SPECTROSCOPY • RADIOMETRY AND PHOTOMETRY
vinyls, vinylidines, and cis and trans 1,2-disubstituted • RAMAN SPECTROSCOPY
P1: GRB/MBQ P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN007C-340 July 10, 2001 14:45
BIBLIOGRAPHY McClure, ed., American Society for Testing and Materials, Philadel-
phia, PA.
Diem, M. (1993). “Introduction to Modern Vibrational Spectroscopy,”
Chalmers, J. M., and Dent, G. (1997). “Industrial Analysis with Wiley—Interscience, New York.
Vibrational Spectroscopy,” Royal Society of Chemistry, Analytical Griffiths, P. R., and de Haseth, J. A. (1986). “Fourier Transform Infrared
Spectroscopy Monographs, Cambridge CB4WF, UK. Spectrometry,” Wiley, New York, NY.
Coleman, P. A. ed. (1993). “Practical Sampling Techniques for Infrared Lin-Vien, D., Colthup, N. B., Fately, W. G., and Grasselli, J. G. (1991).
Analysis,” CRC Press, Boca Raton, FL. “The Handbook of Infrared and Raman Characteristic Frequencies of
Colthup, N. B., Daly, L. H., and Wiberley, S. E. (1990). “Introduction Organic Molecules,” Academic Press, New York, NY.
to Infrared and Raman Spectroscopy,” 3rd ed. Academic Press, New Messerschmidt, R. G., and Harthcock, M. A. (1988). “Infrared Mi-
York, NY. crospectroscopy,” Marcell Dekker Inc., New York, NY.
Crocombe, R. A., Olson, M. L., and Hill, S. L. (1987). “Computer- Pouchert, C. J. (1985). “The Aldrich Library of FT-IR Spectra,” Vols. 1
ized Quantitative Infrared Analysis,” pp. 95–130. ASTM STP 934. and 2, The Aldrich Chemical Company, Inc., Milwaukee, WI.
P1: ZCK/MBQ P2: GQT Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
Liquid Chromatography
Neil D. Danielson
Miami University
I. Analytical HPLC
II. Small- and Large-Scale HPLC
III. Separation Techniques
IV. Development of Separation Strategy
V. Sample Derivatization Chemistry
673
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
C. Injector
In liquid chromatography, introduction of the sample ide-
ally on the column or very close to it is important to
minimize sample diffusion and band broadening. Sample
sizes are small, ∼5–100 µl, with 20 µl being a common
size. Most HPLC injectors used today are loop valves as
shown in Fig. 3. A syringe is used to overfill a sample loop
while the mobile phase or eluent is bypassed directly to
the column (Fig. 3a). Upon switching the valve, the elu-
ent is diverted, displacing the sample from the loop and
onto the column (Fig. 3b). Since the valve is designed
to withstand 6000 psi, the mobile phase flow need not
be stopped. Convenience and good reproducibility are the
primary advantages. Error values less than 0.5% are possi-
FIGURE 2 A single-piston reciprocating pump. (a) schematic di- ble. The rotor seal is prone to scratching from particulates
agram (b) operation of a check valve. [From Yost, R. W., Ettre, and prefiltration or centrifugation of the sample should
L. S., and Conlon, R. D. (1980). “Practical Liquid Chromatog- be carried out if deemed necessary. Seal wear can gen-
raphy,” Perkin–Elmer Corporation, Norwalk, Connecticut, p. 146. erate small black particles which need to be taken out of
Reprinted with permission.]
the mobile phase before they reach the column using a
low dead volume on-line filter placed after the injection
solution output, (3) a flow rate range of 0.1 to 10 ml/min valve.
with 0.5% relative reproducibility, and (4) noncorrosive
wetted parts. Although a variety of pumps such as the D. Column Dimensions and Hardware
syringe and pneumatic types have been used, the recip-
rocating pump comes closest to fulfilling the above re- Most analytical HPLC columns are made of precision,
quirements at a modest cost. A schematic diagram of a smooth-bore stainless steel tubing. Uniformity of the in-
simple reciprocating pump is shown in Fig. 2a. Basically, side walls is important to minimize channeling effects
the operation involves a motor rotating a cam that pushes at the wall/packing interface. Dimensions for stainless
a sapphire piston back and forth in a small chamber. Ruby
ball check valves (Fig. 2b) permit solvent flow in only one
direction. The pumping rate is generally adjusted by con-
trolling the speed of the motor. The flow rate is often com-
puter controlled through a DAC and is generally set be-
tween 0.5 and 3 ml/min. Advantages of this pump include
good flow-rate reproducibility and compatibility with gra-
dient mobile phases. The primary disadvantage is the def-
inite pulselike nature of the delivered solution stream that
can cause detector baseline noise at sensitive settings. To FIGURE 3 A rotary sample valve. (a) Valve position for filling
alleviate this problem, dual-head and more recently triple- sample loop and (b) for introduction of sample into the column.
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
steel columns are generally 5–30 cm in length with a able to design HPLC instruments particularly suited for
2.3–6.2-mm inside diameter. The length of the column separations of biological samples.
is often determined by the particle size of the packing
since pressure drop is inversely proportional to the square E. Column Packing Supports
of the particle diameter (d p). The pressure drop (p) of
a packed bed can be more completely described as Column packing supports for HPLC can be divided into
two main classes, silica and polymer types. Silica (Fig. 5a)
p = ηL µφ/d p 2 , has several desirable properties such as excellent pressure
where η = mobile phase viscosity, L = column length, stability, high surface area, good control of size and shape,
µ = linear velocity, and φ = flow resistance factor (≈500). and ease of availability. Of the silica packings, pellicular
Small diameter columns can have better peak sensitivity and porous are the two major kinds. Pellicular packings,
(see Section I.E) but sacrifice sample capacity. developed first as an alternative to classical large particle
Stainless steel zero dead volume fittings such as a LC supports, basically consist of a solid glass bead on
1
in. → 161
in. union (Fig. 4a) are often required to make which a 1–2 µm porous layer of silica has been deposited.
16
connections. The 14 in. → 16 1
in. reducing union fittings Interaction of the sample components occurs only in this
(Fig. 4b) are used to connect the column to the injector and thin layer. Originally 40-µm size particles were made
detector. Both types of fittings can easily withstand 6000 but now smaller particles are available. Pellicular parti-
psi pressure. Pressed in the column fittings are 0.2-µm cles, fairly inexpensive in price, are sometimes used in
frits to contain the packing particles in the tube and prevent guard columns. Guard columns, usually 3–5 cm in length,
any particulates from reaching and disturbing the packing protect the analytical column from irreversibly adsorbed
bed. The connecting volume between the column-injector sample constituents often found in complex biological ma-
and column-detector must be kept at a minimum because trices. In addition, a silica guard column can protect a silica
bandbroadening (variance or σ 2 ) is directly proportional analytical column from dissolution due to a high pH mo-
to the fourth power of the tubing radius but only propor- bile phase by saturating the mobile phase with dissolved
tional to the column length, and inversely proportional to silica. Porous silica particles that are 10, 5, or 3 µm in
the flowrate. Therefore, 16 1
in. stainless steel tubing with size are more commonly packed in analytical columns.
a 0.01 in. or smaller inside diameter is used for all fitting Recently, columns packed with 2-µm particles have been
connections. Recently, metal free column hardware and reported in the literature. The 3–10-µm size particles can
polyethyletherketone (PEEK) tubing have become avail- be commercially obtained either irregular or spherical in
shape with a surface area ranging from 50 to 500 m2 /g.
FIGURE 6 Effects of particle size on column performance: (a) 10 µm, (b) 5 µm, and (c) 3 µm. Ultrasphere C-18
columns: (a) 30 cm × 4.6 mm, (b) 15 × 4.6 mm, and (c) 7.5 × 4.6 mm. Mobile phase: 60–40 methanol–water. Flow
rate = 1 ml/min. Temperature = 30◦ C. Pressure: (a) 810 psi, (b) 1600 psi, and (c) 2250 psi. Peaks: (1) Phenol, (2)
Benzaldehyde, (3) Acetophenone, (4) Nitrobenzene, (5) Methylbenzoate, (6) Anisole, (7) Benzene, and (8) Toluene.
[Reprinted with permission from Beckman/Altex Scientific.]
The pore size of the silica particles must be large enough to be effective as column packings for the reversed-phase
to permit easy entrance and exit of the sample molecules. separation of proteins and are commercially available.
However, since pore size is inversely proportional to the As can be seen in Fig. 6, the advantages of columns
surface area of the packing, the pores should not be exces- packed with smaller particles are faster analysis times,
sively larger than the sample components of interest. For improved solute sensitivity, and decreased solvent con-
relatively small organic or inorganic molecules, a pore size sumption. It can be shown that the peak height maximum
of 60 Å is sufficient. For the separation of large molecules (Cmax ) can be calculated from the following equation,
such as polymers or proteins, a pore size of at least 250 Å Cmax = (Cs Vs /Vr )(N /2π )0.5 , where Cs and Vs are the con-
is preferred. centration and volume of sample injected respectively, Vr
Packing materials for the HPLC separation of is the retention volume, and N = number of theoretical
biomolecules that have not only 500–1500 Å pores but also plates. The number of theoretical plates can be calculated
a network of 6000–8000 Å transecting tunnels have been easily from the chromatographic data as N = 16(tr /wb )2 ,
developed by Regnier. These highly porous materials, where tr = peak retention time and wb = width of the peak
when packed into columns, permit mobile phase veloci- at baseline in time units. Because Vr is proportional to the
ties 2–5 times higher than those for conventional wide pore volume of the column (πr 2 × L, where r = column ra-
silica columns. For the rapid reversed-phase separation of dius and L = column length), Cmax will increase propor-
peptides and proteins, both small 2-µm porous wide pore tionally to the ratio of the square of the radii as column ra-
(200 Å) silica and 2-µm pellicular silica microspheres dius decreases. Going from a column of 4.6 mm to one of
packed in 3-cm columns have been developed. For a five 2.1 mm and assuming all other variables are constant, the
component mixture of proteins, separation times of less improvement in Cmax can be predicted to be almost 5 times.
than a minute were possible. Nonporous monodisperse An efficiency comparison of different octyldecyl (C-18)
1.5-µm silica beads developed by Unger have been shown modified silica columns using plate count (N ) is shown
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
TABLE I Quantitative Column Comparison at Optimum tion of acidic compounds such as anti-inflammatory drugs
Velocitya (aspirin, acetoaminophen, and ibuprofen) or basic drugs
Column type N/column N/m N/sec N/psi such as antihistamines. Zirconia columns provide higher
plate counts than polymeric columns and comparable ef-
Conventional 21800 87200 36.8 12.5 ficiencies to silica based columns. Titania has also been
5-µm C-18 silica
(250 × 4.6 mm)
investigated as a stable HPLC column packing but is less
Microbore 19100 38200 9.05 63.7
well studied.
10-µm C-18 silica Polymeric packings for HPLC have been developed
(500 × 1.0 mm) largely to overcome the mobile-phase pH constraints of
Small particle 11900 119000 99.2 3.81 silica. In general, for routine use of silica-based pack-
3-µm C-18 silica ings, the pH of the mobile phase must be maintained
(100 × 4.6 mm) between pH 2 and 7.5. This is because the silica itself
a From McCoy, R. W., and Pauls, R. E. (1982). J. Liq. Chromatogr. can dissolve at alkaline pH, and cleavage of the silox-
5, 1869. Reprinted with permission from Marcel Dekker, Inc. ane bonds holding the functional groups occurs at acid
pH. Another advantage of polymeric packings is that
few residual polar groups, such as nonreacted hydrox-
in Table I. As predicted by the Van Deemter equation, the yls found in silica, are present that can cause peak tail-
number of plates per meter increases as particle diameter ing through hydrogen bonding with certain solutes. Most
decreases. The Van Deemter equation can be expressed as polymer HPLC packings are polystyrene-divinylbenzene
(PS–DVB) resins cross-linked at 10% or greater to en-
HETP = A + Bu + Cm u + Cs u ,
sure sufficient particle rigidity (Fig. 5). Even so, the col-
where HETP = height equivalent of a theoretical plate, umn pressure limit for most polymers is limited to about
u = mobile phase velocity, A = eddy diffusion term 3000–5000 psi. Most PS–DVB HPLC packings are spher-
dependent on the particle diameter of the packing ical in shape and 5 or 10 µm in size. The pore size and
(dp ), B = longitudinal diffusion term dependent on surface area of PS–DVB particles can also be controlled
the diffusion constant of the analyte in the mobile similar to the range of values given for silica. However,
phase (Dm ), Cm = the mass transfer term involving the generally polymeric packings are less efficient (20,000–
solute from the bulk mobile phase to the surface of the 50,000 plates/m) than silica packings. Because of the
stationary phase and is dependent on dp2 /Dm , and Cs aromatic (moderately nonpolar) nature of PS–DVB, it is
is the mass transfer of the solute into and out of the used primarily for reversed-phase HPLC unless chemi-
stationary phase and is dependent on the square of the cally modified. Functionalization of PS–DVB resins for
film thickness (df ) of the stationary phase. A good rule of use as ion-exchange packings is the other dominant use of
thumb is N ≈ 3000 L /d p, where L = column length (in these supports. Other polymers such as polyvinylpyridine,
centimeters) and d p = particle diameter (in micrometers). polyacrylamide, polyvinylalcohol, and various floropoly-
The small particle column was lowest in plates generated mers such as derivatized polychlorotrifluoroethylene have
per unit pressure, indicating high-column back pressure also been used as reversed-phase packings. Further de-
could be a problem with some mobile phases. tails of the importance of polymeric supports to ion-
Silica can be used as is for normal phase HPLC in exchange chromatography in particular are provided in
which the mobile phase is nonpolar in nature, such as hex- Section III.
ane/chloroform and the column-packing surface is polar.
For reversed-phase HPLC in which the mobile phase is
F. Column Packing Technique
polar, such as water or methanol, and the column-packing
surface is nonpolar, the silica must be chemically altered. Both the dry-fill and wet-fill packing methods have been
Silanization reactions are carried out to covalently attach used to prepare HPLC columns. The dry-fill approach
long chain hydrocarbon groups to the silica surface. Fur- involves vertical tapping of the column with simultane-
ther details of the importance of silica supports are pro- ous rapping along its side to cause good consolidation
vided in Section III, “Separation Techniques.” of the column bed. This procedure is recommended for
Porous zirconia particles coated with polybutadiene the packing of rigid particles with a diameter greater than
make a reversed-phase HPLC column packing that offers 20 µm, such as pellicular spheres. Small porous particles
both excellent pressure stability and chemical inertness have high surface energies with respect to their mass and
throughout the entire pH range. Because of the complex clump together when attempted to be packed dry. Prepara-
surface chemistry of zirconia, mobile phase additives such tive HPLC columns are often made using this “tap-filled”
as phosphate or fluoride are added to facilitate the separa- method.
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
The wet-fill or slurry packing techniques are recom- be well shaped with no shoulders or split tops. This is
mended for particles less than 20 µm in diameter. In a sign of channeling and the column must be repacked.
this approach, the slurry is pumped quickly under high Excessive peak width is usually a sign of a void volume
pressure into the column blank. The solvent chosen to and again the column should be repacked.
prepare the slurry must wet the particles and keep them
well dispersed. In addition, it is desirable to choose a sol-
G. HPLC Detectors
vent with a density that approaches the particle density,
the so-called balanced density situation. Although sus- A wide variety of HPLC detectors have been developed to
pected cancer causing agents such as tetrabromoethane try to fill both the high-sensitivity and universal-detection
or chloroform can be used, this method is recommended requirements. Some of these are adaptations of well-
primarily for 10 µm particles. Particles 5 µm or less set- known GC detectors such as flame ionization and electron
tle very slowly, and other solvents should be considered. capture. This discussion will focus primarily on the design
Methanol or isopropyl alcohol are commonly considered and appropriate applications of the seven most common
for silica packings or polymers. For small reversed-phase and commercially available HPLC detectors. A listing of
silica particles, acetone-hexane mixtures, because of their these detectors is shown in Table II. The HPLC detectors
low viscosity, have been employed. Because of the possi- can generally be classified as either responsive to a change
bility of charging nonpolar reversed-phase packings due in the property of the mobile phase when a solute (sample
to the rapid liquid flow from the packing pump, 80% component) is present or to a property of the actual solute
methanol–20% sodium acetate (0.1%) in water has been itself.
recommended. An example of the latter is the most widely used HPLC
A high-pressure pneumatic pump capable of being pres- detector, the UV–VIS spectrophotometer (Fig. 7). Its pop-
surized to 8,000 to 10,000 psi by closing a shutoff valve ularity is due to a wide range of applicability, excellent
is generally necessary to pack efficient columns. Upon stability, and low cost. This detector is designed to be
opening the valve, the released pressure forces the pack- equipped with a low-volume (8 µl or less) flow cell usually
ing slurry from a stainless steel reservoir into the column 1 cm in pathlength. The low volume of the cell is impor-
blank in a few seconds. This “slamming” process gener- tant to minimize bandbroadening, as is the radius of the
ally will permit a uniform packed bed of particles in the connecting tubing to the column. Lenses are used to focus
column. The column bed is considered stable if the liquid as much light as possible through the 1- or 2-mm diameter
flow from the column is constant after several repressur- cell. Most filter HPLC spectrophotometers are equipped
ization cycles. After no liquid is observed to drip from the with a mercury lamp. This source emits 90% of its radia-
column, it is carefully removed from the packing reser- tion at 254 nm, which is an excellent wavelength for the
voir. The column ends are cleaned off with a razor blade detection of aromatic compounds. Different filters and/or
to permit the fittings to be attached. The column should special phosphor converters placed after the lamp allow
be checked with a test mixture and mobile phase chosen other wavelengths to be selected. The zinc lamp is used for
to give peak retention times from 5 to 15 min. A column shorter wavelengths such as 214 nm. Usually the light is
plate height two to three times the particle diameter should focused through dual flow cells, one considered the sample
be obtainable if the column is well packed. Peaks should side and the other the reference side. If the mobile phase
TABLE II Characteristics of Liquid Chromatography Detectorsa
Temperature Linear Noise Maximumd
Detector basis Typeb sensitive? range levelc sensitivity
2nd ed.” Wiley-Interscience, New York, p. 162. Reprinted by permission of John Wiley & Sons, Inc.
b G = General; S = selective.
c AU = Absorbance units; RIU = refractive index units.
d Sensitivity for a favorable sample in g/ml.
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
FIGURE 7 Diagram of a photodiode array instrument. For HPLC, the cuvette is replaced by a flow cell similar to that
used for the standard UV–VIS detector. [From Siouffi, A-M., Chapter 1, in “Food Analysis by HPLC.” (L. M. L. Nollet,
ed.), Marcel Dekker, New York.
does not absorb, which is often the case, the reference side which is a function of the solute concentration. Although
is usually air. However, occasions arise where it is advan- the RI detector is versatile, it is not particularly sensitive
tageous to pump the mobile phase through both cells for (microgram level) and is very prone to temperature fluc-
subtraction purposes. The detector is quite stable to tem- tuations. Careful temperature control of the detector cell
perature fluctuations; however, air bubbles that lodge or with a water jacket is crucial for maintaining a stable base-
pass through the flow cell will cause baseline spikes. A line. Modifications in the electronic design and the use of
UV–VIS HPLC detector equipped with a monochromator a laser source have also improved the detectability of the
for specific wavelength selectivity is more versatile but RI detector.
also more expensive. Deuterium and tungsten lamps are Luminescence detectors can be either based on fluo-
used to provide ultraviolet and visible light, respectively. rescence or chemiluminescence. The fluorescence HPLC
Often multiple wavelengths can be monitored simultane- detector is basically a fluorometer equipped with a flow
ously. The photodiode array (PDA) UV–VIS detector has cell slightly larger in volume than the UV–VIS detec-
now become quite common as the standard HPLC de- tor to permit more fluorescent light from the solute to
tector ordered with a new instrument. The PDA detector reach the photomultiplier. Again, the primary modifica-
employs a reverse optics design which allows the entire tion is the use of lenses to focus the excitation source light
spectrum of light to be dispersed onto a diode array (see onto the flow cell. Filter instruments are often more sensi-
Fig. 7). Each diode of the array is responsible for detect- tive than monochromator instruments because of greater
ing light of a narrow wavelength range depending on the light throughput. Again, lasers have allowed the use of
desired spectral range to be covered and the speed of data small volume flow cells with lower detection limits. The
acquisition. This instrument can take a spectrum over a fluorescence detector, although useful for only a select
preselected wavelength range in a fraction of a second as class of aromatic hydrocarbons or derivatives, is about
the solute elutes from the column. Therefore, qualitative 100 times more sensitive than a UV–VIS detector. For
(peak purity) as well as quantitative information about the example, polyaromatic hydrocarbons can be determined
unknown sample component can be obtained. at the ppb level. Signal stability with respect to temper-
The differential refractometer is perhaps the second ature is good. The coupling of fluorescence to HPLC
most widely used HPLC detector because of its universal has generally minimized background signals from im-
nature. Potentially, any substance with a refractive index purities and oxygen quenching effects. If the lamp of
(RI) different from the mobile phase is detectable. The the fluorescence detector is turned off, it can operate as
deflection type RI detector consists of a light source pass- an effective chemiluminescence detector. Postcolumn ad-
ing two beams of monochromatic light through a double dition of reagents such as luminol and a metal catalyst
prism that constitutes the sample cell and reference cell. for the detection of an oxidizing agent such as H2 O2 is
If the mobile phase composition changes, the altered re- necessary.
fractive index causes the beam to be deflected from its The electrochemical (EC) HPLC detector is basically
initial position on the photomultiplier detector. The elec- a small electrode composed usually of glassy carbon
trical signal produced is proportional to the light position, mounted in a flow cell (Fig. 8a). The auxiliary electrode is
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
reliability but also the lower cost. The most common com-
mercial LC/MS instrument is based on the electrospray
interface (Fig. 10). The mobile phase containing the sep-
arated sample components is converted to an aerosol of
electrically charged droplets by means of strong shear
forces of the nebulizing gas and the strong electrostatic
field (2–6 kV) in the spray chamber. Ions of one polar-
ity are attracted to the droplet surface by the electrostatic
field causing the droplets to disperse in a fine spray or
electrospray. For example, if the solution contains acetic
acid and the nebulizer needle is held at a positive potential,
positively charged sample molecular ions (MH+ ) will be
formed. Because the sample solution is not heated when
the aerosol is created, thermal decomposition of the ana-
lytes is not a problem. Before the ions can be mass ana-
lyzed, the solvent must be removed to yield the base ion. A
counterflow of heated nitrogen gas causes the desolvation
process forcing the like-charged surface ions closer to-
gether until a Coulombic explosion produces tiny droplets.
Continued evaporation of the remaining solvent is rapid
FIGURE 9 Diagram of an evaporative light-scattering detector until base analyte ions are produced that can then enter the
[Reprinted by permission from Alltech Associates.] metallic opositely charged end of the glass capillary lead-
ing to the mass spectrometer. The nebulizer and capillary
temperature control of the evaporation tube. Minimizing are set at right angles to each other to minimize mobile
particulate impurities in the mobile phase is also impor- phase or sample matrix contamination of the mass spec-
tant. Common reversed-phase organic solvents and water trometer. Because the glass capillary separates the atmo-
up to 25% even with volatile buffers can be effectively spheric pressure region of the source from the vacuum re-
removed. Typical applications include lipids, sugars, and gion of the mass spectrometer, the analyte ions are pushed
triglycerides. by a pressure gradient through the capillary as a supersonic
Mass spectrometry (MS) has now become a user- jet flow. As the ions exit, they are electrostatically repelled
friendly detector for LC not only because of the increased to the first skimmer by the like-charged metallic end of
FIGURE 10 Diagram of the Bruker orthogonal electrospray LC/MS interface. [Reprinted by permission from the
Bruker-HP Esquire LC Operations Manual, Version 3.1.]
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
FIGURE 14 Separation of a model mixture containing 5 parabenes, 6 PAHs, and thiourea as the unretained compo-
nent. Conditions: 250 mm to the detector (335 mm total length) × 0.1 mm CEC Hypersil C-18, 2.5 µm, acetonitrile-25
mM MES, pH = 6 (80 : 20), 20 kV, 10-bar pressure applied to both ends of capillary. Plate numbers 60,000–77,000.
[From Dittman, M. M., and Rozing, G. P. J. (1996). Chromatogr. A 744, 63–74.]
C. Preparative HPLC (3) ion exchange, and (4) size exclusion. For the first three,
the liquid mobile phase has a major role in governing so-
Preparative HPLC use 25-cm columns with large diame-
lute retention. The fundamental principles and important
ters of 25–150 mm while semipreparative HPLC columns
packings for each of these LC modes will be described as
of the same length but 10–30 mm in diameter are used to
well as providing typical application chromatograms. In
separate 0.05–1 g samples. Generally the column packing
addition, chiral separations will be discussed in a separate
is the more inexpensive 40–50-µm silica and flow rates
section.
anywhere from 20 to 300 ml/min are considered reason-
able. Because gram quantities of the sample are injected,
the detector need not be very sensitive and therefore the RI A. Adsorption LC
detector is commonly employed. The goal of preparative
HPLC, often in support of organic synthesis research, is to Adsorption LC or liquid–solid chromatography (LSC) is
separate as much sample as possible into its components. principally carried out in the normal phase mode. It in-
Therefore, peak resolution is often sacrificed to permit volves no partitioning of the sample solute in the station-
larger injected samples, even to the point of column over- ary phase. Instead, the polar groups of each organic solute
load. The chromatography triangle of speed, resolution, interact through primarily hydrogen bonding forces with
and sample capacity indicates that if capacity is of great the polar sites of the stationary phase. Therefore, careful
interest, speed will also likely be sacrificed. As long as adjustment of the polarity of the mobile phase for sta-
the peaks can be distinguished for fraction collection, the ble activity of the polar sites is needed for reproducible
separation is considered adequate. Often analytical HPLC, separation.
NMR, and IR are used to confirm the purity of the col- The most common packing materials for LSC are
lected fractions. An example chromatogram comparing porous silica (SiO2 )x or alumina (Al2 O3 )x . Both of these
analytical and preparative HPLC is shown in Fig. 15. materials have numerous surface hydroxyls that act as
the adsorption sites. Silica particles have a high surface
area averaging 400 m2 /g and are quite acidic in nature with
III. SEPARATION TECHNIQUES a surface pH of about 5. This pH is usually not so low as to
cause acid-catalyzed solute degradation reactions during
The different types of liquid chromatography can be clas- the chromatography. As expected, retention of organic
sified into four main classes based on the solute-stationary bases such as various anilines or nitrogen heterocycles is
phase interaction. These are (1) adsorption, (2) partition, particularly good on silica. Alumina on the other hand, is
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
FIGURE 15 Analytical (a) and preparative (b) isolation of vitamin B-12 intermediates. (a) Column, 180 × 0.2-cm i.d.,
Corasil II, 37–50 µm; mobile phase, hexane/isopropanol/methanol. (b) Column, 240 × 2.3 cm i.d. 37–80 µm silica:
mobile phase, hexane/isopropanol/methanol (5 : 2 : 1), flow rate 34 ml/min; injected sample, 5 g. [From Snyder, L. R.,
and Kirkland, J. J. (1979). “Introduction to Modern Liquid Chromatography, 2nd ed.” Wiley, New York, p. 655. Reprinted
with permission.]
quite basic with a pH of about 12 and generally is lower in Z ads solute molecules adsorbed on the packing, and Sm
surface area with larger pores. Base catalyzed degradation solvent molecules in the mobile phase.
reactions on alumina can be a problem. However, good A quantitative log relationship between retention factor
retention of acidic organic compounds such as phenols k
and mobile phase strength NB follows.
and carboxylic acids is possible on alumina. Florisil, a
magnesia–silica coprecipitate, which is strongly acidic log k
= k
B − (Ax /n B ) log NB ,
in nature, has also been used for LSC. However, silica
accounts for about 80% of all applications. Free (non- where kB
= retention factor in a pure nonpolar solvent,
hydrogen bonded) hydroxyl groups are more reactive to Ax = adsorption cross section of analyte X, n B = adsorp-
solute polar groups and provide most of the retention. The tion cross section of solvent molecule, and NB = number
siloxane, Si–O–Si, bonds are very weak in their adsorption of polar solvent molecules. If the slope representing the
properties. The presence of a polar solvent in the mobile number of analyte molecules/number of solvent molecules
phase such as water will promote hydrogen bonding, de- displaced is large, B is either a very polar solvent and/or
creasing the number of active sites available for solute the analyte is weakly retained. The converse is true if the
retention. The LSC retention mechanism can be summa- slope is small.
rized as a competition between the solute molecules (Z ) This adsorption–desorption equilibrium is in opera-
and the solvent molecules (S) for the adsorption sites. tion continuously as the solute molecules pass down the
column. The more polar the mobile phase, the more
Z m + nSads
Z ads + nSm ,
adsorption sites will be blocked by the solvent, causing
where Z m represents solute molecules in the mobile the solute molecules to remain in the mobile phase and
phase, Sads solvent molecules adsorbed on the packing, decreasing retention.
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
TABLE III Properties of Solvents for HPLCa The usual order or elution of organic solutes is depen-
Viscosity at 25◦ C dent on the type of polar functional groups, number of
Solvent εob UV cutof (nm) [centipoise (cP)] groups, and orientation. A listing of functional groups
from low k
(capacity factor which is proportional to re-
Pentane 0.0 195 0.22 tention) to high k
follows: Alkane < olefins < aromatic
Isooctane 0.01 197 0.47 ≈ organic halides < sulfides < ethers < nitro compounds
Cyclohexane 0.04 200 0.90 < esters ≈ aldehydes ≈ ketones < alcohols ≈ amines
Carbon tetrachloride 0.18 265 0.90 < sulfones < sulfoxides < amides < carboxylic acids.
p-Xylene 0.26 290 0.60 As expected, this order roughly reflects the eluotropic se-
Toluene 0.29 285 0.55 ries in Table I. A greater number of polar groups will
Benzene 0.32 280 0.60 promote retention unless their close proximity permits in-
Ethyl ether 0.38 218 0.24 tramolecular hydrogen bonding.
Chloroform 0.40 245 0.53 One of the strengths of LSC is its ability to separate
Methylene chloride 0.42 233 0.41 isomers, particularly aromatics functionalized with polar
Tetrahydrofuran 0.45 212 0.46 groups, in the retention order ortho < meta < para. The or-
Acetone 0.56 330 0.30 tho compound is retained the least due to intramolecular
Ethyl acetate 0.58 256 0.43 hydrogen bonding. The meta functional groups can inde-
Aniline 0.62 310 3.8 pendently interact with the stationary adsorption sites but
Acetonitrile 0.65 190 0.34 not often at the same time. The para isomer is retained
Dimethylsulfoxide 0.75 268 2.0 longest because the two opposite functional groups can
Isopropanol 0.82 205 1.9 “sit down” on the adsorption surface and both simultane-
Ethanol 0.88 210 1.1 ously interact with the active sites. A chromatogram of
Methanol 0.95 205 0.54 nitroaniline isomers is shown in Fig. 16. As required for
Water large 191 0.90 all types of liquid chromatography, the sample must be
a Most of this data was taken with permission from a similar table in soluble in the mobile phase. Therefore, LSC is generally
Snyder, L. R., and Kirkland, J. J. (1979). “Introduction to Modern Liquid used for organic solvent extracts of solid or aqueous sam-
Chromatography,” Wiley, New York, p. 248. ples as well as characterization of product solutions from
b Eluotropic series for alumina (similar rank for silica).
organic synthesis.
O CH3 O CH3
dry
toluene
Si H ClSi R Si O Si R HCl,
N2
O CH3 O CH3
O O R
H2O
Si OH R2SiCl2 Si (Si O)nH
O O R
FIGURE 20 Three-dimensional presentation of the geometries of cyclodextrin. [From Braithwaite, A., and Smith, F.
J. (1996). “Chromatographic Methods, 5th Ed.” Chapman & Hall, London.]
log k
= −(y/x) log[E x ] + (log B)/x, fore, buffered solutions are almost always the major com-
ponent of a mobile phase for ion-exchange LC. For weak
where y = charge of the solute, x = charge of the eluent, acidic or basic solutes, the mobile phase pH controls their
and B = the product of the capacity of the packing and ionized state and ability to interact with the resin. The
the equilibrium constant for the ion-exchange process. capacity of weak ion-exchange resins is in addition in-
Although silica has been used, the most common ion- fluenced by pH. All other factors considered equal, the
exchange supports are PS–DVB resins because of their greater the capacity of the resin, the greater the ion reten-
stability at pH extremes. The non-cross-linked benzene tion. Finally, the pH as well as the buffer salt can contribute
rings are available for functionalization. Sulfonation of significantly to the overall ionic strength of the mobile
PS–DVB resin yields the strong cation exchanger, Res– phase. Ionic strength is calculated by taking one-half of the
SO− +
3 X , while chloromethylation and subsequent ami- sum of the ion concentration times their charges squared.
nation forms the strong anion exchanger, Res–CH2 –N+ As the ionic strength increases, the amount of counter ion
(CH3 )3 X− . The capacity of these resins, the number of in the mobile phase increases, driving the equilibrium back
exchangeable groups per gram of resin, can range from to the left. This competition of the counter ions for the sta-
0.1 to 2 meq/g, depending on reaction conditions. Surface tionary ionic sites results in a reduced retention of the so-
agglomerization is a convenient method to prepare low lute ions. The lower the resin capacity, the smaller the ionic
capacity ion exchange packings for ion chromatography. strength that is required to elute the solute ions from the
For example, sulfonated PS–DVB microspheres (5–40 column. The ionic strength is often intentionally increased
µm) are contacted with colloidal anion exchange particles gradually to improve the separation of weakly and strongly
(100–1000 Å) to electrostatically form a surface agglom- retained ions in a mixture (see Section IV, Fig. 26).
erated anion exchange resin. The ion-exchange capacity The nature of the ionic solutes often affects their ion
of the resin can be controlled by changing either the size of exchange retention. As expected, polyvalent ions are held
the microspheres or of the colloidal particles, as well as the more tightly than monovalent ions. Within a given charge
degree of functionalization of the latter particles. The cor- group, retention generally increases with the size of the
responding weak anion exchanger Res–NH+ (CH3 )2 X− ion but decreases with the size of the hydrated radius. Sol-
and cation exchanger Res–COO− X+ have also been de- vated ionic radii limit coulometric interactions between
veloped for use in the separation of labile molecules such ions and energy must be put into the system to strip the
as proteins. water away. The retention order for the alkali metals is
The mobile-phase factors of pH and ionic strength pri- Cs+ > Rb+ > K+ > NH+ + + +
4 > Na > H > Li . Because of
marily control the retention of ion exchange resins. There- its greater hydration, Li is retained less than H+ .
+
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
The mobile phase viscosity should be less than 1.0 cP volves the gradual increase of the mobile-phase solvent
(see Table III) to minimize high column back-pressures. strength with time to increase the speed of peak elu-
Optimization of the sample separation with respect to res- tion. Usually the percent composition of a component
olution, speed, and capacity should be the goal. However, of the mobile phase is changed. However, addition of a
compromise of each of these factors to some degree is new stronger solvent can also be done. Gradient elution
necessary; if not, undue cost or labor is expended. Finally, should be used instead of an isocratic mobile phase if
the conditions of an HPLC method can often be decided Tg /Tg ≥ 0.25, where Tg represents the difference in
simply by checking the scientific literature for similar pre- retention times between the first and last eluting peaks,
vious studies. and Tg is gradient time. The optimum gradient time can
be predicted from the equation Tg = 20 VM φ/F, where
VM = column dead volume, φ = change in volume
B. Programming an LC Variable
fraction of the strong solvent during the gradient, and
If the sample components have a widely varying degree F = flow rate. An average capacity factor of 5 and the
of retention, separation times can be excessively long for isocratic parameter for small molecules equal to 4 must
reasonable peak resolution. To improve the separation, be assumed. The shape of the gradient program can be lin-
programming of some variable during the course of the ear, concave, convex, or stepwise. In choosing a solvent
separation is necessary. The most common parameters are pair, the only requirements are the solvents be nonreac-
flow rate, solvent strength, temperature, and column type. tive and miscible. For adsorption LC, methylene chloride
Flow programming involves initially using a low flow is often added to hexane. For reversed-phase LC, ace-
rate to better resolve the early peaks and then increas- tonitrile is added to water. For ion-exchange LC, a buffer
ing the flow rate to elute well-retained components. It or salt solution is added. A gradient program can be op-
can be carried out step-wise or continuously if desired. timized by following these steps: (1) Run a linear pro-
Basically, improvement of front-end resolution at the ex- gram at about 2%-min of the strong solvent B to deter-
pense of back-end resolution is the result. Decreasing the mine if and where all the components elute, (2) choose
flow rate by 75% will yield a resolution improvement of the initial % solvent B composition to be half of the
about a factor of 1.4. The primary advantages of flow pro- %B where the first peak elutes, (3) optimize the pro-
gramming are its ease of implementation and no need for gram by starting at the front of the chromatogram, in-
column reequilibration between samples. serting sharper %B rates at the appropriate times, and (4)
Solvent strength programming or gradient elution is the fine-tune the program adding short %B holds to main-
most effective and popular programming method. It in- tain good peak resolution. Computer-assisted methods
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
such as DRYLAB are commercially available. An ex- matic compounds in a few minutes using a short polybu-
ample of how gradient elution programming can im- tadiene (PBD) modified zirconia column with only super-
prove chromatographic time and resolution is shown in heated water as the eluent (see Fig. 27). For polymeric
Fig. 26. packings, subtle temperature changes in the 35 to 40◦ C
Temperature programming has been used to a limited range can reduce peak retention by several minutes. In
extent to control peak retention times. As in GC, an in- fact, polymeric columns are often isothermally tempera-
crease in temperature will reduce retention time. This re- ture controlled to improve peak retention reproducibility.
sults in sharper peaks with better sensitivity and often an Finally, stationary phase programming of column
improvement in column efficiency. Because the mobile switching can be used to generate either longer or shorter
phase viscosity is reduced, column backpressure drops. retention of certain sample components as desired.
For silica-based columns, dramatic increases in temper- Schematic diagrams of these common switching modes
ature from 25 to 70◦ C have been used to separate pep- are shown in Fig. 28. A series arrangement of columns
tides and polyaromatic hydrocarbons. High temperatures (Fig. 28a). permits good separation of early and late elut-
of 150–200◦ C have been used for the separation of aro- ing peaks in a reasonable time. The early eluting compo-
nents are allowed to pass through both columns to the de-
tector while the late eluting components pass only through
the first column. The option of eluting the mobile phase to
waste (Fig. 28b) is also possible if undesired sample ma-
trix components need to be stripped off rapidly. A parallel
arrangement of columns (Fig. 28c) permits true stationary
phase programming in which the sample, depending on
type and composition, can be directed to column 2 or 3 for
complete separation. Finally, the configuration in Fig. 28d
permits recycle chromatography or the process of passing
the samples through the column repeatedly to improve res-
olution. This technique is particularly useful in preparative
FIGURE 28 Different experimental configurations for column switching. [From Snyder, L. R., and Kirkland, J. J.
(1979). “Introduction to Modern Liquid Chromatography, 2nd ed.” Wiley, New York, p. 697. Reprinted with permission.]
chromatography. Alternatively, this setup could be used means. Precolumn or postcolumn derivatization chem-
for column backflushing, which involves running the mo- istry is often an inexpensive and effective procedure to
bile phase backwards through the column. Sample com- remedy this situation. Precolumn derivatization chem-
ponents retained strongly at the front of the column can be istry is carried out prior to and usually separate from the
eluted easily to the detector. This same procedure can be chromatographic operation. The following conditions are
used to clean the front end of the column of irreversibly desirable for precolumn derivatization.
retained compounds.
1. Reaction stoichiometry and product structure are
V. SAMPLE DERIVATIZATION CHEMISTRY known.
2. The reaction should be reasonably fast and the
The previously described separation strategies are all derivatives stable in solution.
directed to improving analysis time and peak resolution, 3. The derivatives and excess reagent must be readily
assuming the separated components can be detected. separable and stable during the chromatographic step.
However, many compounds of interest, such as aliphatic
carboxylic acids or amino acids, are difficult to detect The primary advantages of precolumn derivatization is
by optical absorbance, fluorescence, or electrochemical that a wide variety of reagents are available for use since
P1: ZCK/MBQ P2: GQT Final Pages
Encyclopedia of Physical Science and Technology EN008B-382 June 30, 2001 18:58
959
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
Nuclear magnetism Magnetic properties acquired by a commercially until about 1982, it is estimated that by 2000
substance when there is a net alignment of the nuclear there were more than 10,000 magnetic resonance (MR)
spins. scanners in use worldwide and more than 20 million MR
Pixel Individual picture element. The numerical value of scans were being done each year. On the order of 150 mil-
the pixel brightness is used to represent it in the image lion diagnostic MR scans were performed between 1982
display. and the end of the year 2000.
Precession Motion of a rapidly spinning object when it During the 1980s and 1990s, thousands of clinicians
is subjected to a twisting force or torque. The axis scientists and engineers were engaged in intense efforts to
of the spin moves at a constant rate along a conical enhance the capabilities of MR scanners by reducing their
surface. costs, improving image quality, shortening scan times, and
Relaxation time Measure of the time required for an ex- increasing the number of clinically useful applications.
cited system to decay or relax toward its equilibrium These activities have been extremely productive and it
state. Longitudinal relaxation refers to the longitudinal seems likely that they will continue for years to come and
magnetization and is measured by T1 . Transverse re- the number of scanners in service and the number of scans
laxation refers to the transverse magnetization and is performed each year will continue to increase.
measured by T2 .
Resonance Enhanced response of a physical system that
occurs when the frequency of an applied driving force
I. MAGNETIC RESONANCE AS
is equal or nearly equal to some natural frequency of
A PHYSICAL PHENOMENON
the system.
Rotating frame Coordinate system rotating around the
A. Current Status in Medicine
direction of the static magnetic field at or near the
Larmor frequency. It is useful for depicting the mo- Although the term nuclear magnetic resonance has been in
tion of the magnetization vector. use in the technical community since before World War II,
Selective excitation Excitation of the spins in a limited some changes in terminology have accompanied the onset
region, usually a thin slice, within a sample to be of the medical applications in the 1980s. Partly to avoid
imaged. any fear on the part of patients that radioactive isotopes
Spectrometer Device for resolving, measuring, and char- might be involved (they are not) and partly to forestall any
acterizing electromagnetic energy produced when mat- confusion with the distinct discipline of nuclear medicine
ter changes between allowed energy levels. Examples (which does utilize radioactive isotopes), the adjective nu-
include X-ray spectrometers, optical spectrometers, in- clear is often dropped. The term NMR is still used in the
frared spectrometers, and NMR spectrometers. more technical discussions of the subject, however, as it
Spin Quantum mechanical property of a nucleus or an will be in this article. Two broad general areas of medical
electron that gives it an intrinsic angular momentum applications of NMR have arisen: one is called magnetic
and magnetic moment. resonance imaging (MRI) and the other is called magnetic
Spin–warp technique Method of using gradient fields resonance spectroscopy (MRS). Almost all MRI studies
to encode position-dependent information into NMR are done using signals from hydrogen nuclei (protons),
signals, then using Fourier transform mathematics to although in it is possible in certain specialized cases to
decode the signal and produce an image. use signals from other nuclei, such as sodium, to con-
Voxel Volume element in the object being imaged. struct MR images. Proton MRI, as the name suggests,
The strength of the signal from a voxel determines utilizes the NMR signal, arising mainly in the protons of
the brightness of the corresponding pixel in the the water-containing and fatty tissues of the body, to pro-
image. duce cross-sectional images of the internal anatomy. MRS
utilizes the fact that the resonant frequency of a given nu-
cleus varies slightly depending on what chemical molecule
NUCLEAR MAGNETIC RESONANCE (NMR) is a it is located within—a phenomenon called the chemical
subtle physical phenomenon that can be used to observe shift. Some MRS studies utilize signals from protons to
and exploit a very weak form of magnetism associated distinguish chemical compounds, such as lactic acid and
with the nuclei of certain chemical elements. Beginning the amino acids, that are present in much lower concen-
in the late 1970s and early 1980s, it became possible to uti- trations in the body than is water. But MRS studies more
lize NMR signals from within human patients as a means commonly utilize nuclei such as 31 P (phosphorus) and
13
of imaging internal disease states for the purpose of medi- C (carbon) whose density in tissues is also much lower
cal diagnosis. Although MR scanners were not introduced than that of the water protons. In proton spectroscopy it is
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
a small, dense, positively charged nucleus around which in use, and when the frequency of the oscillation matched
orbit light, negatively charged electrons. The nuclear atom that of the separation of the nuclear magnetic energy lev-
is reminiscent of the solar system, with planets orbiting els a measurable beam deflection occurred. Although the
around the sun. The model has certain defects, which were beam method had important applications to fundamental
removed with the development of quantum theory. Unfor- physics, because it permitted the accurate measurement of
tunately, the quantum theory replaced the previous con- the strength of the magnetic moments of various nuclei, it
cepts of atomic structure with abstruse mathematical con- was still a very long way from permitting the detection of
cepts, and these cannot be depicted in any fashion that is nuclear magnetism in bulk materials.
at once rigorously correct and also visually informative. An unsuccessful attempt to observe magnetic reso-
In other words, quantum theory does not permit pictorial nance directly in a bulk material was made in 1936 by
representations of atoms to be taken literally. Nonetheless, the Dutch physicist C. G. Gorter, but he was thwarted by
the concept of the nuclear atom still provides a convenient technical difficulties. Many advances in radio-frequency
and useful aid, particularly in visualizing magnetic phe- electronics, however, occurred during World War II,
nomena, and we shall make use of it for this reason. particularly as a result of research on radar technology.
Electric currents (i.e., charges in motion) are known to In 1946, separate groups headed by E. M. Purcell at
produce magnetic fields. The nuclear atomic model sug- Harvard and F. Bloch at Stanford detected NMR in bulk
gests one obvious source of electrical current—the nega- materials by use of resonance techniques. This marked
tively charged, orbiting electrons—and, this is, in fact, one the beginning of modern NMR activity. Spectrometers
major source of the magnetic behavior of materials. The were developed that permitted relatively straightforward
other two examples of moving charges within the atom NMR experimentation on test-tube-sized samples of a
are the electron and nuclear spins. large variety of materials. Because the NMR frequency
Much of the experimental information on atomic prop- turns out to be very sensitive to local magnetic fields at
erties came through detailed study of the light emitted the nuclei being studied, it became a major tool in analytic
from excited atoms as the orbital electrons changed states, chemistry for molecular structure determination and for
or orbits. Very careful studies of such spectra with spectro- the identification of unknown compounds. For example,
scopes of the highest possible resolution showed that the by 1955 the NMR spectrometers had revolutionized
spectra of many atoms were split into several components the structural analysis of organic compounds. This
placed very close together. This phenomenon, called the application of small-bore spectrometers to analytical
hyperfine structure, led W. Pauli in 1924 to suggest that chemistry continues unabated to the present time.
atomic nuclei possessed a spinning motion about a cen- At various times several researchers (e.g., J. R. Singer
tral axis. The hyperfine structure was then explained as a in 1959, T. R. Ligon in 1967, and J. A. Jackson in 1968)
weak nuclear magnetic perturbation of the electron ener- reported NMR measurements of one sort or another on
gies. In 1925 S. Goudsmit and G. E. Uhlenbeck made a human tissues. In a 1971 report, which created substantial
similar proposal, suggesting that the electrons also pos- interest, R. Damadian described the use of NMR to detect
sess an intrinsic spinning motion, which provides them cancer in rat tissues. In the years following that report,
with both angular momentum and a magnetic moment. a controversy developed regarding the ability of NMR to
Thus the nuclear atomic model postulates three types of make a specific distinction between benign and malignant
electric current that can act as sources of magnetic forces: tissues. The tissue is not yet entirely settled, although most
the orbital motion of the electrons and the spinning mo- workers would agree that, with present capabilities, NMR
tions of the electrons and the nucleus. instruments operating alone cannot make this distinction.
The spin magnetic moment of nuclei is usually at least Current work on MRI in humans can be traced to the
1000 times smaller than that of electrons, and conse- report of P. C. Lauterbur in 1973 wherein he proposed
quently, the expected strength of the forces produced by the use of gradient fields to permit the incorporation of
nuclear magnetism are extremely small and essentially un- position-dependent information into NMR signals. He
detectable. During the 1930s, I. I. Rabi was able to detect created the name zeugmatography to describe the pro-
and study nuclear magnetic moments of atoms that were cess. Perhaps not surprisingly, the concept has flourished
moving as atomic beams through an ultrahigh vacuum. An while the name has faded from use (although it is in-
array of magnets was used to slightly deflect the beam by cluded in the recently published second edition of the
an amount dependent on the nuclear magnetic moment. Oxford English Dictionary). The first NMR images were
In 1939, Rabi greatly refined the beam deflection method of small capillary tubes filled with water and placed in
and reported the first instance of nuclear magnetic reso- an appropriately modified spectrometer. These were soon
nance. To accomplish this, a weak oscillating magnetic followed by images of human anatomy. The quality of
field was added to that of the permanent magnets already these images rapidly increased, and many modifications
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
of Lauterbur’s original suggestions were made. In 1977, some of the details of the internal structure of the nucleus.
W. S. Hinshaw and his colleagues published a high-quality The spin of the nucleus is characterized by a spin quan-
wrist image. Their technique used time-dependent gra- tum number I . The quantum theory of angular momentum
dient fields to localize signal generation to a “sensitive shows that the value of I is fixed for a given nucleus
point,” which could be scanned through the patient. In and can have only integer or half-integer values (i.e.,
1978, Damadian and associates published an axial cross I = 0, 12 , 1, 32 , . . .). The magnitude of the nuclear angular
√
section of the human chest using a “field focusing” tech- momentum J is I (I + 1) h (h = 1.055 × 10−34 J sec is
nique that has not been widely utilized in subsequent in- Planck’s constant divided by 2π ). The angular momentum
vestigations. By 1980, W. A. Edelstein and co-workers had in a given direction can only assume the 2I + 1 discrete
produced cross-sectional images of the human head and values −I h, −(I − 1)h, up to I h. Note that these results
body by the spin-warp technique they had developed. This display one of the peculiar features of the quantum theory;
was a modification of the two-dimensional Fourier trans- although the total magnitude of the magnetic moment is
√
form method proposed in 1975 by A. Kumar and others. I (I + 1)γ h, the maximum projection in the direction
The spin-warp technique has become widely standardized of the applied field, which is the only portion that can be
in modern NMR scanners. The 1980s witnessed a rapid physically observed, is the somewhat smaller value I γ h.
proliferation of manufacturers and users of MR scanners. Protons, electrons, and neutrons have been found to
These scanners are almost certainly the most significant each have I = 12 . Nuclei with even numbers of both protons
new diagnostic equipment introduced into medicine dur- and neutrons all have I = 0. The nuclei with I = 0 have no
ing this time. angular momentum and no magnetic moment and, there-
Thus, in the 70-year period from 1911 to 1981, NMR fore, cannot produce nuclear magnetism. This accounts for
went through the stages of being an unsuspected property the large number of chemical elements that cannot be stud-
of unknown subatomic structures, through a period of im- ied by NMR. Odd–odd nuclei all have integral values for
portance only to basic physics, followed by a period during I while even–odd and odd–even types have half-integral
which it was developed as a tool for specialized chemical values for I . Table I is a listing of the spin and magnetic
applications, and finally becoming a crucial medical tool
useful to millions of patients and employing thousands of
workers. TABLE I Magnetic Properties of Nuclei of Current or Poten-
tial Medical Interesta
Resonant Magnetic
C. Nuclear Magnetism frequency Nuclear moment Natural
Nucleus at 1 T (MHz) spin, I (10−26 A m2 ) abundance (%)
Nuclear magnetism is an aggregate property of enormous
numbers of identical nuclei responding in step to exter- 1H 42.57 1
2.44 99.985
2
nally imposed magnetic fields. All magnetic phenomena 2H 6.54 1 0.61 0.015
occurring within matter, in the final analysis, can be ex- 3H 45.41 1
2.61 0.0
2
plained only by quantum mechanical methods: analyses 13 C 10.71 1
0.61 1.10
2
based on strictly classical methods are known to lead 14 N 3.08 1 0.29 99.63
to significantly erroneous conclusions. The most useful 15 N 4.31 1
−0.25 0.37
2
method of characterizing phenomena of nuclear mag- 17 O 5.77 5
−1.13 0.038
2
netism, however, involves the use of macroscopic magne- 19 F 1
40.05 2 2.30 100.0
tization vector M. The behavior of M is governed by the 23 Na 3
11.26 2 1.45 100.0
Bloch equations, which are classical in form, but which 25 Mg 2.61 5
2 −0.51 10.0
incorporate, in an empirical fashion, the more fundamen- 31 P 1
17.23 2 0.99 100.0
tal principles of quantum theory. The quantum theory of 33 S 3
3.27 2 0.42 0.75
nuclei in a magnetic field, and in thermal equilibrium with 35 Cl 3
4.17 0.54 75.7
their surroundings, is presented now as a prelude to dis- 39 K
2
3
1.99 0.26 93.258
cussing the Bloch equations. 41 K
2
3
1.09 0.14 6.73
Each different atomic nucleus is viewed as consisting 2
43 Ca 2.86 7
−0.75 0.135
of protons and neutrons. The properties of the nucleus in- 57 Fe
2
1
1.36 0.079 2.2
clude its charge, mass, and size, as well as its spin. The spin 127 I
2
5
8.51 1.68 100.0
endows the nucleus with an angular momentum J and a 2
magnetic moment m. The vectors m and J are proportional a Adapted from data in Walker, F. W., et al. (1984). “Chart of the
to one another m = γJ, where γ , the gyromagnetic ratio, Nuclides,” General Electric, San Jose, CA; Lederer, C. M., and Shirley,
varies from one nucleus to another and presumably reflects V. S. (1978). “Table of Isotopes,” 7th ed., Wiley, New York.
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
moment properties of several nuclei of current or poten- large magnetic moment, and for a rather intense field of
tial medical interest. Note that most of the nuclei in this 1.5T , this formula shows that, for every 10 million protons
table have a positive magnetic moment that corresponds in a sample, there are only 50 more nuclei in the favored,
to a spinning positive charge. Magnetic moments with a parallel state than in the higher energy, antiparallel state.
negative sign correspond to a spinning negative charge. If Other things being equal, it would improve the sensitivity
a magnetic field B is applied to the sample, the nuclei will of MRI if this population difference could be increased.
be given an energy −m · B. Thus, the energy of a state Equation (1) shows that this difference can be increased
depends on the orientation of m, and therefore J, with by increasing the field strength or by lowering the tem-
respect to the applied field. perature. It is clearly impractical to achieve a significant
The projection of the vector m in the direction of B must decrease in patient temperature, however, and there are
take one of the 2I + 1 values −γ h I , −γ h(I − 1), up to substantial technical difficulties in achieving body-sized
γ h I . The nucleus, therefore, has available to it 2I + 1 magnets much stronger than those currently in use.
states with different energies. The energy of these states When a material is magnetized, the strength and di-
will be equally spaced from one another by an amount rection of the effect is given by the vector M, called the
E = γ hB. The state with the lowest energy (the most magnetization, which is defined as the total magnetic mo-
favored state) has the magnetic moment most nearly par- ment per unit volume. If a region of volume V contains
allel to the applied field and vice versa. a largenumber of individual magnetic moments mi , then
Quantum mechanics predicts that if an oscillating mag- M= mi /V , where the sum is over all the sources in the
netic field is applied to the spin system, there will be a region. The dimensions of m are amperes times square me-
resonant exchange of energy between the field and the ters (A m2 ); therefore, the dimensions of M are amperes
spins when the quantum energy hω in the oscillating field per meter. If, as in the present case, the magnetization
corresponds to the separation between adjacent energy is proportional to the applied field, the susceptibility χ
levels. Transitions between nonadjacent energy levels are (which is dimensionless) is defined by the formula
not allowed. This criterion predicts a resonant interaction
when hω0 = γ hB or, equivalently, ω0 = γ B. It is impor- M = χ B/µ0 , (2)
tant to note that this resonant frequency does not depend where µ0 = 4π × 10−7 H/m is a constant called the per-
on either h or I . This ties in with the fact that the nonquan- meability of free space. For any material the total suscep-
tum, classical analysis, to be discussed later, of magnetic tibility will be the sum of the contributions from each of
moments in a magnetic field gives the same value for the the relevant sources of magnetic moment; the orbital elec-
characteristic frequency. tron motion, the electron spin, and the nuclear spin. In the
The states with the magnetic moment in the direction of present case, of course, we are particularly interested in
the field have a lower energy than those with the opposite χn , the contribution of the nuclei to the total susceptibility.
orientation. Consequently, if the spin system can come to Statistical analysis of the distribution of the nuclei among
equilibrium with its surroundings at a temperature T , the the available energy states shows that a nucleus with a
lower energy states will become more populated than the spin I , magnetic moment m (A m2 ), and a density of
higher energy states and the substance as a whole will take ρ (spins per cubic meter) will have a nuclear magnetic
on a net nuclear magnetization. This represents an aggre- susceptibility given by
gate effect of the tendency of all of the individual nuclei
to orient themselves parallel to the applied magnetic field. µ0 ρm 2
This tendency toward alignment is, of course, opposed by χn = . (3)
3kT
the randomizing effects of the thermal energy present in
Pure water has a density ρ of 55 moles/liter, or equiva-
the material.
lently, 6.62 × 1028 protons/m3 . Using the values in Table I,
The difference between energy levels caused by the ap-
the nuclear magnetic susceptibility for the protons in wa-
plied magnetic field is quite small in comparison to the
ter is found to be 3.86 × 10−9 . The magnetic behavior of
thermal energy. For a system with only two levels (I = 12 ),
water is particularly relevant to MRI because most of the
the ratio of the number of spins in the lower energy state
signal derived from human tissues originates from water
n + to that in the higher energy state n − is given by
molecules. Note that m in Table I and Eq. (3) refers to
n + /n − = eE/kT = eγ hB /kT . (1) the total magnitude of the magnetic moment vector. An-
other convention often used is to refer to the maximum
Here, k is Boltzmann’s constant and T is the absolute observable component of the magnetic moment vector as
temperature. At body temperature 37◦ C or 310 K, kT = the magnetic moment. √ In the notation used here this com-
4.28 × 10−21 J. Even for protons, which have a relatively ponent is equal to m I /(I + 1).
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
which normally last a few milliseconds or less. When- relaxation processes to be described later. These latter ef-
ever Eq. (4) is valid, vector calculus states that the time fects normally produce changes in M that occur on the
rate of change of M must be perpendicular to M because millisecond to seconds time scale.
of the properties of the vector cross product. This means The most common approach to creating the magnetic
that, under the action of the externally applied fields, M resonance phenomena is to use a strong field, which we
cannot change in length. Therefore, if by some means, a shall designate as B0 , in the z direction and to add to
nuclear magnetization has been created within a specimen it a weaker oscillating field B1 that is oriented at right
at an initial time, the externally applied forces will cause angles to the z axis. If the oscillating field has both x
the magnetization to move with time, but only in a way and y components, B = B1 (cos ω1 t ı̂ − sin ω1 tˆ), then in
that keeps the length of the magnetization vector constant. a frame rotating at the frequency ω1 (moving clockwise
The simplest situation is when the external field is simply when viewed from the positive z direction) it is just a con-
a constant B directed along the z axis. If at t = 0 there stant B1 along the rotating x direction. Because such a
is a transverse component of M, Mt , pointing along the field contains two components at right angles to one an-
x axis, the solutions to Eq. (4) are other, it is referred as a quadrature excitation field and is
said to be circularly polarized with a clockwise rotation.
Mx = Mt cos ω0 t and M y = −Mt sin ω0 t,
It is not hard to show that a linearly polarized field with
where ω0 = γ B and Mz is constant. That is, the component twice the amplitude (e.g., B = 2B1 cos ω1 t ı̂) will have the
of the magnetization along the z axis remains unchanged same effect on the spins as the circularly polarized field
as time goes on, while the transverse magnetization rotates above. Interestingly a field rotating in the “wrong” direc-
at a rate (called the Larmor frequency) about the direction tion B = B1 (cos ω1 t ı̂ + sin ω1 t,ˆ) will have essentially no
of the applied magnetic field. Thus the total M vector at effect on the spins. In this article we will assume that a
any point moves steadily, at a constant rate, sweeping out a quadrature B1 is used since this simplifies the analysis
cone whose axis is the direction of B. This motion is analo- somewhat.
gous to the motion of a rapidly spinning gyroscope (such The exact solution of Eq. (4) in these circumstances
as a top) responding to its own weight. This precession is not difficult but the results are more complex than we
is of basic importance to the detection of nuclear mag- wish to present here. The essential features of the solution
netism that, as has been previously mentioned, is much are that the B1 field has a negligible effect on the mo-
too weak to be detected directly by the magnetic forces tion of M unless its frequency ω1 is close to the Larmor
it exerts. The precession of the transverse magnetization frequency ω0 . More specifically, unless ω1 is within a fre-
produces a time-dependent magnetic field and, therefore, quency range γB1 of ω0 , the oscillating field will be in-
by Faraday’s law, a time-dependent electric field. This effective. If ω1 is equal to ω0 , the motion in the rotating
electric field can be detected as a voltage in a coil situated coordinate system is very simple. Then the magnetization
outside the sample. It is of fundamental importance that vector will rotate about the B1 field, which will be constant
this induced voltage turns out to be large enough, in many in this frame at the rate γB1 . Thus, if the magnetization is
cases, that the precessing nuclear magnetization can be along the z axis at time t = 0, it will rotate about B1 and
detected electronically. Note that, although different lines will make an angle θ = γB1 t with the z axis after time t.
of reasoning are used, both the quantum mechanical and Thus, if the oscillating field operates for a time t equal
the macroscopic approaches lead to a characteristic fre- to π/(2γB1 ), the magnetization will rotate 90◦ (π/2 rad)
quency given by ω0 = γ B. and will be located in the transverse plane. If it operates
A concept that is often utilized to describe the motion of for twice this time, the magnetization will be rotated 180◦
the magnetization vector is a coordinate system rotating at and would be inverted from its initial position.
or near the Larmor frequency. If the new coordinate sys- If a sample is placed in a strong magnetic field B0 , the
tem is rotating at exactly the Larmor frequency, then in it, earlier analysis shows that initially the magnetization is
for the example previously given, there is no motion of the zero, but should increase with time to an equilibrium value
magnetization vector. To specify the direction of rotation, M0 = χn B0 /µ0 . The gyroscopic equations cannot describe
we note that nuclei with positive values for the gyromag- this process since Eq. (4) shows that the length of the M
netic ratio precess in a clockwise direction when viewed vector cannot be changed by the external fields. The inter-
from the positive z direction. The advantage of the rotat- nal fields provide the answer to this paradox as they permit
ing coordinate system is that it disentangles the very rapid an exchange of energy between the surroundings, referred
precession motion, which usually takes place at mega- to somewhat loosely as the lattice, and the nuclear spin
hertz rates, and that is caused by the strong static field, system. Because the internal fields are the result of rapid,
from the much slower motions produced by weak, super- essentially chaotic, motion of the atoms of a liquid relative
imposed, oscillating external magnetic fields and from the to one another, it is extremely difficult to calculate, from
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
first principles, what those fields are as a function of time, ena within human tissues. In more general applications
much less their effect on the magnetization. In 1946 Felix the Bloch equations have been found to provide good pre-
Bloch cut the Gordian knot by lumping the total effect dictions for the behavior of liquid or liquidlike samples.
of all these internal processes into two parameters called In solids, however, the Bloch equations require substan-
the relaxation times. The effect of one of the parameters tial modification to give accurate results. The protons
called T1 , the spin–lattice relaxation time, is to govern the in biological tissues behave, from a magnetic resonance
rate at which Mz approaches it equilibrium value M0 : standpoint, as though they were located in a liquid
environment.
dMz /dt = (M0 − Mz )/T1 . (5)
In equilibrium, there is no transverse magnetization,
F. Relaxation Times
consequently the internal fields must act to reduce any Mx
and M y that may be present. Bloch proposed quantifying The range of values taken by T1 and T2 in human tis-
this process by using a second relaxation time T2 : sues is crucial to determining the practicality of MRI for
human tissues. If an unmagnetized sample (e.g., a human
dMx /dt = −Mx /T2 , dM y /dt = −M y /T2 . (6) patient) is placed in an uniform magnetic field, it is ini-
The reason that different relaxation times are needed to tially unmagnetized. The Bloch equations show that the
describe transverse and longitudinal relaxation is that the nuclear magnetization will gradually build up along the
strong external field biases the response of the spin system z direction and approach M0 asymptotically according to
so strongly that these two magnetization components the exponential expression:
respond differently to the weak, internal fields. Analysis Mz (t) = M0 (1 − e−t /T1 ). (8)
of the microscopic mechanisms responsible for relaxation
show that T2 will always be shorter than, or at most equal Thus, Mz will achieve 63.2% of its final value (M0 ) in a
to, T1 . time equal to T1 , 86.5% of M0 in 2T1 , and so on. If it is
Bloch conjectured that the total motion of the magneti- desired to achieve 99% of the total possible magnetization
zation vector can be described as the superimposed effects it is necessary to wait for a time of 4.6T1 . If T1 is too long,
of the gyroscopic motion (driven by the externally applied a prohibitively long period can be required to achieve a
fields B0 and B1 ) and the relaxation processes (associated useful magnetization. The fact that Gorter failed to detect
with internally generated magnetic fields). This combina- nuclear magnetism in 1936 may have been the result of
tion leads to the final form for the Bloch equations in the an unfortunate choice of material, which had too long a
stationary frame with unit vectors ı̂ ,ˆ, and k̂: value for T1 . It should be noted that there are materials that
have T1 values as long as hours or even days. Fortunately,
dM Mx My M0 − M z
=− ı̂ − ˆ + k̂ +γ M ×[B0 +B1 (t)]. mobile protons in biological tissues have T1 values of, at
dt T2 T2 T1 most, a few seconds (Table II).
(7)
The only time a signal can be detected from the nuclear
The Bloch equations give a complete description of spins is when a transverse magnetization is present. This
the behavior of the magnetization within a body. Qualita- can be achieved by using a short burst of radio-frequency
tively, they express relatively simple ideas, the transverse (rf) energy, the B1 field or rf pulse, at, or very near, the
magnetization Mx ı̂ + M y ˆ is constantly relaxing toward Larmor frequency. A 90◦ pulse will rotate a magnetization
zero while precessing rapidly at the Larmor frequency that is initially along the z axis into the transverse plane. If
γB0 . The longitudinal magnetization Mz is constantly re- the B1 field is then turned off, the transverse magnetization
laxing toward its equilibrium value M0 . If B1 is not zero Mt = Mx2 + M y2 will precess at the Larmor frequency. Its
it is constantly rotating M about an axis parallel to B1 in amplitude will decay according to the relation
the rotating frame. In practice, the solution to the Bloch
Mt (t) = M0 e−t /T2 cos ω0 t . (9)
equations may be relatively complicated, particularly if
the frequency of B1 (t) is not exactly equal to the Larmor The electric signal picked up during this time is called the
frequency. free induction decay or FID (Fig. 2). If T2 is too short,
The relaxation times T1 and T2 provide only an em- the signal will decay away so rapidly that no useful in-
pirical treatment of the effects of the internal mag- formation can be extracted from the FID. Note that as
netic fields. In practice, they must be found by ex- soon as B1 disturbs the longitudinal magnetization from
periment rather than by calculation. Once T1 and T2 its equilibrium value M0 , Mz starts to rebuild according
have been determined, experimental results indicate that to Eq. (8). Therefore, it is possible to put the spin sys-
the Bloch equations provide a completely satisfactory tem through a periodic excitation cycle using a series of
description of all nuclear magnetic resonance phenom- 90◦ pulses. Between pulses the longitudinal magnetization
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
TABLE II Proton Relaxation Times for Various Substances in level populations given by Eq. (1). Once saturation has
(37◦ C)a occurred it is necessary to wait for a time approximately
T 1 (msec) equal to T1 to permit the magnetization to rebuild [Eq. (8)]
before further signals can be obtained from the specimen.
25 MHz 64 MHz T 2 (msec)
An additional time consideration is the duration of the
Pure water (deoxygenated) 4500 4500 2900 rf pulse necessary to carry out the desired rotations of the
Cerebrospinal fluid 4300 4300 2200 magnetization. The length of this pulse is usually a few
Blood (oxygenated) 860 940 350 milliseconds or less. This is much shorter than T1 or T2 for
Skeletal muscle 585 870 47 the tissues of interest. Thus, there is no significant change
Brain in the length of the magnetization vector during the time
White matter 520 557 72 that the rf pulse is being applied.
Cortical gray matter 690 920 100 The relaxation times of protons in several human tissues
Liver are indicated in Table II. P. A. Bottomley and his coworkers
Normal 345 490 43 have carried out extensive analysis of the published data,
Hepatoma 810 1080 84 and have concluded that a large variability exists from one
Fat 220 260 84 investigation to another in published values for relaxation
Multiple sclerosis plaque times. Therefore, the values in Table II should not be re-
(mean) — 1315 174 garded as precise—standard deviations of ±10% or more
in the data are not unusual. It is likely that the improved
a Adapted from data in Bottomley, et al. (1984). Med. Phys. 11,
instrumentation now available, and the recently acquired
425–448; Bottomley, et al. (1987). Med. Phys. 14, 1–37; Hopkins, et al.
significance of this data will lead to a rapid improvement
(1986). Magn. Reson. Med. 3, 303–311; Brooks, and DiChiro, (1987).
Med. Phys. 14, 903–913; Hardy, et al. (1986). Magn. Reson. Med. 3, in the precision and accuracy with which T1 and T2 values
935–940; Larsson, et al. (1988). Magn. Reson. Med. 7, 43–55. for human tissues are known. In general, it appears that for
most human tissues T1 is substantially larger than T2 . For
rebuilds toward its equilibrium value. At the same time protons T1 increases in a significant way as the static field
transverse magnetization is precessing and producing a strength, and thereby the Larmor frequency is increased.
FID signal. Of course, Mz will not have completely However, T2 tends to remain constant, or decrease slightly,
achieved its steady-state value M0 by the time of the subse- as the frequency is increased.
quent 90◦ pulse. If the pulses are repeated too rapidly, the As an example of the use of relaxation times to discrim-
Bloch equations show that the magnitude of the FID goes inate between normal and malignant tissues, the values for
to zero, an effect called saturation. From a quntum me- both liver and for a hepatoma, a tumor derived from liver
chanical point of view, saturation is the result of an exces- tissue, are given in Table II. In many cases, however, the
sive amount of B1 excitation that eliminates the difference relaxation time differences between normal and malignant
tissues are not so clear cut.
The concept of the T2 decay is usually explained using
the idea of the dephasing of the nuclear spins present in the
sample. Consider a transverse magnetization that has been
created by a 90◦ pulse. All the spins in the sample have
experienced the same applied magnetic field and imme-
diately after the pulse they all have the same phase. That
is, their individual nuclear moments are all pointed in the
same direction and the induced voltage they produce in
the receiver coil is at its maximum. Once the externally
applied B1 field is removed, however, the individual nu-
clei are still subjected to the weaker, but persistent, effects
of their different local environments. At any instant this
leads some nuclei to be precessing faster, and some slower,
than the average rate of precession, which is given by the
Larmor frequency, set by the external field B0 . The result
FIGURE 2 Free induction decay. The FID is a damped oscillation.
of this is that the spins get increasingly out of phase with
The case illustrated here shows a beat pattern between the signal
from spins precessing at the Larmor frequency and a reference one another as time goes on. This causes the transverse
frequency shifted from it by 370 Hz. The T2 is 30 msec. (Courtesy magnetization and, consequently, the induced voltage to
of GE Medical Systems.) decay exponentially toward zero.
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
It should be realized that the discussion of dephasing located in higher fields will precess more rapidly than
provided here is rather casual. Since we are inferring that those in lower fields. The net result is again a dephasing
the dephasing occurs among individual nuclear spins and of the spins across the sample and an undesirable decrease
that this determines the time course of the relaxation of of the induced voltage in the receiver coil. However, al-
the macroscopic transverse magnetization, the argument though the total magnetization of the sample may have
should include a quantum mechanical justification. Such decrease to a low value, those spins relatively close to
an argument would be too lengthy to provide here. This one another in space will still be nearly in phase one an-
topic is discussed thoroughly in the book by Slichter. For other. Thus, although it is not evident externally, a form
qualitative purposes, the idea of individual nuclear spins of spin order still exists within the sample, even after the
gradually dephasing with one another remains a useful external signal is no longer detectable. In 1950, E. Hahn
one and we will continue to employ it. showed that this remnant order can be detected by apply-
Two important physical consequences associated with ing a 180◦ pulse to the precessing spins at a time τ after
dephasing are motional narrowing and spin echoes. It the 90◦ pulse that created the transverse magnetization.
might be thought that nuclei that are diffusing rapidly After such a pulse, the spins that had advanced in phase
among their neighbors during an FID would experience beyond the average by an amount of δφ are now behind
more rapid dephasing than nuclei that are relatively fixed the average by the same amount. Because they are still
in position. In fact the opposite is true. The explanation precessing more rapidly, however, these spins will catch
is that if a spin is relatively fixed in position, it will be up with the average at a time τ after the 180◦ pulse. At this
forced to respond to whatever its local magnetic environ- time all the spins will be back in phase with one another
ment happens to be for an extended period. Unlike the and the receiver will detect a signal called a spin echo.
externally applied fields, however, there is no reason for Of course, all of the microscopic dephasing mechanisms
the local magnetic fields to be identical from one location are operating continuously so that the amplitude of the
to another because the local environments are not corre- spin echo is reduced by a factor e−2τ/T2 from the initial
lated. Therefore, if the sample consists of a large number amplitude of the FID.
of localized spins, as in a solid, they will tend to drift In the early days of NMR, spin echos were used to over-
rapidly out of phase with one another. On the other hand, come the relatively high inhomogeneities of the available
if the nuclei are not fixed but move from one location to magnets. In MRI systems today, however, the magnets
another rapidly, as in a liquid, the local fields vary from have sufficient homogeneity that this is not usually nec-
instant to instant, causing the phase of the nuclear pre- essary. In imaging systems, however, special coils that
cession to increase almost as often as it decreases, and produce gradients in B0 are extensively utilized. Spin
the overall dephasing of the total system proceeds more echoes are often used, as a routine part of the imaging
slowly. Therefore, protons located on fixed sites experi- sequence, to reverse the dephasing produced by the de-
ence a very rapid dephasing, which means a short T2 , while liberately employed gradient fields. If the gradient field is
those on freely diffusing water molecules have much larger constant in time, a 180◦ rf pulse may be used to produce an
values for T2 . After Fourier transformation a larger value echo as just described. On the other hand, if the gradient
for T2 corresponds to a narrower resonance line. There- field itself is reversed after being applied for at time t, an
fore, this important phenomenon wherein nuclei in highly echo will occur at time 2t without the need for a rf pulse.
mobile molecules manifest long T2 values is called mo- This is called a gradient recalled echo.
tional narrowing. This remarkable phenomenon was first The physical environment within biological tissues is
explained by N. Bloembergen and coworkers in 1948. It extremely heterogenous. Within a single cell there are
is a very strong effect. The T2 of protons in ice, for ex- known to be numerous microscopic structures includ-
ample, is shorter than that of protons in water by a factor ing the nucleus, the cell membrane, mitochondria, and
of 100,000. Therefore, signals from solid tissues, such as microtubules. It is likely that the magnetic environment
teeth and dense bone, decay almost instantly after excita- varies somewhat from location to location as a result of
tion and cannot be detected by MRI. As a consequence, this heterogeneity, and consequently, it might be expected
MRI provides images based on the distribution of mobile that the proton relaxation times would vary with intra-
protons only. Because the local feilds are random from one cellular location. Experimentally, however, it is found
nuclear location to another, there is no hope of reversing that, although the relaxation times vary from one organ
this form of spin dephasing and it consequent decrease of to another, there usually does not appear to be a signif-
voltage in the receiver coils. icant deviation from single exponential decay within a
Another source of dephasing of the spins is caused by single histological region. The explanation involves the
macroscopic inhomogeneities in the static applied field rapid self-diffusion of water molecules among one an-
B0 . If B0 varies from one location to another, the spins other. At body temperature the self-diffusion coefficient
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
for water is D = 3.7 × 10−5 cm2 /sec. During a time t II. NUCLEAR MAGNETIC
the random-walk motion associated with self-diffusion RESONANCE SCANNERS
will√give each molecule an average displacement equal AS MEDICAL INSTRUMENTS
to Dt. For t = 10 msec this gives an average displace-
ment of 6 µm and for t = 100 msec, 19 µm. These dis- A. System Considerations
tances are comparable with average cell sizes. This shows
The overall operation of a NMR scanner is controlled by
that during a typical T2 interval a given water molecule
a computer (Fig. 3). It provides the pulse-timing infor-
has time to encounter much of the intracellular and local
mation for the gradient and radio-frequency transmitter
extracellular environment, and therefore to average out
coils. It also switches on the preamplifier and the receiver
the inhomogeneities in local field implied by the cellular
circuitry during the time when data is being acquired from
heterogeneity.
the nuclear spins. The data are acquired as free induction
decay or, more commonly, spin echo signals. Extensive
mathematical calculations, usually a two-dimensional
G. Chemical Shift
Fourier transformation, are required to convert the FID
As discussed, the electrons orbiting each atom respond in or spin echo data into an image and the calculations are
a diamagnetic fashion when an external field is applied. done by the computer. For permanent magnet systems
As a result, the total magnetic field at the nucleus of an no magnet power supply is required; for superconducting
atom is some what less than the externally applied field. magnet systems the magnet power supply is needed only
The exact value of this reduction depends on the details initially when the magnet is being energized.
of the electron orbits in the given molecule and will be The NMR signal is sufficiently weak that interfering
different in different molecules. Consequently, in a given electrical noise is a major consideration. To minimize
external field and for a given type of nucleus, the nuclear outside electrical interference, some form of electrical
resonant frequency will vary from one type of molecular screening is normally placed around the scanner or the
environment to another. This change in frequency δ f is scan room, and electrical filters are used on those circuits
called the chemical shift. It is usually small, and δ f / f 0 that could transmit outside noise to the receiver. Figure 4
can be specified in parts per million. These small shifts shows a patient within a screened room being positioned
are easily measured, however, and in MR spectroscopy for scanning. In Fig. 5 the patient is seen within the magnet
they are the basis for discriminating one molecular species bore in the location that permits scanning to be done.
from another. In MRI the chemical shift leads to an image The strength of NMR signal generated by the precess-
artifact that slightly confuses the boundaries between fat ing spins is proportional to the degree of alignment of
and water-containing tissues. Protons in water and fat have the spins [M0 in Eq. (2)] and to their rate of precession
resonant frequencies differing by 3.5 ppm. (γ B0 ). Since both those quantities are proportional to field
B. Magnets
The magnet is probably the most significant portion of a
FIGURE 4 Preparation for scanning. The superconducting mag-
whole-body scanner. The magnets used vary substantially
net shown here operates at 1.5 T. (Courtesy of GE Medical
Systems.) in terms of field strength, shape, and basic design. How-
ever, they all must meet certain basic requirements. One
obvious requirement is that the magnet be large enough
strength, the signal increases as B02 . The noise in the sys- to admit a human body and produce a field strength that
tem can be brought to a very low level by proper design is intense enough to produce a strong proton NMR signal.
of the electronics. However, there is some electrical noise Equally important, though not so obvios, is the require-
that is generated within the patient and that cannot be re- ment for high homogeneity, or uniformity of the magnetic
moved by improved circuit design. The ultimate source of field strength from one part of the imaging volume to an-
this noise is the random motion of ions and charged macro- other. It is essential that, unless gradient coils are being
molecules within the patient’s body and is of thermal ori- used to deliberately modulate it, the Larmor precession
gin. The noise voltage increases approximately linearly proceed at the same rate throughout the entire portion
with increasing frequency. of the patient that is to be imaged. The homogeneity re-
A key parameter in determining overall image quality is quirment puts strong limitations on the acceptable magnet
the signal-to-noise ratio (SNR). The arguments just given designs. Superconducting and resistive systems are de-
indicate that the SNR should increase linearly as the field signed to provide a cylindrical symmetry. A cylindrical
strength is increased. This is the basis for the use of strong opening, called the room temperature bore, is available to
magnetic fields in NMR scanners. If it is desired to achieve permit placing the patient into the region of strong and
high-resolution proton images, thin slices and small pic- highly homogeneous field.
ture elements (as will be discussed later) must be used; At the present time clinical proton imaging is usually
however, this leads to a decrease in the amount of signal done using magnetic field strengths in the order of 0.2 to
available to determine the pixel brightness numbers. If this 1.5 T, although, in about 2000, commercial MRI systems
process is pushed too far, the images will become grainy operating at 3 T began to become available. Later, in the
because of the effects of the noise. By increasing the field section on current trends, research scanners operating at
strength of the magnet, the available signal is increased; even higher field strengths will be described. The types
this can be used to support higher resolution imaging. of magnet that have been used to produce clinical im-
ages have included permanent magnet systems, iron-core
electromagnets, and resistive and superconducting mul-
ticoil magnets. Each of these designs has certain advan-
tages. However, for producing fields greater than about
0.5 T, only the superconducting systems are practical;
field strengths in this range are beyond the capabilities
of present-day permanent magnet materials. In resistive
systems the coils are made of a conventional conductor,
such as copper or aluminum. If enough current is run
through them to produce whole-body sized fields much
above 0.2 T, so much heat is generated in the windings
that even with water cooling there is a likelihood of burn-
ing the insulation or actually melting the wire.
FIGURE 5 NMR scan. The patient has been positioned in the The phenomenon of superconductivity provides the best
center of the magnet and can be seen through a window above approach to achieving strong magnetic fields. Since 1911
the operator’s console. (Courtesy of GE Medical Systems.) it has been known that some materials when cooled to
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
regulations. During the 1980s several manufacturers suc- use of static fields up to 4.0 T and, under appropriately con-
cessfully sought FDA approval for their scanners. trolled conditions, an average heat input to the patient’s
As additional experience was gained during the 1990s, body (specific absorption rate or SAR) of 4.0 W/kg. The rf
FDA approval was granted for a variety of scanner en- heating that occurs under standard imaging conditions is
hancements, such as the use of higher field strengths and comparable to that resulting from normal metabolic activ-
a wider variety of RF and gradient coils. ity and is unnoticeable to the patient. Likewise, the possi-
NMR scanners place the patient in an environment that ble effects of dB/dt on nerve excitation and the magneto-
is quite unlike that of any other medical instrument. Ini- hydrodynamic interaction between flowing blood and the
tially, there were several areas of concern that, with expe- static magnetic field appear to require conditions far be-
rience, have become better understood and appear not to yond those used in modern scanners. At the present time
represent a danger to patients. The area of greatest contin- these concerns represent hypothetical, rather than actual,
uing concern is the interaction between the strong, static, hazards of the scanning process.
magnetic field and ferromagnetic substances inadvertently
brought into the region of the scanner. One version of
this problem arises from the fringing field surrounding
III. IMAGING TECHNIQUES
the magnet, which can be treacherously strong. Many ob-
jects common in hospitals (e.g., oxygen bottles, mops,
A. Selective Excitation
fans, and hairpins) contain enough magnetic material that
they can be drawn into the magnet with great force and One of the key capabilities of MRI scanners is the ability
rapidity. Such flying objects are extremely dangerous to to excite a single, thin slice of spins within the patient. This
anyone in their path. For this reason, most manufacturers permits the construction of images that have the character
and users of the scanners go to great lengths to limit access of two-dimensional cross-sectional cuts through the pa-
to the vicinity of the scanner. Permanent magnet systems, tient’s anatomy. NMR imaging resembles CT scanning in
and some superconducting systems that have magnetic producing this type of anatomical image. However, MRI
shielding around them, have smaller fringing fields and has a substantial advantage over CT in that it permits the
are less susceptible to this effect. A second version of this location of the slice to be chosen electronically by the op-
problem comes about because some patients have ferro- erator without moving the patient or any components of
magnetic substances implanted within their bodies. This the scanner. In MRI, imaging planes of any orientation
is usually the result of a prior surgical procedure, such as may be chosen, and these planes may also be moved elec-
the clipping of a diseased blood vessel, but in some cases tronically from side to side, top to bottom, or front to back
iron fragments (e.g., shrapnel) have become embedded through the patient’s anatomy.
in a patient’s tissues during some sort of accident or ex- The key to selective excitation is carrying out the rf exci-
plosion. Patients are not always aware of the presence of tation in the presence of a gradient field. Suppose the static
these objects. Therefore care has to be taken before scan- field is uniform across the patient. If a rf pulse is applied at
ning to exclude those patients with possibly dangerous fer- the Larmor frequency, ω0 = γ B0 , for a time long enough
romagnetic implants. Implanted cardiac pacemakers can to create a 90◦ pulse, this will excite spins over a large
malfunction, or conceivably, be permanently damaged be- volume of the patient. If, however, the z-gradient coil is
cause of exposure to strong magnetic fields. Therefore used to apply an additional static field, Bz = G z z, at the
patients with these devices in place are not normally can- time of the rf pulse, the resonance condition will be met
didates for NMR scanning. strictly only in the plane z = 0. Spins far from this plane
Other areas of initial concern were the effects of the are well off resonance and essentially will be unaffected
static field on normal tissue function; the possibility that by the rf pulse. Spins at z = 0 will be rotated by 90◦ just
electric fields generated by the rapidly changing gradient as if no gradient field were present. Spins close to, but not
fields (i.e., the “dB/dt” effect) could cause nerve stimula- at, the plane z = 0 will be partially excited. To predict the
tion or irregularities in the cardiac rhythm; tissue heating exact behavior of the excitation, as a function of z, near the
associated with the rf excitation field; and possible effects origin it is necessary to carry out a solution to the Bloch
on blood pressure resulting from forces of interaction be- equations. It turns out that spins in the selected slice, but
tween blood and the static magnetic field. In the scanners slightly off the center plane, will also be flipped through
presently used, all of these effects appear to be readily tol- approximately 90◦ but will have phase differences with
erable, and in most cases, negligible. The FDA continues those spins at z = 0. This will reduce somewhat the signal
to receive and evaluate designs for more advanced scan- generated by the slice. It can be shown that a more perfect
ners, and the regulatory aspects of scanner safety continue slice profile will result if, instead of using a rf pulse of
to evolve. Of particular note are guidelines approving the constant amplitude, the pulse amplitude is modulated by
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
A. Proton Imaging FIGURE 10 Sagittal head image. The image represents a 3-mm
thick slice near the midline of a normal volunteer’s head. Like all
Once clinical MR scanners became available certain ad- the images shown here this was taken at 1.5 T. (Courtesy of GE
vantages and disadvantages of their use became estab- Medical Systems.)
lished. Among the chief advantages of MR are the ability
to image almost any region of the body, the very high con- available on CT scanners. The contrast in these images
trast available between soft tissue structures, the ability to is said to be T1 weighted—that is, the scan repetition time
vary the plane of imaging at will, the ability to vary the tis- TR was short enough that not all tissues had time to mag-
sue appearance by varying the scan parameters, the lack of netize completely. Thus, tissues with larger values of T1 ,
the need for any invasive step such as the injection of con- such as the cerebrospinal fluid, appear dark on these scans.
trast agents, and the absence of any ionizing radiation (i.e., Good contrast is seen between gray and white matter of
X rays). The relative disadvantages include the cost, time the brain. MR is completely free of any interference from
requirements, and the inability to detect certain materials. bone. This makes imaging of the cerebellum, near the
The cost of the examination varies with the time required base of the brain at the back of the head and surrounded
to complete it, the number of separate images required, and by rather dense bone, more effective with MR than with
other factors. A large portion of the scanner cost is associ- X-ray methods such as CT.
ated with the magnets, which require expensive materials
such as large amount of superconducting wire. Each im-
age acquisition requires a time ranging from less than one
second to as much as 20 min. During this time it is nec-
essary for the patient to remain still to avoid blurring the
image. CT scans can be taken much more quickly than this.
Certain materials (calcium, in particular) that are readily
seen in X-ray studies such as CT do not give a NMR sig-
nal, and therefore appear only as voids on MR images.
This is a drawback especially in the diagnosis of certain
tumors. One consequence of this balance of advantages
and disadvantages is that CT and MR have become com-
plementary imaging modalities with neither one showing
signs of displacing the other.
The advantages of MR scanning have been especially
pronounced in imaging of the brain and spinal cord. At
the present time a large fraction, perhaps around 70%,
of all MR studies are done for central nervous system
indications.
FIGURE 11 Coronal head image. The subject was in the same
Figures 10 and 11, both done on normal volunteers, position as in Fig. 10 but by interchanging the gradients used for
illustrate several advantages of MRI for brain imaging. selective excitation an image of a slice at right angles to that shown
Neither the sagittal nor coronal section are routinely in Fig. 10. (Courtesy of GE Medical Systems.)
P1: GRB/GWT P2: GPJ/GAX QC: GAE/FYD Final Pages
Encyclopedia of Physical Science and Technology EN008M-395 June 29, 2001 15:52
FIGURE 12 Eye image. A surface coil was placed over the right
eye to obtain a high-resolution image.
B. Current Trends
Magnets in which the patient is placed within a horizon-
tal gap between two magnetic pole faces are now being
at these field strengths scanners operating at 3–4 T now Chen, C.-N., and Hoult, D. I. (1989). “Biomedical Magnetic Resonance
have the potential of serving as high performance general- Technology,” Adam Hilger, Bristol and New York.
purpose clinical scanners. Edelstein, W. A., Hutchinson, J. M. S., Johnson, G., and Redpath, T.
(1980). Spin-warp imaging and applications to human whole-body
In addition at a few sites have begun studies using imaging. Phys. Med. Biol. 25, 751–756.
whole-body research scanners operating at fields well Fukushima, E., ed. (1989). “NMR in Biomedicine: The Physical Basis,”
above 4 T. For example, in December 1997 an 8 T whole- American Institute of Physics, New York.
body scanner was installed at Ohio State University in Hayes, C. E., Edelstein, W. A., Schenck, J. F., Mueller, O. M., and Eash,
Columbus, Ohio (Fig. 20). Since that time several 7 T M. (1985). An efficient highly homogeneous radiofrequency coil for
whole-body NMR imaging at 1.5 T. J. Magn. Reson. 63, 622–628.
whole-body scanners have been installed and whole body Hinshaw, W. S., Bottomley, P. A., and Holland, G. N. (1977). Radio-
scanners planned to operate in the range of 9–10 T are graphic thin-section image of the human wrist by nuclear magnetic
currently under discussion. Although these magnets have resonance. Nature (London) 270, 722–723.
patient bores large enough to admit the entire body, most Hollis, D. P. (1987). “Abusing Cancer Science,” Strawberry Fields,
of the early research applications have been concerned Chehalis, OR.
Hoult, D. I., and Lauterbur, P. C. (1979). The sensitivity of the zeug-
with brain imaging using head coils. Whole-body scan- matographic experiment involving human samples. J. Magn. Reson.
ners operating at 7 T and above will probably be used 34, 425–433.
predominately for research, rather than for clinical diag- Jin, J.-M. (1998). “Electromagnetic Analysis and Design in Magnetic
nostic purposes, for several years to come. Resonance Imaging,” CRC Press, Boca Raton, FL.
Kleinfeld, S. (1985). “A Machine Called Indomitable,” Times Books,
New York.
Lauterbur, P. C. (1973). Image formation by induced local interactions:
SEE ALSO THE FOLLOWING ARTICLES examples employing nuclear magnetic resonance. Nature (London)
242, 190–191.
Ogawa, S., Tank, D. W., Menon, R., et al. (1992). Intrinsic signal changes
ELECTRON SPIN RESONANCE • FERROMAGNETISM •
accompanying sensory stimulation: functional brain mapping with
FOURIER SERIES • MAGNETIC MATERIALS • MICROWAVE magnetic resonance imaging. Proc. Natl. Acad. Sci. USA 89, 5951–
MOLECULAR SPECTROSCOPY • NUCLEAR MAGNETIC RES- 5952.
ONANCE • QUANTUM MECHANICS Robitaille, P.-M. L., Warner, R., Jagadeesh, J., et al. (1999). Design and
assembly of an 8 tesla whole-body MR scanner. J. Comput. Assist.
Tomogr. 23, 808–820.
Schenck, J. F. (1996). The role of magnetic susceptibility in magnetic
BIBLIOGRAPHY resonance imaging: magnetic field compatibility of the first and second
kinds. Med. Phys. 23, 815–850.
Abragam, A. (1961). “Principles of Nuclear Magnetism,” Oxford Uni- Schenck, J. F. (2000). Safety of strong, static magnetic fields. J. Magn.
versity Press (Clarendon), London and New York. Reson. Imaging 12, 2–19.
Atlas, S. W., ed. (2001). “Magnetic Resonance Imaging of the Brain and Schenck, J. F., Dumoulin, C. L., Redington, R. W., Kressel, H. Y., Elliott,
Spine, 3rd ed.,” Lippincott, Williams and Wilkins, Philadelphia, PA. R. T., McDougall, I. L. (1992). Human exposure to 4.0-tesla magnetic
Black, P.-McL., Moriarty, T., Alexander, E., et al. (1997). Development fields in a whole-body scanner. Med. Phys. 19, 1089–1098.
and implementation of intraoperative magnetic resonance imaging and Shellock, F. G., ed. (2001). “Magnetic Resonance Procedures; Health
its neurosurgical applications. Neurosurgery 42, 831–845. Effects and Safety,” CRC Press, Boca Raton, FL.
Bronskill, M. J., and Sprawls, P., eds. (1993). “The Physics of MRI,” Stark, D. D., and Bradley, W. G., Jr. (1999). “Magnetic Resonance Imag-
American Institute of Physics, Woodbury, NY. ing,” 3rd ed., Mosby, St. Louis, MO.
P1: GTM/GRL P2: GNH Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN009N-406 July 18, 2001 23:32
Mass Spectrometry
Kenneth L. Busch
Kennesaw State University
I. Instrumentation
II. Launch Points for the Next Fifty Years
145
P1: GTM/GRL P2: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-406 July 18, 2001 23:32
remain intact, and the supporting instrumentation contin- that “new capabilities catalyze new demands,” and the in-
ues its steady and innovative evolution. But applications in tertwined relationship of modern mass spectrometry, its
biological mass spectrometry have expanded at an extraor- instrumentation, and its market is aptly described.
dinary rate, and fundamental new approaches to the cre-
ation and interpretation of mass spectrometric informa-
A. Sample Introduction Systems
tion are being developed to address new questions. It has
been estimated that a billion mass spectra are recorded The mass analysis and ion detection functions of a mass
daily (Busch, 2000a). The growth and demand for high- spectrometer are completed in a vacuum, usually of 10−5
throughput mass spectrometry suggests that even that ex- to 10−6 torr (a slightly higher pressure prevails in the
traordinary number is an underestimate, if not now, then operation of an ion trap mass spectrometer). The pro-
certainly within the next few years. Classically, mass spec- cesses of electron ionization, chemical ionization, and
tra were interpreted to provide details of molecular struc- matrix assisted laser desorption ionization (MALDI) also
ture revealed through the unimolecular dissociations of operate under vacuum, while the electrospray ionization
the molecular ion. Increasingly, however, mass spectral (ESI) source operates at atmospheric pressure, and ESI-
data in other forms are being recorded and manipulated. generated ions then pass through a set of differential pres-
Mass spectrometry is now used to study interactions be- sure apertures into the mass spectrometer. Mass spectrom-
tween complex molecules in the gas phase to provide eters must operate under vacuum so that the ions move
clues in combinatorial investigations. Mass spectrometric through the instrument in a controlled manner rather than
data is transparently summarized in databases that support being scattered by collisions with residual gas molecules.
repetitive high-throughput proteomics analysis. Multidi- The mean free path of the ion should be larger than the
mensional MS/MS data is stored in arrays examined with dimensions of the mass spectrometer itself. The issues of
chemometric-based routines that mine the data in search vacuum in mass spectrometry are more thoroughly dealt
of answers for complex questions of pattern and structure with elsewhere (Busch, 2000b); the many orders of mag-
that have just now begun to be formulated. nitude difference between instrument operating pressure
This overview covers relevant issues in instrumentation and the atmospheric pressure of the laboratory, or the pres-
for modern mass spectrometry, an overview of techniques sure regime of a column-based separations method, is a
used in both classical and newer areas of mass spectromet- basic design parameter of the sample introduction system.
ric investigation and a preview of growth areas in mass Mass spectrometry is unique among many modern an-
spectrometry research and application. Current applica- alytical methods in that sample molecules are physically
tions are best reviewed by examination of the manuscripts and irretrievably introduced into the instrument. The sam-
published in the current journals of mass spectrometry, ple handling devices must be transport devices that accom-
and review of the papers presented at the annual profes- modate a wide dynamic range of sample quantities, effi-
sional meetings, especially meetings dealing with biolog- ciently transporting all sample molecules from the outside
ical chemistry, and the annual meetings of the American world into the ionization source of the mass spectrometer.
Society for Mass Spectrometry. Given the diversity of sample types, there is a concurrent
diversity of sample introduction systems.
I. INSTRUMENTATION
1. Direct Insertion Probe
Francis W. Aston won the Nobel Prize in Chemistry in The direct insertion probe (or direct probe) is a device to
1923 for his prolific work in the discovery of isotopes introduce small amounts of solid or liquid samples into the
using mass spectrometers that he had constructed in the ionization source of the mass spectrometer without chro-
Cavendish Laboratory. Aston was experienced and skilled matographic separation. The direct probe is appropriate
in the construction and maintenance and operation of in- for use when the sample is known to be pure or of lim-
struments. But, in his 1942 book “Mass Spectra and Iso- ited complexity, or when a chromatographic separation
topes” Aston described his mass spectrometer as an in- is impossible because of sample volatility, or when such
strument that “behaves at times in the most capricious and a separation would be too time-consuming. About 10−5
unaccountable manner.” Where would the field of mass to 10−4 torr of sample pressure is sufficient to record a
spectrometry be today if this were still the case, if, for ex- mass spectrum in an electron ionization or chemical ion-
ample, there had been fewer instrumental developments ization source. This sample pressure can be achieved by
of reliable and capable ionization sources, mass analyz- evaporation of the sample into the vacuum at room (in-
ers, and ion detectors? A well-known adage states that strument ambient) temperature, or by increasing the tem-
“new instrumentation begets new chemistry”; add a codicil perature of the direct insertion probe, which can be heated
P1: GTM/GRL P2: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-406 July 18, 2001 23:32
electrically to 400◦ C. Automation of a direct probe sample different from performing the same specification at levels
introduction system has recently been described (Manura of 0.01%. With increased resolution, chromatography is
and Manura, 2000). With the automated system, samples better able to isolate components present in mixtures at
can be introduced to the mass spectrometer at the rate of lower levels. With increased sensitivity, mass spectrome-
15–20 samples per hour. This is a faster rate of sample try is better able to identify such components. Finally, with
analysis than is normally possible than with chromatogra- increased emphasis on the measurement of analytical in-
phy coupled with mass spectrometry, in which the column formation for both screening and regulatory purposes, the
separation time accounts essentially for all of the time of demand for trace analyses has increased significantly.
the analysis.
a. Gas chromatography. The combination of gas
chromatography with mass spectrometry (GC/MS), itself
2. Chromatographic Columns
the topic of an excellent text by McFadden in 1973, is
The ability of mass spectrometry to identify a sample com- now realized within an integrated, low-cost, widely avail-
pound is maximized when the sample is pure; the combi- able analytical instrument. There is no longer any need to
nation of mass spectrometry with a chromatographic inlet review the history of development of GC/MS instrumen-
system has therefore become a mainstay of instrumental tation, nor the transition from packed-column to capillary
analysis. Chromatography separates sample components column GC. Instead, it is relevant to consider the simple
in time, passes them into the mass spectrometer, and each physical interface between the methods, and then con-
is then characterized via measurement and interpretation straints on the operation of the methods in the combined
of its mass spectrum. The first gas chromatography/mass GC/MS instrument, and finally, to preview the information
spectrometry (GC/MS) instruments of the middle 1970s available to the analyst from the GC/MS combination.
used packed columns, a 5-mm outer diameter, 1-m long In the modern GC/MS instrument, there is no interface
glass tube filled with the packing material (silica or di- per se between the capillary column and the ionization
atomaceous earth) onto which the stationary phase was source of the mass spectrometer. The influx of helium
coated. An enrichment device was needed to separate the carrier gas is of such low amount that it can be accommo-
molecules of the sample from the great excess of the he- dated without difficulty by the vacuum pump connected
lium carrier gas flow. The resolution achieved with such to the source, while still maintaining high vacuum in the
columns was relatively low; peak widths could be tens of ionization source and the mass analyzer. The capillary col-
seconds wide, and the occurrence of overlapping peaks in umn is terminated by direct connection to the ionization
the separation of a complex mixture common. If we were source, with all sample from the column passed directly
still using packed columns for the separation of complex into the source. Termination of the column at the pres-
mixtures, mass spectrometry would be limited to analyz- sure of the mass spectrometer rather than the usual at-
ing mixtures of 10–20 nonoverlapped components, all ex- mospheric pressure of many other GC detectors does not
hibiting the proper thermal stability and volatility for GC change the retention time significantly, since the pressure
characterization. drop occurs in only the last few cm of the column. It is
Clearly, the capabilities of modern chromatographic important that a bonded, stable stationary phase be used
techniques have been vastly improved. Packed column in the capillary column to minimize the amount of column
GC has been replaced by capillary column GC. Similarly, bleed. Column bleed is elution of the stationary phase itself
the large columns of normal-phase liquid chromatogra- into the ionization source of the mass spectrometer. Sen-
phy (LC) are replaced by microcolumn reverse phase LC sitivity is compromised, as is unambiguous identification
columns. Capillary electrophoresis (CE) is an entirely new of compounds due to the presence of extraneous ions in
means of separating small amounts of more complex, and the mass spectrum formed from the stationary phase. Al-
charged sample molecules, and has evolved into several though background subtraction can remove or reduce the
distinct forms with unique capabilities. Mass spectrom- contribution of these bleed ions, this requires additional
etry coupled with different forms of chromatography is spectral processing time.
now applied to the analysis of many mixtures, of higher As eluting peaks from a GC become narrower (as sep-
complexity, and more disparate sample types. aration resolution increases), the need for faster scanning
It is appropriate here to revisit the meaning of “mixture of the mass analyzer becomes more stringent. Even for
analysis.” A mixture becomes increasingly complex as GC peaks only a few seconds wide, at least a few com-
measured component levels decreases, and as the mixture plete mass spectra should be recorded so that they can be
is examined with increasingly sensitive and sophisticated averaged together to form an approximation of the mass
methods. Specification of mixture components with levels spectrum measured with a steady sample concentration in
greater than 1% provides a snapshot of composition vastly the ionization source. Modern mass analyzers, including
P1: GTM/GRL P2: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-406 July 18, 2001 23:32
sector-based analyzers, quadrupole, and ion trap instru- is used rather than GC specifically because the sample
ments provide the requisite fast scanning capabilities. No molecules are relatively nonvolatile, and cannot be evapo-
change in the usual operating conditions of the GC is re- rated without decomposition. Therefore, other ionization
quired, and the mass spectrometer is usually operated at methods have been developed that produce ions directly
near the maximum allowable scan rate consistent with ex- from the solvent (electrospray ionization), or use additives
pected ion signal strength. within the solvent to cause a chemical-ionization-like re-
The usual graphical output of a GC/MS analysis is the action (for example, ammonium acetate buffer was used to
TIC trace, where TIC is an acronym for total ion current. cause protonation in the thermospray ionization source).
The output is graphically similar to the output of a single- Today LC/MS usually involves the use of the ESI
channel GC detector such as a flame-ionization detector source, described in detail in the next section. The com-
or an electron-capture detector. In GC/MS, the TIC trace position and flow of the solvent is constrained within
represents the sum of ion intensities across the scanned limits set by stable ESI source operation. A flow rate of
mass range, a single sum value for each scan number of the 1–10 µl/min of common LC solvents (methanol, ethanol,
GC/MS run. When compounds elute from the GC column, isopropanol, or acetonitrile) generates a stable ESI spray.
the number of ions formed increases, and the summed ESI interface designs used to accommodate higher LC
TIC value increases. The TIC trace is used to determine flow rates include pneumatically assisted electrospray, in
the retention times of compounds. Then a complete mass which a concurrent flow of warmed gas aids in confining
spectrum is assembled by adding together, and then aver- the spray and speeding desolvation, allowing flow rates
aging, mass spectra recorded during scans recorded across of 100–200 microliters/minute. Higher flow rate columns
the eluting peak. Finally, the averaged mass spectrum is can be coupled to an ESI source through a flow splitter.
searched against the library of electron ionization mass The effluent of lower flow rate capillary columns can be
spectra. GC/MS data can also be processed in many other augmented by additional liquid for a stable spray, or one
ways, with the reconstructed ion chromatogram a power- of the newer designs for a nanoelectrospray source can
ful means to extract additional information from the data be interfaced to the capillary LC column. Capillary LC
recorded. The reconstructed ion chromatogram (RIC) is columns with an integral spraying tip can also be used for
a postrun routine that plots intensities of mass-selected direct electrospray ionization at the end of the LC column,
ions vs scan number for a GC/MS run. Ions that “belong” with appropriate connection to a potential source. In all
together in the mass spectrum of an eluting compound of these interfaces, there is a direct connection between
will maximize in intensity at the same time (the retention the column and the ionization source, with minimized op-
time of the compound). Background ions have a steady or portunity for sample loss. Sensitivities are therefore max-
slowly changing intensity. Unresolved GC peaks can be imized, and are mitigated only by the convolution of the
discerned by slightly different intensity maxima in the RIC scanning speeds of the mass analyzer and the widths of
profiles. Data processing in other forms of chromatogra- the LC peaks, as in GC/MS.
phy/mass spectrometry is similarly accomplished, with
the TIC trace providing an overall snapshot of the sepa- c. Capillary electrophoresis. CE was developed in
ration, spectral averaging across the width of the eluting the 1990s, and was soon interfaced to mass spectrome-
peak generating a characteristic mass spectrum of the com- try to meet the need for structure-specific identification
pound, and the use of reconstructed ion chromatograms of eluting compounds. Often, the sensitivity of the mass
to investigate the purity of the mass spectrum. spectrometric analysis is mentioned as a driving force for
CE/MS coupling, but until development of the latest gen-
b. Liquid chromatography. LC/MS requires an int- eration of electrospray ionization sources, laser-induced
erface and an ionization method that accommodates the fluorescence routinely provided lower limits of detection
polar solvent that carries the separated sample mix- than did mass spectrometry. The potential-driven move-
ture components through the (usually) reverse-phase col- ment of ions in solution is the basis for CE separations.
umn. Just as GC evolved from larger packed columns to The rate of movement of ions is determined by the sum of
smaller, higher resolution capillary columns, LC also pro- electrophoretic and electroosmotic flow. Both neutral and
gressed from columns with large flow rates of solvent to charged compounds move through the column, migrat-
smaller columns that operate with solvent flow rates of ing at different rates, and maintaining a high separations
a few milliliters per minute, and eventually to microbore resolution due to the shape of the flow gradient in the
columns with flow rates of 20–100 µl/min. Electron and small column. The flow rate through a CE column is a low
chemical ionization (vide infra) would require that the 1–2 microliters/minute. The sample loading on the CE
solvent be completely removed, and that stable gas-phase column is also low, with perhaps a few ng of sample at
neutral molecules of the sample be formed. However, LC most available for detection.
P1: GTM/GRL P2: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-406 July 18, 2001 23:32
The movement of sample through the CE column is ization process is the direct result of the interaction of an
driven by a potential difference between the entrance and energetic electron with the electrons in the molecule of
the exit of the CE column; this applied potential must interest. The electrons are emitted from a metal (usually
be taken into account when designing an interface that rhenium) filament through which 3–4 Amperes of current
directs the effluent into the ionization source of the mass is passed. This current heats the filament to about 2000◦ C;
spectrometer. CE/MS is exclusively coupled with the ESI electrons are released from the metal and accelerated into
source, which has its own requirements for imposition of a the source. The classic “70 eV” electron ionization mass
high potential on the needle from which the spray emerges. spectrum of an organic compound is obtained when the
Several early designs were explored for integration of both potential difference maintained between the filament and
solvent flow and imposed potentials. Integrated designs the source block is 70 V, with the block being maintained at
now use the exit of the CE column itself as the spraying a more positive potential. Variations in the measured mass
tip for electrospray. A relevant point in all designs is the spectra and in the ionization cross sections of organic com-
fact that the CE column exits at atmospheric pressure, pounds with changes in the electron energy were studied
which preserves the pressure profile within the column, early in the development of electron ionization mass spec-
and maintains the inherently high CE resolution. trometry. A value of 70 V was chosen so that mass spectra
recorded at this electron energy did not vary greatly with
small changes in the electron energy, and the sensitivity
B. Ionization Sources
(number of ions produced per amount of sample intro-
Mass analyzers cannot manipulate neutral molecules. A duced into the course) is also essentially constant about
positive or negative charge is necessary for interaction of this value.
the ion with magnetic and electric fields. The differing The incident electrons must have an energy greater than
magnitudes of responses for ions of different masses is the ionization energy of the target gas molecule M, defined
the basis for their separation in the mass analyzer. Most as the energy required to remove the electron held most
samples exist initially as neutral molecules, and gas chro- weakly within the molecule. The electron ionization pro-
matography and liquid chromatography are used for the cess can be written for the gas-phase sample molecule M:
most part to separate mixtures of components that are q
M (gas) + e−
filament −→ M
+
+ e− −
filament + emolecule .
neutral in the gas phase, and neutral in the liquid used
as the LC solvent. The ionization source converts these Electrons in molecular orbitals are moving at velocities
neutral molecules into ions, or extracts ions from solu- of about 108 m/sec. As the very fast filament electron ap-
tion, and passes the ions into the mass analyzer of the proaches the molecule, it causes the release of one of the
instrument. The physical means to do so depends on the slower molecular electrons, forming a positively charged
q
initial form of the sample. Electron and chemical ioniza- molecular ion. The molecular ion M+ (the superscripted
tion are the older and more traditional ionization meth- dot denotes an unpaired odd electron) may subsequently
ods, both widely used with GC. Liquid chromatography dissociate, since the ionization imparts more energy to the
is used predominantly with electrospray ionization, as is molecule than that required for ionization alone. The ex-
capillary electrophoresis. MALDI is an ionization method cess energy can cause the dissociation of the molecular
q
that creates ions directly from a solid mixture deposited on ion M+ , or it can be retained in the ion as excess internal
a surface (consistent with planar chromatography, but not energy. Since an electron is far too light to transfer kinetic
directly with the column-based chromatographic meth- energy to the sample molecule in a collisional process, the
ods that dominate modern mass spectrometry). MALDI process of electron ionization involves electronic excita-
is included here since it is one of the rapidly growing ap- tion of M. The assumption that the atoms do not move as
plications areas in biological mass spectrometry, and its the transition to an excited electronic energy state occurs
use may revitalize some forms of planar chromatography, is known as the Franck–Condon principle. The assump-
q
including forms of planar gel electrophoresis and affinity tion states, in effect, that the molecular ion M+ retains the
chromatography. original structure of the molecule M, at least at short times
after its formation. If dissociations of the molecular ion
are prompt, therefore, we can assume that the dissociations
1. Electron Ionization
represent those of the original molecule and not a struc-
Electron ionization was the first ionization method de- turally reorganized isomer. Some fraction of the molecular
veloped for mass spectrometry, and it remains the most ions formed will be stable enough to pass into the mass
widely used. The most extensive mass spectral libraries analyzer; their measured m/z ratio is a direct indication of
assembled are those of electron ionization mass spectra the molecular mass of the sample molecule itself. Ratio-
recorded under a “standard” set of conditions. The ion- nalization of the processes that lead to the fragmentation
P1: GTM/GRL P2: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-406 July 18, 2001 23:32
processes provides clues to the original structure of the it encounters will most likely be a methane molecule. The
molecule. A balance must be attained between fragmenta- ion/molecule reaction between the two involves a pro-
q
tion extensive enough to provide reasonable clues to struc- ton transfer to create CH+ 5 and CH3 . Several other sub-
ture of the neutral molecule, and a relative intensity of the sequent reactions occur, and the final distribution of ions
molecular ion large enough so that the molecular ion can depends explicitly on the source temperature and pres-
be identified, and the mass of the sample molecule de- sure, but the primary reactant ion is usually CH+ 5 . This
termined. Determination of molecular mass is a physical ion acts as a strong gas-phase acid that protonates any-
measurement, in contrast to the insight and intuition of thing more basic than methane; neutral sample molecules
q
interpretation. The appearance of both M+ and fragment are sufficiently basic to accept a proton to form the proto-
ions derived from the molecular ion—the deduction of nated molecule. The mass of the (M + H)+ ion reveals the
both molecular mass and molecular structure—is the core molecular mass, and the protonated molecule then frag-
use that has supported the growth of electron ionization ments in accordance with the amount of internal energy
mass spectrometry. contained within it. Methane is not the only reagent gas
that forms a reagent ion that transfers a proton to the gas-
phase sample molecule. Isobutane introduced into the CI
2. Chemical Ionization
source at a pressure of about 1 torr generates C4 H+ 9 as an
In electron ionization, an encounter with an energetic elec- acidic reagent ion, which also transfers a proton to the neu-
tron causes the ejection of a single electron from a gas- tral sample molecule, but gives it less energy so that the
phase sample molecule M to form the odd-electron molec- protonated molecule fragments less. Ammonia reacts un-
q q
ular ion M+ . If too much energy is deposited into M+ der CI conditions to create NH+ 4 . The ammonium ion can
during the ionization process, or if the molecule is prone transfer a proton to the sample molecule, or it can trans-
to dissociate, fragment ions may be seen in the mass spec- fer the intact ammonium group to create (M + NH4 )+ . A
q
trum, but the M+ may be reduced to such a low inten- simple rule in CI mass spectrometry is use methane to ob-
sity that it is indistinguishable above the background sig- tain the protonated molecule and some fragmentation, and
nal level. Without the molecular ion, the determination of then use isobutane when you want to minimize the frag-
molecular mass is difficult. Chemical ionization (CI) was mentation and concentrate ion current in the protonated
developed to overcome this difficulty and provide molec- molecule (as might be required when low limits of detec-
ular ions for compounds used for determination of the tion are sought). Since the ammonium ion will transfer
molecular mass. its proton only to compounds more basic than ammonia
Chemical ionization involves a collision and reaction itself, selectivity in ionization can be achieved.
between an ion and a molecule, both in the gas phase. The
ion is called the reagent ion and the molecule is the neutral
3. Electrospray Ionization
sample molecule. Since the process is an ion/molecule re-
action, parameters such as source temperature and source Among the ionization methods discussed here, ESI is
pressure, sample pressure and reagent gas pressure, and unique in that it generates ions directly from within a so-
the presence of impurities in the gas stream that enters lution that is sprayed from a fine needle at atmospheric
the source must be carefully controlled. The CI source pressure. At the core of the ionization source is a stain-
is a variation of the standard EI source, with modifica- less steel capillary tube that carries solvent (effluent from
tions required to achieve a higher source pressure (about a LC column, typically) at a flow rate of 2–5 µl/min. A
1 torr) while keeping the mass analyzer pressure within potential difference of 3000–4000 V is maintained be-
acceptable limits. The source filament emits electrons, but tween the needle and a counter electrode, which can be
these are now accelerated to several hundred volts (250– a wall of the source, or a skimmer cone with an aperture
500 eV is typical) compared to the standard 70 eV of that passes the ions into the mass spectrometer. A spray is
an EI source. This higher energy allows the electrons to generated at the tip by the solvent flow emerging at atmo-
penetrate through the higher gas pressure in the source. spheric pressure, and the potential difference ensures that
Since the pressure of 1 torr is mostly the reagent gas (of- the droplets emerging from the needle are charged, aiding
ten methane), the electron emitted from the filament is in their dispersal. A supplemental flow of gas is also some-
likely to encounter a methane molecule. When it does, an times used to aid in desolvation of the droplets. The ions
electron ionization process occurs, viz. that are within the bulk solvent, or more accurately, those
q ions that are formed in the droplet as the last evaporation
CH4 + e− + − −
filament −→ CH4 + efilament + emethane . of neutral solvent molecules occurs, are focused by ion
q
The CH+ 4 ion does not travel far before it encounters a neu- lenses, and sampled through apertures, to be passed into
tral gas molecule, and at 1 torr of methane, the molecule the mass spectrometer.
P1: GTM/GRL P2: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-406 July 18, 2001 23:32
As the solvent emerges from the charged capillary, it mass spectra, and even spatially resolved mass spectral
first forms a cone (called a Taylor cone) that results as maps of a surface. Organic and biological molecules de-
the droplet adopts a shape to mimimize Coulombic re- compose under such intense irradiation, and despite much
pulsions between the charges on the surface of the liq- early work in the area, suitable conditions for successful
uid. The initially formed cone then dissociates into small direct analysis were not found. In MALDI, a matrix is used
droplets, each now isolated in the gas phase (and still to moderate on a molecular level the energy deposited at
at atmospheric pressure), and each carrying an excess the surface, and promote ionization of the sample molecule
charge on its surface. Desolvation involves the loss of without decomposition or excessive dissociation. In prac-
neutral solvent molecules from the droplet, and proceeds tice, the sample is mixed in solution with a large excess
rapidly at atmospheric pressure. As the droplet decreases of the matrix (typically a smaller molecule chosen for
in size, the charge density increases until an instabil- high UV absorbance, appropriate volatility, and ability to
ity limit is reached, and the droplet dissociates into still promote ionization), and an aliquot of a few microliters
smaller highly charged droplets. Residual solvent quickly volume is deposited on an inert surface to co-crystallize.
evaporates, leaving only the charged ions themselves to be The surface is irradiated with short (10–20 nsec) pulses
transferred into the mass spectrometer. The ionic popula- of laser light at 337 nm; a tight focus on the surface is
tion within a microdroplet, and then a nanodroplet, is not not necessary. The irradiance power achieved is approxi-
known explicitly. The charge imposed on the droplet as it mately 106 W/cm2 .
departs the needle must be carried by a chemical species in The ions observed in the MALDI mass spectrum (they
the solution. If the droplet is positively charged, the charge have been termed the “survivor” ions) are the minority
carriers will be predominantly protons. As the droplet de- within all the species that leave the surface as a result of the
creases in size, the “pH” rises exponentially (pH is not laser beam irradiation, comprising perhaps 0.01% of the
an accurate description of the situation since there is no desorbed/ablated/expelled species (Zenobi and Knochen-
equilibrium). As the droplet decreases in size, the protons muss, 1999). Since the presence of the matrix is requisite
are forced to ionize the sample molecules. Protonation, for MALDI, investigators quite logically seek to establish
and in fact multiple protonation, is commonly observed. the role of the matrix by changes to its chemical nature,
Key aspects of the ESI process are the formation of with consequent changes to its intrinsic proton affinity,
multiple charged ions of the sample molecule, and mini- its absorbance spectrum, its ionization potential, and its
mum fragmentation of these molecular ions. Positive ions crystal structure. While the analyte-to-matrix ratio may
of the general form (M + nH)n+ are formed by multiple be controlled on a macroscopic level, it varies across a
protonation of larger biomolecules (molecular mass is des- broad range on the microscopic level, and will also vary
ignated by M) such as peptides and proteins. One effect spot to spot. The incident laser beam interrogates a surface
of multiple charging is to bring multiply charged higher on which crystals of different size and different properties
mass molecules within the mass range of commonly used are dispersed. The microscopic morphological effects that
mass spectrometers, since the mass analysis is actually a result from a change in gross sample preparation proce-
m/z measurement. Further, since M is constant between dure are uncharacterized. The simple fact of the matter is
the series of peaks observed as adjacent multiply charged that MALDI works amazingly well under a wide variety
ions, the multiple measurements of mass of these ions of conditions, and it is perhaps not necessary to control
constitute a series of simultaneous equations that can be all of them rigidly. It has become recognized that in ad-
solved to determine M, the molecular mass, to a pre- dition to the direct desorption of species from the laser-
cision of ±0.005%. This mass measurement capability irradiated surface, secondary ionization processes occur
was not an improvement on exiting methodology, but was in the selvedge above the surface. This is a reiteration of
a totally new capability, without precedent and without the concept of the selvedge introduced to explain reactions
competition from other analytical methods. This unique- in secondary ion mass spectrometry and desorption ion-
ness makes accurate prediction of future applications ization techniques in general (Cooks and Busch, 1983).
and developments in mass spectrometry challenging and Reactions that transfer, preserve, or dissipate charge de-
exciting. termine what ions finally survive to form the MALDI mass
spectrum. Proton transfer reactions that may involve the
matrix form protonated molecules. Alkali cations at the
4. Matrix-Assisted Laser Desorption Ionization
surface as impurities can form cationized species. In a pro-
A pulsed laser can be used to deposit a very large amount cess reminiscent of ESI, multiply charged species can be
of energy into a small spot in a short time. This energy formed, but the high population of free electrons, and the
causes ionization, and if the desired ions are atomic ions, high mobility of electrons within the selvedge, promotes
then laser desorption proves a useful method for creating reduction processes that reform the singly charged ions.
P1: GTM/GRL P2: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-406 July 18, 2001 23:32
Once the ions are formed through a process of electron can be accommodated in other mass analyzers. This very
ionization, they are trapped within the QIT by application low pressure can be reached, but engenders additional dif-
of appropriate voltages (DC) and rf signals on the elec- ficulty in creating and transporting ions from the outside
trodes. Again, in a simple case, the DC voltages can be world into the cell. The interfacing of an ESI source to a
at ground potential. A rf drive potential of about 1 MHz FTMS has been accomplished, but the practical difficulty
frequency is applied to the ring. Ions within a broad mass of transporting ions efficiently through a pressure differ-
range trace stable orbits near the center of the QIT. In most ential of 10–12 orders of magnitude is considerable. The
instruments, helium gas at a pressure of approximately advantages of high-resolution mass spectral data are suf-
10−3 torr aids in keeping the ions in orbits near the center ficient impetus for development of this instrument and its
of the trap. As the amplitude of the rf drive potential is expanding applications. The presence of multiple charges
increased, the ion motions progressively change. Eventu- on an ESI-generated ion increases the mass range of FTMS
ally, ions develop an unstable trajectory along the z-axis of by, in effect, making the ion easier to manipulate in the
symmetry (see figure), and pass through the ion exit holes cell and easier to detect. In that the ion is not destroyed in
in the end-cap electrode to be detected by the electron its detection, the sensitivity of FTMS is high. The limit-
multiplier. Appropriate potentials are used so that ions are ing steps are maintaining charge on the ions (larger ions
ejected from their orbits in mass order, and thus the device have a tendency to relinquish their charge unimolecularly)
scans across the mass range of the mass spectrum. and maintaining those ions within a stable and coherent
orbit. With enough time, the signal frequency can be de-
termined for a group of only 100–1000 ions of the same
5. Fourier Transform Mass Spectrometers
mass, which may correspond to only a very small amount
The basis of high resolution Fourier Transform mass spec- of sample, or a very low abundance ion in a mass spec-
trometry (FTMS) is the measurement of the frequency of trum. In fact, the presence of too many ions complicates
an ion orbiting in a static magnetic field. The ions are not the measurement through introduction of secondary space
destroyed by this measurement of frequency. Instead, the charge effects that distort the measured frequency. Capa-
ions orbit continuously within the confining cell of the bilities of FTMS are discussed again in the final section
FTMS instrument (held there by a combination of poten- of this overview on launch points for the next fifty years.
tials applied to a cubic cell and a static magnetic field) until
removed by application of an external electrical pulse, or D. Detectors
scattered out of a stable orbit by collisions with neutral
With the exception of the Fourier transform mass spec-
residual gas molecules. The key to an accurate measure-
trometer described above, ion detection in other mass
ment of frequency, and therefore an accurate determina-
spectrometers is the destructive event in the sequence.
tion of mass, is maintenance of a coherent ion orbit. With
Each individual ion carries such a small charge and is
extended measurement time, the impact of random errors
of such low absolute mass that direct detection is difficult
in the determined frequency value are reduced (the usual
(but not impossible). Considerable amplification of the ion
measurement statistic in which random noise decreases
charge or the ion mass is necessary to make mass spec-
with the square of the number of measurements). The keys
trometry practicable. Electron multiplier detectors, known
to the maintenance of ion orbit are the homogeneity of the
since the 1950s, provide the requisite amplification of the
static magnetic field, the symmetry of the trapping poten-
charge into an easily manipulated current.
tials generated within the ion confinement cell, and the
base pressure within the FTMS instrument. At low base
1. Electron Multipliers
pressures within the FTMS instrument, the ion orbits are
stable for periods of seconds and longer. Extraordinary The electron multiplier detector is a transducer that con-
resolutions have been achieved not only for simple atomic verts the impact of ions on its front surface into an am-
ions but also recently for complex ions derived from com- plified electron current then sampled by modern digital
plex organic and biomolecules, as well as for ions from electronics. The sensitivity of mass spectrometric analysis
polymers and organometallic compounds. The maximum is supported by the high gain (as high as a millionfold)
resolution of 100,000 reached by sector instruments has provided by modern electron multiplier. This impressive
been surpassed easily by a resolution of 1,000,000 readily gain is achieved by a combination of the electron emissive
achievable in FTMS. properties of the active surface of the multiplier and the po-
Perhaps the most significant practical impediment to tential difference maintained along its length. The active
the use of FTMS has been the need to maintain a low surface of the modern Channeltron version of the electron
pressure in the cell where the ions orbit. A pressure of multiplier consists of emissive layers of silicon dioxide
10−9 torr is desirable, as compared to the 10−6 torr that overlying a conductive layer of lead oxide supported by
P1: GTM/GRL P2: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-406 July 18, 2001 23:32
ions are collected. Once these positive ions are accelerated tributes such as exact mass and measuring their intensity
into the body of the electron multiplier, the electron release as formed from a dissociating molecule. In interpreting a
and amplification process occurs as described above. For mass spectrum, we explore issues of chemical reactivity,
the efficient detection of high-mass ions, the conversion often based on kinetics, but expanding more recently into
dynode may be held at a very high potential to acceler- thermochemical arenas. This chemical focus represents
ate larger ions to a velocity sufficient to release on impact the fifty years of modern mass spectrometry in contrast to
a large number of lower-mass positive ions that are then Aston’s world of isotopic discovery that dates from 1919.
detected by the electron multiplier. Prognosticators look to the next fifty years, lest the con-
tent of an overview such as this become archaic even as it
2. Array Detectors appears.
The electron multiplier is a single channel device. The
mass analyzer selects the ions by mass, and the ion cur- A. Exact Mass Measurement
rent is generated through amplification as described above,
Since the exact masses of individual atomic ions are known
measured, and recorded as the value for the ions at that
with high accuracy, the exact mass of an ion of a given
mass. Since the mass analyzer has to scan through the
empirical formula (a known combination of atoms) is also
mass range, at any given instant, all ions except those
known with high accuracy by simple summation. In mass
of the one mass passing through to the detector are lost.
spectrometry, measurement of the exact mass of an ion is
The Mattauch–Herzog geometry of double focusing mass
used to deduce information about the empirical formula
spectrometer used a focal plane detector in which ions of
(not the molecular formula) of the ion. If the total number
different masses were brought simultaneously into focus
of possible combinations is small, and the measurement of
at a planar detector that most often was photographic film.
the exact mass sufficiently accurate, the measurement can
The sensitivity of this integrating detector was high. With
be used to derive an ion empirical formula. The traditional
newer developments in multichannel or multipoint col-
exact mass measurement is usually limited to the more
lectors and their associated electronics, photographic film
abundant ions observed in the mass spectrum, since higher
has been replaced with array detectors, which are assem-
mass resolution in sector instruments was achieved at the
blages of small devices that act individually much like the
cost of lower ion signal.
electron multiplier described above.
High-performance sector instruments could provide
The term array is used to describe an assemblage of
mass resolutions of up to 100,000, but the typical daily op-
small single point ion detectors (sometimes called ele-
eration was a resolution of a few thousand. Excursions to
ments) arrayed in a plane. Each of these elements acts as
higher resolutions required concerted efforts with a clean
an ion current amplifier. Each element acts as a detector
and stable instrument, and the availability of a relatively
for the ion mass that is directed onto that particular space,
large amount of sample. As described above, the basis of
and that particular element. All ions of all masses are de-
high resolution FTMS is the measurement of the frequency
tected simultaneously, and the array detector acts as an
of an ion orbiting in a static magnetic field. The ions are
integrating detector (ignoring the electronics of readout
not destroyed by this measurement; the measurement can
and measurement, and transfer of values into the data sys-
be completed on only a few hundred ions kept within the
tem). Clearly the array detector will provide an increase
cell for an extended period.
in sensitivity over a single-channel detector. It is therefore
The ability to routinely make exact mass measure-
useful in trace analysis, or in analyses in which a fraction
ments as exemplified by FTMS is not a simple exten-
of the mass spectrum (perhaps across the molecular ion
sion of the use of such values to derive empirical formula.
region) should be recorded without scanning that cover
There are two related areas in which the mass measure-
a wide mass range. More recently, these detectors have
ment data is used in situations in which other information
become useful in mass spectrometers in which the pro-
provides additional restraints on possible empirical for-
duction of ions in the source is discontinuous, as in a laser
mulas, and a synergistic extension of abilities. Rodgers
desorption ionization source. Time variabilities in ion pro-
et al. (2000) show that stable isotope incorporation into
duction are muted by the integrating nature of the detector.
selected biomolecules increases the upper mass limit at
which accurate mass measurement can provide an empir-
II. LAUNCH POINTS FOR THE NEXT ical formula composition for the ion. Simply, the nom-
FIFTY YEARS inal mass difference between a “natural abundance ion”
and the corresponding 13 C-enriched (99% enrichment) ion
The mass spectrometer represents our microscope into of the same form immediately yields the number of car-
the world of individual ions, recording their physical at- bon atoms in the molecule. Once that value is known, the
P1: GTM/GRL P2: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-406 July 18, 2001 23:32
number of possible empirical formulas that can sum to the we now follow the same process with MS/MS data. At
accurately measured ion mass is greatly reduced. The ap- the same time, new instrumental capabilities (specifically
proach is general to any particular isotopic incorporation. with ion traps and Fourier transform mass spectrometers)
It is not a development in instrumental capability; it is a mean that multiple-stage MS/MS can be completed, and
development in which instrumental capability is synergis- the sequential induced dissociations of a mass-selected
tically coupled with reasoned use of other information. parent ion into product ions can be measured. It is easy to
The same coupling is evident in the ability of FTMS to record and easy to store multidimensional MS/MS data.
provide accurate mass measurements for ions generated What does this data mean? It is most certainly a series of
in multistep dissociations in MS/MS. It is not necessary insights into the structure and reactivity of the ions. But in
that the value of exact mass measurement differentiate a larger sense, it is also a collection of a pattern of informa-
between all possible ion empirical formulas, but only be- tion that can reach a complexity sufficient that the pattern
tween possible empirical formulas for product ions that itself becomes a specific identifier for a particular molec-
can be formed from the mass-selected parent ion for which ular structure and identity. This concept is not unfamiliar
the empirical formula is already known. As a simple ex- to users of nuclear magnetic resonance (NMR) data. Tar-
ample, if the parent ion contains no sulfur atoms, then the geted sifting of a complex database of multidimensional
product ion cannot either. Further, “weak” points within a MS/MS has not yet impacted the analytical community,
molecular structure, and therefore probable sites of cleav- and the clear advantages for unambiguous sample identi-
age, are usually apparent to experienced analysts. With fication have not yet been reaped. It is merely a matter of
the number of possibilities thus even further reduced, the time, and closer to five years than fifty.
value of the exact mass measurement is amplified further.
The possibilities narrow as the number of stages of induced
dissociation increases, relaxing the need for accurate mass SEE ALSO THE FOLLOWING ARTICLES
measurement as the ion signal grows smaller.
ANALYTICAL CHEMISTRY • DATABASES • GAS CHRO-
B. Multidimensional MS/MS MATOGRAPHY • ION KINETICS AND ENERGETICS • LIQUID
CHROMATOGRAPHY • MASS SPECTROMETRY IN FOREN-
The central analytical role of mass spectrometry continues SIC SCIENCE
to be twofold: what is the sample and how much of it
is there? Certainly there are extraordinary new insights
into chemical behavior between complex species in BIBLIOGRAPHY
solution and in the gas phase that will be part of the
next fifty years of mass spectrometry, made possible Aston, K. W. (1942). “Mass Spectra and Isotopes,” Arnold, London.
by advances in ESI and MALDI. Forecasting in the Busch, K. L. (2000a). Spectroscopy 15(11), 30–39.
new realms of biological mass spectrometry is difficult. Busch, K. L. (2000b). Spectroscopy 15(9), 22–25.
But in looking exclusively to these new and exciting Cooks, R. G., and Busch, K. L. (1983). Int. J. Mass Spectrom. Ion Phys.
areas, the implications of modern mass spectrometry 53, 323.
Mamyrin, B. A. (1994). Int. J. Mass Spectrom. Ion Phys. 131, 1–19.
for the “tried and true” are often overlooked. In the Manura, J. J., and Manura, D. J. (2000). American Laboratory 33(3),
introduction, it was stated that a billion mass spectra are 40–53.
recorded daily around the world. Most certainly, the vast McFadden, W. (1973). “Techniques of Combined Gas Chromatography/
majority of these are never examined by human hand and Mass Spectroscopy: Applications in Organic Analysis,” John Wiley
mind, neither assessed nor interpreted. Some decision Interscience, New York.
Rodgers, R. P., Blumer, E. N., Hendrickson, C. L., and Marshall, A. G.
of some value is reached, perhaps automatically, and (2000). J. Amer. Soc. Mass Spectrom. 11, 835–840.
the mass spectra are then archived. With tremendous Zenobi, R., and Knochenmuss, R. (1999). Mass Spectrom. Rev. 17(5),
advancements in computing and data storage capabilities, 337–366.
P1: GSR/GLE P2: FXY Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
Mass Spectrometry
in Forensic Science
Jan Schuberth
National Laboratory of Forensic Chemistry,
University Hospital (Emeritus)
I. Data Generation
II. Data Evaluation
159
P1: GSR/GLE P2: FXY Final Pages
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
Quadrupole A type of mass filter. It has an open space Mass spectrometry is, in short, a method by which one
between four rods where an oscillating electric field is generates charged molecules and molecular fragments and
generated to direct the flow of ions made up by charged then measures the mass of each or, more rightly, their
fragments, formed from an analyte, and accelerated mass/charge ratio. The uncharged gas molecules, which
into the field. Only those ions with mass/charge ratios are introduced into the mass spectrometer, move randomly
that suit the electric potentials of the oscillating field around in space, so to control their motions they have to
pass the rods to become recorded. be ionized (i.e., a charge is added to the molecules). Also,
energy in excess of the ionization must be supplied to break
some of the covalent bonds holding a molecule together.
MASS SPECTROMETRY is a main tool in forensic This is to split it into specific fragments that may be used
chemistry for the analysis of chemicals in samples taken to obtain structural information about the molecule or to
to reveal a possible crime. Its main value rests on the fact unfold its characteristic mass spectrometric fingerprint.
that mass spectrometry allows for an unbiased search with An outline of the process is shown in Fig. 1. Our dis-
high sensitivity and specificity for a variety of substances cussion will include a general description of the hardware
with wide ranges of different physicochemical properties. to generate the MS raw data (the introduction of the sam-
The principles for making a substance suitable for mass ple at atmospheric pressure into the mass spectrometer,
spectrometric analysis are presented. These include the which is under vacuum), the formation of the mass frag-
sample inlet from atmospheric pressure to the low pressure ments and their separation from one another, focusing, and
in the apparatus, the conversion of the sample molecules detection. The raw data thus obtained require some form
to ionized particles, their separation from one another in a of evaluation to become intelligible, a process that calls
magnetic or oscillating electric field, and their detection. In for automatic and/or manual computer work. A few exam-
forensic chemistry the mass spectrometer is most often fo- ples of real-life forensic studies will help explain common
cused on a wide range of different mass fragments, and the approaches used in such a pursuit.
data gotten from a test, therefore, need refinement to be-
come intelligible. Examples from real-life forensic work
I. DATA GENERATION
in toxicology, arson analysis, and environmental forensics
have been chosen to illustrate such processes.
A. Sample Inlet
Forensic science plays a key role for law-enforcing bod-
ies. Its main task is to supply physical evidence pertaining The inlet system is used to introduce the sample into the
to a suspected criminal act. The forensic scientist searches mass spectrometer, to convert it into the gas phase, and
materials collected by the police for evidence a person may to reduce its pressure before ionization. Forensic sam-
have left at the site of a crime or brought with him. The ples are often impure, so the analytes, have to be sepa-
test matter is often made up of agents suitable for chemical rated from the matrix before being inserted into the mass
analysis, such as drugs seized by the police or present in spectrometer. The inlet system is most often an interface
a body fluid, residues of arson accelerants or explosives, between a chromatographic device and the mass spec-
poisons in a dead body, or residues of illegally deposited trometer. By this approach, the analytes are separated
spill oil, to name a few examples. from one another and from the contaminants by either gas
Two distinct approaches can be used for the chemical chromatography (GC) or high-performance liquid chro-
exam. In some surveys the analyst, without having to iden- matography (HPLC), and the isolated compounds in the
tify each single compound, may just compare the pattern of effluents from the column flow directly into the mass
a group of substances in the sample with that of a reference spectrometer.
product that has some bearing on a crime. In some types The combination of GC and MS (GC/MS) is often used
of suspected crimes (e.g., intoxication or drug trafficking), in forensic science. Since the GC separation today is done
the individual substances in a sample must be pinpointed. in a capillary, and not in a packed column as it used to
Regardless of the analytical aim, a method must be used be, introducing the analytes into the mass spectrometer
that will hold up to scrutiny in a court of law. Another has become rather simplified. The flow rates are much
problem the chemist is faced with is that he does not know lower in a capillary than in a packed column, and the
which compounds to look for amid the wide range of pos- entire volume of the effluents can be let into the mass
sible ones with different physicochemical properties. By spectrometer without any losses simply by pushing the
virtue of its high sensitivity and specificity, as well as its loose end of the GC capillary in close to the ion source of
ability to reveal the general unknown, mass spectrometry the mass spectrometer.
(MS) is the chemist’s primary analytical tool for solving Even though HPLC in combination with mass spec-
many forensic problems. trometry (HPLC/MS) is used less frequently than GC/MS,
P1: GSR/GLE P2: FXY Final Pages
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
it has recently gained in popularity. A reason for this de- and breaks some of the chemical bonds of the molecule by
lay in its use in the forensic science arena is the technical putting energy into it. In this process, the analyte molecule
difficulties of interfacing a mass spectrometer with a liq- (M) is impinged with a beam of energetic electrons (e− ),
uid chromatograph. The eluate from the HPLC column is a process that results in the formation of a radical cation
+
a liquid, which expands when it gasifies to become ready with an odd number of electrons (M ). As seen in the fig-
✉
for MS analysis, a situation that adds an extra burden on ure, the end result yields an analyte molecule that has
the vacuum system for reducing the pressure in the mass lost an electron and become a particle with a positive
spectrometer. Moreover, the effluents often carry polar, charge, thus making it feasible for separation in a mag-
heat-labile substances that may taint the ion source when netic or oscillating electric field based on its mass and
they are vaporized. number of charges. The energy of the electrons is gen-
The thermo spray is one of the most often used erally set at 70 eV, an optional value chosen because it
HPLC/MS interfaces. In this system the HPLC effluents, is high enough to exceed the ionization energy at about
upon entering the mass spectrometer, are forced through 10 eV and to split the molecule by breaking its chemical
a pinhole leak to form a jet of liquids which is heated to bonds.
gasify the solvents in the aerosol. The mist of droplets For the identification of an analyte it is often useful to
in the gas phase, which becomes void of solvents, car- know its molecular mass. This piece of information, how-
ries, along with a charge from any ions initially present in ever, is generally not gained by EI; the process often re-
the solution, the less volatile analytes directly into the sults in a complex fragmentation of the molecule, making
ion source of the mass spectrometer to become fit for the mass deduction of the master substance difficult. To
analysis. overcome this problem a milder ionization method must
be used, such as chemical ionization (CI). This process is
done in an apparatus similar to that for EI, but unlike the
B. Ion Generation
EI device, which holds the same low pressure inside as
The mass fragments of a target substance are generated outside the ionization cell, CI is run at a higher pressure
in the ion source of the mass spectrometer. An outline of within the chamber than outside it. Letting the analytes
the simplest ionization method, electron impact (EI), is and electrons into the ionization chamber through small
shown in Fig. 2. This step puts a charge on the molecule holes allows this.
P1: GSR/GLE P2: FXY Final Pages
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
FIGURE 2 Fragmentation of molecules by electron impact. The analyte molecule (M) is bombarded with a beam
+
of energetic electrons (e− ), whereby it loses an electron to become a positively charged particle (M ), feasible for
✉
As shown schematically in Fig. 3 a proton donator methane molecules to generate a stable population of sec-
formed by EI of the reagent gas (e.g., methane) collides ondary ions. These in turn react with the analyte gas, a
in a gas phase with the target substance and then gives process that results in a gas-phase acid–base proton trans-
up its proton to the analyte. The reactions occur in two fer from the secondary ions to the analytes. Since these
steps. In the first one, the primary ions are formed from are present in low concentrations as compared with the
the methane fragments. At a sufficiently high pressure in concentration of methane, the analytes only seldom col-
the ionization cell, the primary ions collide with neutral lide with the electrons, a situation that results in less
P1: GSR/GLE P2: FXY Final Pages
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
FIGURE 3 Fragmentation of molecules by chemical ionization. Gas of the analyte molecules (M) and reagent (CH4 )
+ ✉
are let into the ionization chamber, where CH4 , bombarded with a beam of electrons, forms primary ions CH4 . At a
+
sufficiently high pressure, these collide with CH4 molecules to form the secondary ions (CH5 ). These in turn collide
with the analyte molecule (M), which then is converted to the protonated molecule (M + H)+ , suitable for separation
in a magnetic or oscillating electric field.
fragmentation by CI than by EI. The protonated molecule, ment. The second type, or the mass-selective instability-
which appears as the most abundant fragment at CI, meets operating machine, is known as the ion trap detector (ITD).
the demands for separation in a mass analyzer.
1. Magnetic Sector Instrument
C. Mass Analyzer The working principle for separating charged analyte frag-
ments or molecules in the magnetic sector instrument is
The analyte molecules, which have become ionized, may
shown in Fig. 4. When thrown out from the ion source
now be moved apart from one another in a magnetic or os-
and accelerated, these tend to adjust to orbital movements
cillating electric field. Even though the object is to analyze
when allowed to travel in a magnetic field aimed perpen-
the fragments based on their masses, it is the mass/charge
dicular to the flight of the ions. The trajectory of a fragment
(m/z) ratio that forms the basis for the separation process.
with a given m/z value, thus, depends on the acceleration
This means that fragments with m/z = 200/2 and m/z =
voltage and the strength of the magnetic field. This means
100/1 have the same trajectories and may not be distin-
that, at a given acceleration, only those fragments with m/z
guished from each other. Fragments with two charges are,
values that balance the centripetal force of the magnetic
however, rare, and in practice it is the mass of a fragment
field will travel through the flight tube and reach the detec-
that in general is regarded as the regulating factor of its
tor, whereas the unstable ions will adhere to the path wall.
movement in a magnetic or oscillating electric field. Ap-
pliances can have two fundamentally different operating
2. Quadrupole Instrument
MS modes—for example, mass spectrometers, that detect
stable ions and those, that detect the unstable ones. The Another type of mass filter is the quadrupole, whose
first type, or the beam-type scanning mass spectrometer, working principle is shown in Fig. 5. Its ability to separate
embraces the magnetic sector and the quadrupole instru- fragments with different masses is based on the fact that,
P1: GSR/GLE P2: FXY Final Pages
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
+✉ +✉ +✉
FIGURE 4 Separation of mass fragments with magnetic sector instrument. Ions (M , M 1 , M 2 , . . . ) are accelerated
out of the ion source into the flight tube of the magnetic sector instrument, whose magnetic field, perpendicular to
+ ✉
the trajectories of the ions, forces the charged particles to move in circular orbit. Only those ions (M ) with a given
m/z value will, at a given acceleration and magnetic strength, become stable and exit the magnetic mass separator to
+ +
be detected, whereas the unstable ones (M 1 , M 2 , . . . ) will stick to the wall of the flight tube.
✉ ✉
when ions travel in a beam of an oscillating electric field, 3. Ion Trap Detector
their trajectories become bent. The quadrupole mass filter
is made up by four parallel rods, about 10 cm long and The working principle of a third type of mass filter, the ion
1 cm in diameter, which in a cross section are arranged trap detector (ITD), is illustrated in Fig. 6. It operates like
to form a square box with a gap in between the four bars. the quadrupole based on the concept that the trajectories of
The fragments are allowed to travel in this space along ions, traveling in an oscillating electric field, become influ-
the rods, where they are accelerated in the oscillating enced by the wave frequency. Dissimilar to the quadrupole
electric field set up by dc and ac currents applied to the filter, however, the ion separation of the ITD occurs in a
rods. At a certain ac and dc potential, ions with a specific closed cavity, where the dc and ac currents applied to a
m/z value will become stable and oscillate in a fixed ring electrode and an end cap of the cell set up the elec-
path through the quadrupole to reach the detector to be tric field. The molecules enter the cell in which they are
recorded. The fragments with m/z values that do not suit ionized by an electron beam, and under the control of the
the applied potentials will become unstable and, as in the given electric field the ions are forced to move in an orbit
magnetic sector instrument, will not reach the detector within the space of the cell. When the ac or dc potential
but will stick to the rods. is changed, the motion of some ions becomes unstable,
P1: GSR/GLE P2: FXY Final Pages
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
+✉ +✉ +✉
FIGURE 5 Separation of mass fragments with quadrupole instrument. Ions (M , M 1 , M 2 , . . . ) are accelerated out
of the ion source into the space between four parallel rods with an oscillating electric field. At a certain ac and dc
+✉
potential, ions with a specific m/z (M ) value will become stable and oscillate in a fixed path through the quadrupole
+ +
to reach the detector to be recorded, whereas the unstable ones (M 1 , M 2 , . . . ) will stick to the rods.
✉ ✉
and 50% of these are then ejected from the cell through differs from the beam-type scanning MS by detecting the
a hole in its bottom to be recorded by a detector. Unlike unstable ions, whereas the magnetic sector or quadruple
the quadrupole or magnetic sector instrument, in which instrument scans the stable ones, conditions that may offer
the separation and detection of an ion occur in a contin- rather different analytical possibilities.
ues process, the ITD operates in two steps separated in
time: ion accumulation and mass analysis. The ITD also 4. Tandem Mass Spectrometer
In tandem mass spectrometry (MS/MS), two mass spec-
trometers and a collision chamber are hooked up in se-
ries. Early devices utilized two magnetic sector instru-
ments, but because these instruments became bulky and
high priced, modern tandem devices are nearly all based
on the use of quadrupole or ITD apparatus. The operating
scheme of MS/MS is shown in Fig. 7. An advantage of
MS/MS is that it provides high selectivity and extra infor-
mation about the nature of an analyte, as well as reduced
chemical background noise, which results in an increase in
the signal-to-noise ratio for a detected peak. Even though
this method, as judged from the number of scientific papers
reported, does not seem to be in common use for foren-
sics today, MS/MS will probably become the method of
choice in the years to come.
D. Mass Focusing
FIGURE 6 Separation of mass fragments with ion trap detector.
+ + +
Ions (M , M 1 , M 2 , . . . ) are introduced into a closed cavity and,
✉ ✉ ✉
The MS test can be run in the full scan mode or with se-
under the control of the applied electric field, the ions are forced
lected ion monitoring (SIM). In the full scan mode, the
to move in an orbit within the space of the cell. When the ac or
dc potential is changed, the motion of ions with given m/z values mass analysis covers a range of m/z values, whereas one
+✉
becomes unstable (M ), and they are then ejected from the cell or a limited number of m/z values are selected for the
through a hole in its bottom to be recorded by a detector. exam with SIM. When picking between the two options,
P1: GSR/GLE P2: FXY Final Pages
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
FIGURE 7 Tandem mass separator. Sample ions enter mass selector 1, where the parent ions are separated. The
selected ions next enter the collision chamber, where they collide with gas molecules to form the daughter ions,
which are finally separated in the mass selector 2 and expelled for detection. Either magnetic sector or quadrupole
mass analyzers or both types mixed can make up a tandem mass spectrometer. A single ion trap can also function
as a tandem mass spectrometer performing the same processes as described above in the same location but in
consecutive steps.
operators of the magnetic sector instrument or the produced and, by changing the electric field of the mass
quadrupole have to consider to what extent they are willing spectrometer, these can be monitored. At certain instances,
to trade sensitivity for selectivity or vice versa and whether such as when the target substances have a high affinity
they are searching for the general unknown or for a sus- for electrons, negative ion monitoring can be extremely
pected agent. The reason for this is that the window of the useful, mainly because of the high sensitivity that can
stable m/z values is sequentially swept across the entire be achieved. The approach of using negative ion mon-
m/z range of interest. The ratio of the transmitted window itoring has been particularly fruitful for the analysis of
width to the width of the entire m/z range (i.e., the duty cy- halogenated drug substances, which have been detected at
cle) is in most scanning tests only a fraction of 1%. More 100- to 1000-fold higher sensitivity than when tested by
than 99% of the ions from the target agents are lost. A duty the positive ion monitoring.
cycle of 100% is possible with a beam-type instrument,
but only when run in a nonscanning mode, as with SIM.
E. Mass Detection
With the ITD, on the other hand, deciding whether to
work in the full scan or SIM mode becomes less crucial. In forensic science work, the electron multiplier is the
As described earlier, the ITD monitors the ions with un- most often used tool for detecting the separated ions, and
stable trajectories in two serial steps. Because these oper- a common device is the so-called horn-type electron mul-
ating steps are separated in time, the yield of detectable tiplier; this variety of detector is the most compact and low
ions will become high and rather independent on the scan cost. When the charged fragments enter the detector and
range. Use of the ITD allows scanning with a high sensi- strike the surface area of the horn, as shown in Fig. 8, elec-
tivity the entire mass range that covers a substance group trons are emitted and accelerated by an electrical potential
of interest. Given that forensic scientists often do not know difference in the horn. These in turn hit the surface and
what to look for and therefore need a search method with new electrons are formed, a process that is repeated over
high sensitivity, the ITD should perhaps best meet such and over again to generate a cascade of a progressively
demands. A drawback of the ITD is that its sensitivity is raised number of electrons, which finally are recorded as
more dependent on interfering substances than the beam- a signal. Usually the gain of the emitted electrons is on
type scanning mass spectrometer; the ITD sensitivity thus the order of 104 to 107 per ion entering the detector.
drops with increasing amounts of impurities that may be
co-eluted with the analytes during the chromatographic
separation. The generation of somewhat distorted mass II. DATA EVALUATION
spectra at high analyte concentrations, giving rise to en-
hanced [M + 1]+ peaks, is another ITD problem. Mass spectrometry is used as an analytical tool in many
In addition to the positive ions formed during the ion- forensic situations. In some instances, however, it can be
ization of a molecule by EI or CI, negative ions are also regarded only as a complement to other chemical methods.
P1: GSR/GLE P2: FXY Final Pages
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
A. Toxicology
The search for drug substances, pesticides, poisons, and
their metabolites in body fluids from living persons or
in postmortem organs presents an important and difficult
task for the chemist. In addition to the fact that the forensic
scientist most often does not know what intoxicant to look
for, the main reason for using MS to begin with is the
large number of possible toxic substances. In toxicology
work at the Poison Center in Munich, as many as 8000
different substances have, in fact, been reported in 40,000
investigated objects.
Figure 9 shows an outline of the usual MS approach
for searching a biological sample taken from a human for
alien compounds with pharmacological effects. The ex-
ample selected is a real-life incident of reckless driving FIGURE 9 Search of blood sample for toxics. The mass chro-
by a motorist apprehended by the police on the suspicion matogram at A shows the total ion current (the sum of all frag-
of being under the influence of some drug or drugs. An ments recorded) and the reconstructed mass chromatogram at B
shows the ions with the sum of m/z 91 and 92. The mass spectra
extract of the blood sample was injected into a gas chro-
at C–F depict the library search for identifying the peak at B. The
matograph/mass spectrometer focused on a broad range unknown analyte’s mass spectrum is, after background subtrac-
of mass fragments. The mass chromatogram at A in the tion, displayed at C. The three hottest candidates in the library
figure is made up of the total ion current (the sum of all along with their names and CAS (Chemical Abstracts Service)
fragments recorded) and showed no clear peaks indicative numbers are shown at D–F. At G are shown the chemical formula
of the first ranked candidate and the value for how well the mass
of any drug substances. To raise the signal-to-noise ratio,
spectrum of the candidate fits with that of the analyte and vice
the total ion current was reconstructed with the sum of the versa. A value of 1000 indicates identical mass spectra; zero, no
m/z 91 and 92, and then a peak appeared on the new mass fragments in common.
chromatogram at B.
In the next step of the analytical process, the substance proof of the analyte identity generally must indicate that
generating the peak at B was to be identified, which was at least two fragments and the retention time are the same
achieved by comparing the mass spectrum of the analyte as for the suggested substance. In the example here, the
at C with mass spectra in an on-line library. Out of ten motorist suspected of being under the influence of drugs
candidates picked by the program, three possible ones are was probably a “sniffer,” who had inhaled paint thinner or
shown at D, E, and F. Even though the mass spectrum of the some other solvent containing toluene (methylbenzene)
analyte best fitted that of methylbenzene, it also matched before driving his car.
nearly as well the mass spectra of the two other candidates.
In addition to the recorded fragments at m/z 91 and 92,
B. Arson Analysis
the final identification of the analyte was based on the
fact that the retention time for the analyte was the same The term arson analysis implies the search of materials
as that for methylbenzene. To hold up to legal scrutiny, taken from a fire scene for accelerant residues to establish
P1: GSR/GLE P2: FXY Final Pages
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
Main type
(approximative boiling point) Major components Examples of commercial products Abundant mass fragments
Light petroleum distillates n-Alkanes, branched alkanes Petroleum ethers, pocket lighter, rubber 42, 43, 56, 57, 71, 85
(<120◦ C) cement solvents, lacquer thinners
Gasolines (50–220◦ C) Alkylbenzenes, naphthalenes Automotive gasoline, lantern fuels 91, 92, 105, 106, 115, 116,
118, 119, 120, 134, 141,
142, 148, 156
Medium petroleum distillates n-Alkanes, branched alkanes, Charcoal starters, paint thinners, mineral 43, 55, 57, 71, 82, 83, 91, 105,
(60–200◦ C) alkylbenzenes spirits, torch fuels, dry-cleaning solvents 120, 134, 138
Kerosene (90–290◦ C) n-Alkanes, branched alkanes, Fuel oil, aviation fuel, insect sprays, 57, 71, 91, 106, 120, 128, 134,
alkylbenzenes, naphthalenes charcoal starters 142, 148
Heavy petroleum distillates n-Alkanes, branched alkanes, Fuel oil, diesel fuel 43, 57, 71, 128, 142
(120–410◦ C) naphthalenes
Varia Alcohols, ethers, α-pinene Solvents, turpentine 31, 32, 45, 59, 73, 74, 93
whether the incident was of incendiary origin or not. The ceased were also searched for fuel components. On the
U.S. National Fire Protection Association reported that chromatogram monitoring the sum of all ions (TOT), no
over 100,000 fires in 1994 were arson related and more peaks appeared, but after it had been reconstructed with
than 500 persons lost their lives in these fires. In terms the selected ions at m/z 78 + 91 + 92 a number of aro-
of loss of human life, houses, properties, and goods, this matic hydrocarbons showed up. These could, however,
type of incidence, extracts a large toll from society as stem either from a fuel or from pyrolyzed plastics. The
well as from individuals, and makes the forensic task presence of MTBE at m/z 73 indicated, however, that
urgent. gasoline had been used to set the fire. This piece of evi-
The list of the most common accelerants, shown in dence pointing to arson could not have been brought light
Table I, may at first give the impression that these would without MS.
rapidly gasify along with the burning solid materials and
thus not become detectable, but it is not always so. Traces C. Environmental Issue
of the fuel often remain in some closed areas of the fire
scene even after the temperature has reached perhaps Growing global use and transportation of chemicals with
1000◦ C, and these accelerant residues may be detectable potential deleterious effects on the environment have
after proper sample collection. prompted forensic scientists to develop methods for
By virtue of its high sensitivity and specificity, GC/MS
is well suited for searching trace amounts of test materi-
als for residues of accelerant components. As reported in
the literature on material spiked with accelerants, specific
GC/MS patterns of different types of fuels may be used
to identify an accelerant. A problem with this approach
is that burning plastics and other solid materials release
hydrocarbons that are the same as those in accelerants and
thus may contribute to a false positive result by tainting
the test material. The use of fuel labels may overcome
this hurdle. Methyl-tertbutyl-ether (MTBE), which is an
additive in gasoline, has been suggested as such a marker.
Owing to its low boiling point (55.2◦ C), however, it disap-
pears rapidly during a fire and is therefore only detectable
in postmortem materials from a fire victim who has in- FIGURE 10 Arson analysis of blood sample. The mass chro-
matogram at A shows the total ion current (the sum of all fragments
haled MTBE. Figure 10 shows such an example from the
recorded). The reconstructed mass chromatogram at B shows the
real-life fire of a villa, which was completely destroyed. ions with the sum at m/z 78, 91, and 92 to monitor aromatic hydro-
No accelerant residues were spotted in the ashes around carbons, and the reconstructed mass chromatogram at C shows
the body of a fire victim, so blood samples of the de- the ions at m/z 73 to monitor MTBE.
P1: GSR/GLE P2: FXY Final Pages
Encyclopedia of Physical Science and Technology EN009G-958 July 18, 2001 0:57
detecting the origin of illicit waste discharge. A common and the other from Californian crude oil. The latter source
issue is to link the constituents of, for instance, an oil spill had probably been released in to the sea from storage tanks
sample with its original source. GC or GC/MS are usually by an earthquake that had taken place in the area 40 years
used to do the analysis when the individual hydrocarbons earlier.
are separated and displayed on a chromatogram, whose
pattern is compared with that of the suspected original
source. A major problem in such a survey, however, is SEE ALSO THE FOLLOWING ARTICLES
that some time usually has passed between the waste dis-
charge and the specimen collection. Owing to weathering DNA TESTING IN FORENSIC SCIENCE • GAS CHROMATO-
GRAPHY • ION KINETICS AND ENERGETICS • LIQUID
by evaporation and biodegradation of the components dur-
ing this period, the composition of the oil spill may have CHROMATOGRAPHY • MASS SPECTROMETRY • POLLU-
TION, ENVIRONMENTAL • SPECTROSCOPY IN FORENSIC
changed to the extent that it bears only limited similarities
with that of the origin. SCIENCE • TOXICOLOGY IN FORENSIC SCIENCE
Monitoring the contents of stable isotopes by MS in the
whole sample or in its individual constituents after GC has BIBLIOGRAPHY
isolated them is an approach used to get around the weath-
ering problems. It has been shown that the most common Currell, G. (2000). “Mass spectrometry systems.” In “Analytical Instru-
stable isotope parameter used (i.e., the 13 C/12 C ratio) may mentation, Performance, Characteristics and Quality,” pp. 181–191,
be specific for a source and not very much influenced by John Wiley & Sons, New York.
weathering effects. Technically, the analysis is done after Fifield, F. W., and Kealey, D. (2000). “Mass spectrometry.” In “Principles
and Practice of Analytical Chemistry,” 5th ed., pp. 426–440, Blackwell
the test material has been combusted to carbon dioxide and Scientific, Oxford.
water, and the 13 C/12 C ratio is then determined in the car- Keto, R. O. (1995). “GC/MS data interpretation for petroleum distil-
bon dioxide fraction by MS. To determine the 13 C/12 C ratio late identification in contaminated arson debris,” Journal of Forensic
of the individual components of a sample, they are sepa- Sciences 40, 412–423.
Masucci, J. A., and Caldwell, G. W. (1995). “Techniques for gas chro-
rated by GC, each isolated hydrocarbon goes directly into a
matography/mass spectrometry.” In “Modern Practice of Gas Chro-
combustion chamber, and the carbon dioxide, after having matography,” 3rd ed. (R. L. Grob, ed.), pp. 323–391, John Wiley &
been freed from water, enters the mass spectrometer. An il- Sons, New York.
lustration of the use of this tool to identify oil spill sources McNair, H. M., and Miller, J. M. (1998). “Special topics.” In “Basic Gas
is the stable isotopes of samples taken from the area of Chromatography. Techniques in Analytical Chemistry,” pp. 153–163,
Prince William Sound several years after the Exxon Valdez John Wiley & Sons, New York.
Wasels, R., and Belleville, F. (1994). “Gas chromatographic–mass spec-
disaster. The isotope distribution in the samples as mea- trometric procedures used for the identification and determination
sured on the bulk contents correlated with two distinct ori- of morphine, codeine and 6-monoacetylmorphine,” Journal of Chro-
gins of the oil spill: one coming from the Exxon Valdez oil matography A 674, 225–234.
P1: GNH Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
Microwave Molecular
Spectroscopy
Robert L. Cook
Mississippi State University
I. Background
II. Experimental Techniques
III. Evaluation of the Moments of Inertia
IV. Rigid-Rotor Energy Levels and Spectra
V. Centrifugal Distortion Effects
VI. Rotation–Vibration Interactions
VII. Internal Rotation
VIII. Evaluation of Molecular Structures
IX. Stark Effect and Zeeman Effect
X. Nuclear Quadrupole Hyperfine Structure
XI. Advanced Experimental Methods
799
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
magnetic), the coupling of internal and overall rota- of the electromagnetic spectrum, which lies between the
tion, or the effects of nuclear coupling. conventional radiowave region and the infrared region.
Internal rotation Rotation of two parts of a molecule The distinguishing features that characterize microwave
about a single bond. Such an internal rotation is often spectroscopy are high resolution and highly accurate
hindered by a barrier. frequency measurements. These characteristics, coupled
Klystron, backward wave oscillator, YIG oscillator with a sensitivity to molecular structure, isotopic com-
Sources of tunable monochromatic electromagnetic ra- position, and various other molecular properties, make
diation in the microwave region. it a powerful technique for obtaining detailed molecular
Microwave region Part of the electromagnetic region ex- information.
tending from about 1000 MHz (λ = 30 cm) to 1 mil-
lion MHz (λ = 0.3 mm).
Microwave spectrometer A microwave source, absorp- I. BACKGROUND
tion cell, and detector. Used to investigate the rotational
spectrum. In microwave molecular spectroscopy, absorption spec-
Microwave spectrum Usually an absorption spectrum tra arise from molecular rotation and correspond to tran-
(i.e., the abstraction of energy from the radiation field). sitions between the rotational energy levels associated
The spectrum is associated with transitions between with a given vibrational state of a particular electronic
rotational energy levels in a given vibrational and elec- state (see Fig. 1). The rotational transitions, which fall
tronic state and consists of a series of lines, each char- in the microwave region, are induced through the inter-
acterized by a frequency and an intensity. action of the molecular electric dipole with the electric
Rigid rotor Idealization in which a molecule is treated vector of the radiation. The microwave region extends
as a rigid, nonvibrating rotor. roughly from 1000 MHz (λ = 30 cm) to 1 million MHz
Rotational constants Constants that mainly determine (λ = 0.3 mm). In this spectral region, frequencies are ex-
the rotational energy of a molecule. They are inversely pressed in megahertz (MHz) or gigahertz (GHz) units,
proportional to the moments of inertia, which in turn where the Hertz unit denotes cycles per second; 1 MHz =
depend on the mass and geometry of the molecule. 106 Hz and 1 GHz = 109 Hz. In terms of wavelengths, the
Rotational energy Quantum mechanics requires mole- region may be conveniently divided into the centimeter-
cules to rotate only at certain rates and hence to have wave (λ = 1–30 cm), millimeter-wave (λ = 1–10 mm),
only discrete values of rotational energy and total an- and submillimeter-wave regions (λ < 1 mm). Note that 30
gular momentum. and 300 GHz correspond to 1 cm and 1 mm, respectively.
Rotational quantum numbers Numbers that specify The rotational absorption spectrum depends on the prin-
various angular momenta and distinguish the rotational cipal moments of inertia and, hence, is characteristic of the
energy levels.
Satellite spectrum Rotational absorption lines that arise
from an excited vibrational state rather than the ground
vibrational state.
Selection rules Rules that specify whether or not a
given transition between two particular energy levels is
allowed.
Symmetric top Molecule for which two of the three prin-
cipal moments of inertia are equal. Symmetric tops may
be further divided into prolate and oblate tops.
Waveguide Rectangular-shaped metal pipe used to trans-
mit microwave radiation. Different frequency regions
require different-size waveguides.
whole molecule and very sensitive to the molecular struc- P-band (12.4–18 GHz), K-band (18–26.5 GHz), and
ture, isotopic composition, and numerous other molecu- R-band (26.5–40 GHz). The simplest absorption cell con-
lar properties. Information that can be obtained from the sists of an X-band waveguide 2–5 m long with mica win-
observation and analysis of rotational transitions includes dows at each end to provide a vacuum seal. Small holes
precise molecular structures, dipole moments, centrifu- can be made in the waveguide, usually on its wide face, to
gal distortion constants, vibrational potential functions, allow evacuation of the cell and for sample introduction.
internal rotation barriers, nuclear masses and spins, nu- This cell is used for the various higher frequency bands
clear quadrupole coupling constants, molecular magnetic mentioned previously by employment of tapered transition
moments, conformations of rotational isomers and ring sections to match the cell to the smaller waveguide of the
compounds, magnetic susceptibility and electric polariz- source and detector. Sometimes, the interior of the absorp-
ability constants, molecular quadrupole moments, struc- tion cell is gold-plated to limit chemical decomposition.
tures and other properties of hydrogen-bonded complexes, The microwave radiation passing through the absorp-
rare atom–molecule complexes, molecular ions, nonpo- tion cell is detected by a crystal detector mounted at the
lar spherical-top molecules, and qualitative and quan- end of the waveguide. The small output voltage is ampli-
titative analysis. The field of microwave spectroscopy fied and displayed on an oscilloscope or chart recorder.
continues to expand with new experimental and theo- A microwave spectrum is obtained by sweeping the mi-
retical developments. The contributions of microwave crowave source over a range of frequencies and observing
molecular spectra to the fields of microwave–optical and the small variations in power at the detector. An absorption
microwave–infrared double-resonance spectroscopy, mi- line appears as a sharp dip on the recording system.
crowave molecular astronomy, and structures of weak To increase the sensitivity of a microwave spectrome-
complexes attest to its vitality. ter, Stark modulation may be added. For this purpose, a
thin metal septum is mounted in the absorption cell par-
allel to the broad side of the waveguide, using insulating
II. EXPERIMENTAL TECHNIQUES material such as Teflon (see Fig. 2). An alternating (5–
100 kHz) voltage, usually a zero-based square wave, is
The microwave region has its own methods of generat- applied to the septum. Typically, voltages from a few to
ing, detecting, and measuring microwave radiation. Dur- 2000 V are applied. As discussed later, in the presence of
ing World War II considerable work was done to develop an external electric field, the rotational energy levels and,
microwave radar systems. This led to new sources—the therefore, the absorption lines are split into a number of
klystron and other hardware associated with the propa- components. If, for example, the source frequency is tuned
gation and measurement of microwaves. After the war, to the resonant frequency of an absorption line, the Stark
this provided the impetus for a rapid development of mi- voltage periodically shifts the absorption frequency away
crowave spectroscopy as an area of study. from the frequency of the source, resulting in a modulation
(field on and off ) of the absorption line at the frequency
of the Stark modulator. In general, a signal reaches the
A. Conventional Microwave Spectrometer
amplifier when the source being swept reaches the field-
A typical microwave spectrometer is illustrated in Fig. 2. off absorption frequency or the frequency of the field-on
The essential elements of a microwave spectrometer are a absorption lines (Stark components or lobes). The mod-
microwave source, absorption cell, detection system, and ulated signal is detected and amplified by a preamplifier
a system for measuring the source frequency. Microwave tuned to the modulation frequency. By employing phase-
sources—the klystron and, more recently, the backward- sensitive detection, one can obtain additional sensitivity.
wave oscillator (BWO)—generate a very narrow band of Here, a final amplifier is referenced to the modulation fre-
frequencies so that the source is essentially monochro- quency, and only the noise that has the same phase and
matic. Furthermore, the source frequency can be conve- frequency as the signal is amplified. With phase-sensitive
niently varied and is often phase stabilized to give good detection, the Stark components are displayed in opposite
frequency stability. phase to the zero-field line. Figure 3 shows the appearance
Microwave radiation is conveniently transmitted of a rotational transition obtained with a Stark-modulated
through hollow, rectangular metallic conductors called spectrometer with application of different modulation
waveguides. These are usually made of brass or copper, voltages.
and the rectangular dimensions depend on the frequency Frequency measurements are made by comparing the
range to be transmitted, the size decreasing with increasing frequency of the microwave source with the appropriate
frequency. Typical frequency bands used in the commonly harmonic of a stable oscillator that is calibrated with a
studied low-frequency region are X-band (8–12.4 GHz), frequency standard. The difference frequency between the
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
FIGURE 2 Basic elements of a Stark-modulated microwave spectrometer. The mounting of the Stark electrode in
the absorption cell is shown in the inset.
harmonic signal and the source may be used as the input of 1 part in 106 or better and with resolutions on the order
to a digital electronic counter. Frequency markers are of- of 0.1 MHz. Actually, higher resolution is obtainable with
ten displayed on a chart recorder (see Fig. 2) along with special spectrometers (see Section XI.D).
the absorption line, and these are employed in the fre-
quency measurements. Approximate frequency measure-
B. Millimeter/Submillimeter-Wave Spectrometer
ments of the source frequency can be readily made with a
cavity wavemeter. Absorption lines in the microwave re- As the frequency of the microwave radiation increases, the
gion are commonly measured with frequency accuracies sensitivity of the technique increases rapidly [see Eq. (1)].
FIGURE 3 Effect of the square-wave Stark-modulation voltage on the appearance of the J = 2 → 3 transition of OCS.
Note that as the Stark lobes labeled by |M| are displaced from the field-off line, the intensity of the line increases. The
Stark effect is second order. The energy level diagram for a linear molecule in an electric field is shown in the inset.
The allowed M = 0 transitions are also depicted.
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
Moreover, many small, fundamental molecules have spec- C. Line Intensities and Shapes
tra which fall in the shorter wavelength region. For these
A few comments on the shapes and factors affecting line
reasons, spectrometers have been developed operating at
intensities are in order. An absorption line is not perfectly
higher frequencies. This is accomplished by harmonic
sharp but is usually symmetrical about the resonant fre-
generation of the higher frequencies from a high-powered,
quency. It is characterized by its resonant frequency ν0 , its
lower frequency source using a point-contact crystal diode
shape, and its linewidth 2(ν), where ν is the half-width
harmonic generator. Development of this and various other
at half-intensity points of the line. At very low pressures,
techniques to exploit the millimeter- and submillimeter-
<10 mTorr (1 mTorr = 10−3 Torr = 1 µm = 10−3 mm of
wave regions was pioneered by Walter Gordy and his
Hg), the linewidth arises primarily from Doppler broad-
research group. In the shorter wavelength region, quasi-
ening and is independent of pressure, and the line shape
optical techniques can be used, and more versatile ab-
is Gaussian. This is the region of maximum resolution.
sorption cell designs are possible. The Stark cell is usu-
As the pressure is raised in this region, the peak line in-
ally replaced by a quasi-free space cell. Both metal and
tensity increases, while the linewidth remains constant.
glass cells have been employed. Radiation transmission
As the pressure is further increased, collision broadening
through the adsorption cell is properly matched to the
becomes important. At pressures >10 mTorr, broadening
waveguide by employing waveguide horns equipped with
due to molecular collisions becomes dominant. The line
a Teflon lens (see Fig. 26). A block diagram of a broadband
shape is now Lorentzian, the half-width is directly propor-
millimeter/submillimeter-wave spectrometer is given in
tional to the pressure ν = kp, and the peak intensity is
Fig. 4. The YIG microwave oscillator is phase locked to
independent of pressure. This is the region of maximum
the frequency synthesizer, and a computer is used to sweep
sensitivity.
the synthesizer. The output of the YIG is tripled and am-
A useful quantitative measure of the intensity of a rota-
plified by a TWT amplifier to ca. 1 W. The output is then
tional transition is given by the peak absorption coefficient
further multiplied by a harmonic generator. The harmonics
α0 , defined by
of the driving signal in the millimeter- and submillimeter-
wave regions are propagated quasi-optically through the 8π F j,τ µ2g λg (J, τ ; J , τ ) ν2x
1-m-long absorption cell. The spectrum is detected with α0 = · 0 . (1)
3ck T (2J + 1)
2 2 ν/ p
lock-in detection and digitized and displayed by the
computer. This applies for any class of rotor with FJ,τ , the fraction
of molecules in the lower state J, τ of the transition and
the vibrational state v; µg is the electric dipole moment
component giving rise to the particular transition under
observation, c is the speed of light; k is the Boltzmann
constant; T is the absolute temperature of the gas, ν0 is
the frequency for which the absorption is a maximum; x
is the mole fraction of the absorbing molecular species;
p is the total pressure in the absortion cell; and ν is
the half-width of the line. The above expression summa-
rizes the various factors that affect the line intensity. Since
ν is proportional to p, α0 is independent of total pres-
sure. Furthermore, the intensity increases with frequency,
dipole moment, and λg (J, τ ; J , τ ), the line strength of
the transition. This latter quantity is related to the transi-
tion moment by
|µi j |2 = µ2g λg (i; j) (2J + 1). (2)
The line strength for asymmetric rotors has been tabulated
FIGURE 4 Schematic diagram of a computer-controlled broad- for various values of κ. For a symmetric top a simple
band millimeter/submillimeter-wave spectrometer (νmax ∼ = relation applies:
440 GHz). The YIG oscillator is phase locked and swept by
computer. A helium-cooled indium antimonide detector is used λ(J, K ; J + 1, K ) = [(J + 1)2 − K 2 ]/(J + 1). (3)
for signal detection, and the YIG is frequency modulated for im-
proved signal recovery by lock-in detection. [After Booker, R. A., This also applies to a linear molecule with K = 0. The
Crownover, R. L., De Lucia, F. C., and Helminger, P. (1988). intensity also depends on the fraction of molecules
J. Mol. Spectrosc. 128, 62.] FJ,τ , in the lower rotational state J, τ of the vibrational
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
TABLE I Types of Molecular Rotorsa TABLE II Expressions for the Moments of Inertia of Some
Simple Molecules
Spherical top Ia = Ib = Ic All three moments of inertia
Molecules Moments of inertia
molecules are equal, e.g., CH4
Linear molecules Ia = 0, Ib = Ic Axis a along the internuclear
mXmY
axis, b and c perpendicular Diatomic, XY I = d2
m X + m Y XY
to this axis, e.g., OCS, HCl
1
Symmetric-top Linear, XYZ I = 2
m X m Y dYX + m Y m Z dYZ
2
M
molecules
Prolate top Ia < Ib = Ic Axis of least moment of + m X m Z (dYX + dYZ )2
inertia, a, along the θ
symmetry axis, e.g., CH3 F Bent, XY2 b I x = 2m Y dXY
2 sin2
2
Oblate top Ia = Ib < Ic Axis of largest moment of 2m X m Y 2 θ
Iy = dXY cos2
inertia, c, along the M 2
symmetry axis, e.g., BCl3 Iz = Ix + I y
Asymmetric-top Ia = Ib = Ic All three moments of inertia Pyramidal, XY3 c 2 sin2 θ + m X m Y d 2
I x = I y = 2m Y dXY XY
molecules different, e.g., SO2
2 M
θ
× 3 − 4 sin2
a For a symmetric top, the molecule is designated a prolate or oblate 2
rotor, depending on which inertia axis corresponds to the molecular θ
Iz = 4m Y dXY
2 sin2
symmetry axis. Most molecules belong to the asymmetric rotor case. 2
a The d is the bond distance between atoms i and j; M is the total
ij
mass of the appropriate molecule; and m i is the mass of the ith atom.
principal moments of inertia. The various cases are sum- b The x axis corresponds to the C axis, with the z axis perpendicular
2
marized in Table I. It may be noted that molecules with a to the x y plane and with θ as the Y X Y bond angle.
c The z axis is the C symmetry axis and θ is the Y X Y bond angle.
threefold or higher axis of symmetry are symmetric tops, 3
and this symmetry axis is a principal intertial axis. Also, Note that the acute angle β between the X Y√bond and the symmetry
axis is related to the bond angle by sin(θ/2) = ( 3/2) sin β.
any two perpendicular axes that are perpendicular to the
symmetry axis are principal axes, and the corresponding
moments of intertia are equal. As indicated in the table,
where xi , yi , and z i are the coordinates of the ith atoms
there are two types of symmetric rotors. For the prolate
of mass m i relative to the arbitrary coordinate system and
case, the molecule is elongated like a football, while for
M is the total mass of the molecule. The center-of-mass
the oblate case, the molecule is flattened like a disk. Most
coordinates of the atoms are computed from
molecules are asymmetric tops, and if the molecule has
some symmetry, one or more of the principal axes may xi = xi − x̄, yi = yi − ȳ, z i = z i − z̄. (11)
be selected. If a molecule has a twofold axis of symmetry,
then this axis must be a principal axis. If a symmetry plane The elements of the moment of inertia tensor are evaluated
is present, then two principal axes must lie in this plane and from these coordinates by means of the expressions
the third must be perpendicular to this plane. It often oc-
curs that two moments of inertia are accidentally close to Ix x = m i yi2 + z i2 , Ix y = − m i xi yi ,
each other and the slightly asymmetric top approximates
one of the symmetric tops. In such cases, it is referred to I yy = m i xi2 + z i2 , Ix z = − m i xi z i , (12)
as a near-prolate or near-oblate asymmetric top.
Expressions for the principal moments of inertia of Izz = m i xi2 + yi2 , I yz = − m i yi z i .
some simple molecules are collected in Table II. To evalu-
The Ix x , . . . are the moments of inertia and the Ix y , . . .
ate the moments of inertia in the general case, an arbitrary
are called the products of inertia. The inertia matrix I is
but convenient coordinate system may be chosen in the
symmetric, and diagonalization of this matrix by standard
molecule. The center of mass (x̄, ȳ, z̄) is given by
techniques (such as Jacobi’s rotation method),
m i xi
x̄ = , Ia 0 0
M
R̄IR = 0 Ib 0 , (13)
m i yi
ȳ = , (10) 0 0 Ic
M
m i z i provides the principal moments of inertia, and the transfor-
z̄ = ,
M mation matrix R provides the orientation of the principal
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
axis system relative to the initial arbitary axis system. tailed molecular structures. Because of the sensitivity of
Equivalently, the secular determinant microwave spectroscopy, various isotopic forms can often
be studied in natural abundance.
Ix x − λ Ix y Ix z
Ix y I yy − λ I yz = 0 (14) IV. RIGID-ROTOR ENERGY LEVELS
Ix z I yz Izz − λ AND SPECTRA
may be expanded and solved for its roots, which are the To a good approximation, the energy of a molecule may
principal moments. In general, the above is of third degree be expressed as the sum of the electronic, vibrational, and
in the unknown λ. If, however, one principal axis is known, rotational energies. In pure rotational spectra, transitions
then by taking one of the initial coordinate axes along this take place between rotational sublevels with no change in
direction, one can make the products of inertia associated the electronic or vibrational state. The gross features of the
with this axis vanish, and the secular equation is simplified. pure rotational spectrum of a molecule may be ascertained
The masses employed in calculating the moments and by treating the molecule as a rigid, nonvibrating rotor. Sub-
products of inertia must correspond to a single isotope for sequently, the effects of centrifugal distortion and vibra-
each atom in the molecule. Since the principal moments tion may, in many cases, be included by application of per-
of inertia are different for different isotopic forms of a turbation theory. This is considered further in Sections V
molecule, quite different rotational spectra are obtained. and VI. The molecular geometry, via the moments of in-
In fact, if the molecular mass distribution in a molecule ertia, determines the pattern of the rigid-rotor spectrum.
is changed, the rotational spectrum is affected. The spec- This pattern is relatively simple for linear and symmetric-
trum of 2-chloropyridine can, hence, be expected to be top molecules; however, for asymmetric rotors, there is
quite different from that of 3-chloropyridine. The impli- little regularity to the spectrum except in certain cases.
cations for qualitative analysis of these chemical isomers The frequency of electromagnetic radiation absorbed
are obvious. Rotational isomerism also changes the mass depends on the energy difference between the two states
distribution, which is illustrated in Fig. 10. The sensitiv- and is given by the Bohr relation
ity of the transition frequency to isotopic composition is
ν = (E j − E i )/ h, (15)
shown in Fig. 5 for hydrogen selenide. From a study of the
rotational spectrum of different isotopic forms, additional where h is Planck’s constant and E j and E i are the upper
information is obtained that can be used to evaluate de- and lower rotational energy states, respectively. In what
FIGURE 5 A transition of the asymmetric rotor HDSe. Spectral trace was recorded from left to right and extends
approximately 170 MHz. The five major isotopic species of Se are clearly evident. The spectrum was taken at the
temperature of dry ice; nevertheless, HDSe decomposes in the Stark cell. This is evident from the decrease in line
intensity of 76 Se compared with that of 82 Se.
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
follows, the factor 1/ h will be included in the energy ex- where E is a diagonal matrix of eigenvalues. The transfor-
pression, so that E will be expressed in frequency units. mation matrix T, which diagonalizes the energy matrix,
According to quantum mechanics, gaseous molecules ro- yields the eigenfunctions. Each column of T corresponds
tate at only certain rates and have discrete values of rota- to the set of expansion coefficients an (or eigenvector) for
tional energy and angular momentum. When a quantum of the eigenfunction ψ in the basis {φn }. These define the
energy hν is absorbed, the molecule is raised to a higher set of asymmetric-top functions and are required to calcu-
energy level and rotates at the next allowed rate. The type late the average values of an operator in the asymmetric
of transition that can occur is governed by the electric rotor basis. It may be noted that diagonalization of H is
dipole moment matrix element or transition moment, equivalent to solving the secular determinate
|H − IE| = 0, (20)
µi j = ψi∗ µψ j dτ = (i|µ| j), (16)
which is particularly useful when the order of H is small.
where ψi∗ and ψ j are, respectively, the wavefunctions for Here I is a unit matrix.
the lower and upper states of the transition. The i and
j denote the set of quantum numbers required to char-
A. Diatomic and Linear Molecules
acterize these states. Only those transitions are allowed
for which the above matrix element is nonvanishing. This The Hamiltonian operator for a rigid diatomic or linear
places certain restrictions on the changes possible in the molecule is
quantum numbers, and these restrictions are called selec-
= P 2 /2I, (21)
tion rules. The basic problem of rotational spectroscopy
is the evaluation of the quantized rotational energy lev- where P is the total angular momentum and I the moment
els and selection rules by quantum mechanical techniques of inertia. The operator PZ corresponds to the projection
and the subsequent assignment of the observed transi- of P along a space-fixed Z axis. According to quantum
tions to particular levels characterized by certain quantum mechanics, since the operators , P 2 , and PZ commute
numbers. with each other, they have a common set of eigenfunc-
The general procedure for evaluating the quantum me- tions, which we denote by ψ J,M ≡ |J, M). Classically,
chanical energy levels starts with the formulation of the these quantities are constants of motion. The matrix el-
Hamiltonian , expressed in terms of the angular momen- ements of the angular momentum are
tum operators and coordinates, if required. For rigid linear
(J, M|P 2 |J, M) = h 2 J (J + 1),
and symmetric-top molecules, the Hamiltonian operator (22)
is simple enough that the energy eigenvalue equation may (J, M|PZ |J, M) = h M.
be solved directly,
Therefore, the rotational energy is given by
ψ = Eψ, (17) E J = BJ (J + 1), (23)
to give the eigenfunctions ψ and the energy levels E. For where E J is in frequency units and B = h/8π I . For 2
many problems of interest such as an asymmetric top, the linear molecules, the rotational energy levels are charac-
above eigenvalue equation cannot be solved directly. In terized by two quantum numbers J and M, which are
such cases, the wavefunctions may be expressed as linear restricted to certain integral values:
combinations of a complete set of known functions {φn },
such as a set of symmetric top functions, J = 0, 1, 2, . . . , M = 0, ±1, ±2, . . . , ±J. (24)
ψ= an φn . (18) In the absence of external fields, the rotational energies do
not depend on M, as Eq. (23) implies, and all levels are
According to quantum mechanics, the energy levels are (2J +1)-fold degenerate. However, when an external field
now found by diagonalization of the Hamiltonian or en- is applied, this degeneracy is lifted, and the energy depends
ergy matrix H = [n,n ], where the elements are the ma- on the space orientation quantum number M. A similar
trix elements of in the particular basis {φn }. With the condition holds for symmetric and asymmetric tops. For
availability of high-speed computers, it is usually a rel- rotational absorption of radiation, the selection rule
atively simple matter to diagonalize a symmetric matrix
by appropriate numerical methods. Diagonalization of the J → J +1 (25)
matrix representation of the operator gives the eigenvalues applies, and the rotational frequencies of a rigid linear
of the operator molecule are given by
T̄HT = E, (19) ν = E J +1 − E J = 2B(J + 1). (26)
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
1.063 1.155
H C N 44,315.98 87.2
1.067 1.542
H C P 19,976.01 21.2
1.262 1.159
F C N 10,554.20 5.3
1.285 1.541
F C P 5,257.80 1.0
1.629 1.160
Cl C N 5,970.83 1.7
1.789 1.160
Br C N 4,120.22 8.8
1.129 1.189
N N O 12,561.64 5.4
1.160 1.560
FIGURE 6 The first few rotational energy levels for a linear O C S 6,081.49 1.3
1.053 1.198 1.279
molecule. The allowed transitions and the resulting spectrum, with H C C F 9,706.19 —
1.055 1.204 1.637
approximate intensities, are also shown. H C C Cl 5,684.24 —
1.95 0.96
Na O H 12,567.05 28.7
A series of lines at 2B, 4B, 6B, . . . is thus expected for K
2.21
O
0.91
H 8,208.68 12.2
2.80
a rigid rotor. The energy levels, allowed transitions, and HCN · · ·HFa 3,591.11 5.2
3.07
spectrum of a rigid linear rotor are illustrated in Fig. 6. OC · · ·HFa 3,063.90 9.8
3.54
The molecule OCS, which is commonly used as a stan- Ar · · ·HFa 3,065.71 70.9
3.65
dard for various purposes by microwave spectroscopists, Kr · · ·HFa 2,392.41 31.9
1.11
O]+
1.09
has lines that occur at 12,162.97, 24,325.92, 36,488.80, [H C 44,594.42 82.4
48,651.40 MHz, . . . for the most common isotope. For S
1.4840
O 21,523.56 33.9
a light molecule such as CO, B = 57,635.97 MHz, and a For these complexes the bridge length r (X F) is
the rotational lines are spaced 115,271.94 MHz apart;
given with X = N, C, Ar, or Kr.
thus, high-frequency microwave techniques must be em-
ployed to measure even the 0 → 1 transition, which is
at 115,271.94 MHz. The effect of centrifugal distortion E J,K = BJ (J + 1) + (A − B)K 2 , (29)
is to produce a small shift to lower frequency in each with the rotational constants defined as A = h /8π 2Ia and
transition. Illustrative rotational constants are collected in B = h/8π 2Ib . The energy levels are characterized by the
Table III . quantum numbers J, K , and M, with
J = 0, 1, 2, . . . ,
B. Symmetric-Top Molecules
K = 0, ±1, ±2, . . . , ±J, (30)
The rotational Hamiltonian for a prolate symmetric top
has the form M = 0, ±1, ±2, . . . , ±J.
Bond Bond
Molecule Bond length (Å) Angle angle (deg)
where if B ∼
= C, the parameter is small, b p ∼
= 0. For a near- (J, K , M|P 2 |JKM) = J (J + 1)
oblate top, Wang’s parameter is
(J, K , M|(b p )|J, K , M) = K 2
b0 = (A − B)/(2C − A − B), (36)
(J, K , M|(b p )|J, K ± 1, M) (40)
where if A ∼
= B, b0 ∼
= 0. An alternate measure of the asym- = (1/2)b p {[J (J + 1) − K (K ± 1)]
metry is Ray’s parameter
× [J (J + 1) − (K ± 1)(K ± 2)]}1/2 .
κ = (2B − A − C)/(A − C). (37)
The energy matrix is diagonal in J and has both diag-
For a prolate symmetric top, κ = −1, and for an oblate onal and off-diagonal elements in K , the off-diagonal
top, κ = +1. If B ∼= C, κ ∼
= −1, the asymmetric top may elements coming from the operator (Pc2 − Pb2 ). Since
be classified as a near-prolate rotor. If A ∼
= B, κ ∼ = +1, 1/2(B + C)P 2 contributes only a constant diagonal term,
we have a near-oblate asymmetric rotor. The case κ = 0 a reduced energy matrix may be defined as
corresponds to the most asymmetric top possible. Most
E K ,K = (J, K , M|(b p )|J, K , M), (41)
molecules are prolatelike asymmetric tops. Illustrative ex-
amples of rotational constants for some asymmetric tops with K = K or K ± 1. Diagonalization of this for each
are given in Table V. value of J gives the Wang reduced energies W J,τ (b p ), and
In terms of Wang’s parameter for a prolate asymmetric the total rotational energy is given by
rotor, we can reexpress as
E J,τ = (1/2)(B + C)J (J + 1)
= 12 (B + C)P 2 + A − 12 (B + C) (b p ), (38) + [A − (1/2)(B + C)]W J,τ (b p ). (42)
(b p ) = Pa2 + b p Pc2 − Pb2 . (39) Expressions for W(b p ) for a few low J levels are given
in Table VI. This formulation applies to any asymmetric
Here and in what follows it is convenient to measure angu- top; however, if b p is small, as for a near-prolate rotor,
lar momentum in units of h, Pa2 / h 2 → Pa2 , and so on. The the off-diagonal matrix elements are small, and the matrix
(b p ) is termed a reduced Hamiltonian. In a symmetric is more easily diagonalized. Furthermore, in the case of a
rotor representation the nonvanishing matrices of P 2 and near-prolate (or oblate) asymmetric rotor, the W(b p ) differ
(b p ) are only slightly from K 2 and may be conveniently expressed
TABLE V Rotational and Centrifugal Distortion Constantsa and Structures of Some Asymmetric Rotors
Molecule A B C ∆ J (103 ) ∆ J K (102 ) ∆ K (100 ) δ J (104 ) δ K (102 )
TABLE VI Wang’s Reduced Energiesa The size of the above submatrices depends on the maxi-
J K −1 K 1 W(b p ) mum value of K , which in turn depends on the value of
J being considered. For an oblatelike asymmetric rotor
00,0 0 (κ > 0), A and C as well as a and c are interchanged in
11,0 1 − bp the above relations:
11,1 1 + bp
E J,τ = (1/2)(A + B)J (J + 1)
10,1 0
22,0 2 + 2 1 + 3b2p + [C − (1/2)(A + B)]W J,τ (bo ). (45)
22,1 4 A parallel formulation using κ can also be made and has
21,1 1 − 3b p often been used in the literature. Here the reduced energies
21,2 1 + 3b p are denoted by E J,τ (κ), and the total energy is given by
20,2 2 − 2 1 + 3b2p
E J,τ = (1/2)(A + C)J (J + 1) + (1/2)(A − C)E J,τ (κ).
a Total asymmetric-top energy is given by Eq. (42) for
(46)
any asymmetry b p .
The reduced energies are again found by diagonalization
of the matrices of Eq. (43), with the specific elements now
as an expansion in terms of Wang’s asymmetry parame- a function of κ.
ter. The energy matrix can actually be factored into four The effect of the asymmetry is to remove the twofold
smaller submatrices by taking as basis functions a set of degeneracy of the K levels (K = 0) for a symmetric top.
functions that are linear combinations
√ of the original func- This K -splitting of the levels increases as the asymmetry
tions: (ψ JKM ± ψ J −K ,M )/ 2 and ψ JOM . These functions increases. There are thus 2J + 1 different rotational lev-
may be classified according to the symmetry of the rota- els or values of W (b p ) for each value of J . To distinguish
tional Hamiltonian, the four-group, which has the symme- the sublevels for a given J , the notation Jτ is employed,
try species A, Ba , Bb , and Bc . The form of the submatrices where τ takes on 2J + 1 integer values ranging from −J
is tridiagonal: to +J . These are assigned to the energy levels such that
√
E 00 2E 02 0 0 τ = −J corresponds to the lowest energy, τ = −J + 1 to
√2E E 24 0 the next lowest energy level, and finally τ = J to the high-
+ 02 E 22
E (κ) = , est level. The energy levels for an asymmetric rotor are
0 E 24 E 44 ·
thus characterized by the quantum numbers J and M and
0 0 · · by the pseudo-quantum number τ . An alternate notation
is more frequently employed where two subscripts are
E 22 E 24 0 0
E added to J , that is, JK −1 ,K 1 . A particular asymmetric-top
24 E 44 E 46 0 level converges to a particular prolate symmetric-top level
E − (κ) = ,
0 E 46 E 66 · specified by |K | as κ approaches −1 and to a particular
0 0 · · oblate symmetric-top level as κ approaches +1. The sub-
(43) script K −1 is therefore the |K | value of the prolate rotor
E 11 + E −11 E 13 0 0 limit, κ = −1, and K 1 is the |K | value of the oblate rotor
E 33 E 35 0 limit, κ = 1. The relation between the two notations is
E 13
0+ (κ) = ,
0 E 35 E 55 · τ = K −1 − K 1 . (47)
0 0 · · The possible energy levels for J = 1 may hence be denoted
as 1−1 , 10 , 11 , or more commonly as 10,1 , 11,1 , 11,0 . The
E 11 − E −11 E 13 0 0 first few rotational energy levels of an asymmetric top are
0
E 13 E 33 E 35 illustrated in Fig. 8.
0− (κ) = .
0 E 35 E 55 · For dipole absorption of radiation, the selection rules for
0 0 · · J are as follows, along with their general designations:
with the matrix elements calculated from Eq. (41), taking J → J + 1, R-branch transition,
cognizance of Eqs. (40) and (44). Since the matrices are J → J, Q-branch transition, (48)
Hermitian, (K ||K ) = (K ||K )∗ , we have the relations
J → J − 1, P-branch transition.
E K ,K = E −K ,−K ,
(44) The transitions J = 0, J = −1 can give rise to absorp-
E K ,K +2 = E K +2,K = E −K ,−K −2 = E −K −2,−K . tion lines for an asymmetric rotor, although not for a rigid
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
FIGURE 9 Low-resolution survey scan of the R-band spectrum for 1,2-dimethylenecyclobutane. Stark field
∼3000 V/cm, pressure ∼80 µm, and temperature approximately that of dry ice. Stark lobes are up and zero-field lines
are down.
the spectral patterns, as illustrated by Figs. 6 and 7, are sim- There are a number of useful aids which can help in
ple and easily recognizable. In excited vibrational states, establishing and confirming assignments. The Stark ef-
additional effects can arise (see Section VI), which com- fect, discussed in Section IX, can give Stark effect pat-
plicate the assignment. On the other hand, the spectrum of terns which, when resolvable, may be used to distinguish
an asymmetric rotor, even in the ground vibrational state, a Q-branch from either a P- or R-branch transition and
can be quite complex, as illustrated in Fig. 9. identify the lower J value of the transition. If the molecule
The analysis of any spectrum starts with the evaluation exhibits nuclear quadrupole splitting (see Section X), cer-
of the moments of inertia from a trial structure obtained, tain transitions can exhibit characteristic splitting patterns
for example, from similar molecules. Next, estimates are which can be helpful in assigning the spectrum. Spec-
made of the dipole components µa , µb , and µc from bond tral patterns, particularly for slightly asymmetric rotors,
dipole moments. These, along with the selection rules, can provide useful information for assignments. For ex-
indicate the types of transitions that will dominate the ample, the a-type, R-branch transitions of a near-prolate
spectrum. This information and the moments of inertia asymmetric rotor, JK −1 ,K 1 → (J + 1) K −1 ,K 1 +1 , give rise to
enable the rotational spectrum to be properly predicted. a closely spaced cluster of lines for a given J → J + 1.
The predicted and observed spectra can then be compared. Except for K −1 = 0, there are two transitions for each
Various tentative assignments of the measured lines can be K −1 (2J + 1 lines). Within a given cluster the intensity
made until a correct assignment is found which accounts of K −1 → K −1 lines decreases with increasing K −1 be-
for the details of the measured spectrum. In this way, the cause of the Boltzmann factor in the intensity expression.
rotational constants are evaluated which characterize the Also, the lower K −1 doublet transitions are more widely
molecular structure. split by the inertial asymmetry than the higher K −1 lines.
FIGURE 10 Low-resolution microwave spectrum of crotonic acid. Scan rate 10 MHz/sec, pressure about 15 mTorr,
1000-V Stark modulation. Spectra of two rotational isomers (s-trans and s-cis) are evident. Strongest bands are due
to s-trans. [Courtesy of Hewlett-Packard Co.]
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
1. Double Resonance
In the double-resonance technique, two radiation fields, a
pump (ν p ) and a signal (νs ), are introduced simultaneously
into the waveguide absorption cell. The pump radiation
is high powered, while the signal is low powered. When
both the pump and signal are in the microwave region,
the method is known as microwave–microwave double FIGURE 12 A simplified MMDR spectrometer.
resonance (MMDR). The process can be explained qual-
itatively by reference to the three-level energy system of proper operation of the spectrometer, a filter is used to
Fig. 11. The effect of the high-powered pump ν p (1 → 2) prevent the modulated pump radiation from reaching the
is to reduce the population of state 1 and hence decrease signal detector.
the intensity of the absorption signal νs (1 → 3). On the When the energy levels for the pump transition are
other hand, if the observed transition νs is the 2 → 3 tran- closely spaced, the pump frequency can be in the radiofre-
sition (not depicted), then ν p (1 → 2), which raises the quency region, and radiofrequency–microwave double-
population of state 2, would enhance the observed ab- resonance (RFMDR) experiments can be employed to
sorption signal. Hence, the pump radiation clearly affects simplify the spectrum. The K -doublets of slightly asym-
the microwave absorption signal by virtue of the fact the metric rotors are often split by a few megahertz and lie in
transitions share a common energy level. Knowledge of the RF region. Typically, a conventional Stark-modulated
these energy level connections provides a basis for spectral spectrometer is used, and the strong radiofrequency source
assignments. is square-wave modulated and applied across the Stark cell
Spectrometers have been built around this phenomenon instead of the usual Stark modulation. Without special
to simplify complicated spectra. For example, instead of impedance matching, the upper limit for the RF radiation
Stark modulation, the pump source may be modulated, and is about 100 MHz. Again, only signal transitions will be
this modulatates the signal absorption by means of the observed which share a common energy level with the
double-resonance effect. A narrowband amplifier tuned pump and are therefore modulated by the pump radiation.
to the modulation frequency is employed along with the Other types of experiments based on double-resonance
usual phase-sensitive detector. A simplified version of techniques provide alternate useful information as out-
such a spectrometer is shown in Fig. 12. Clearly, as the lined in Table VIII. Moreover, double-resonance tech-
signal frequency is swept, only transitions with a com- niques have been used with optical and infrared radiation
mon energy level with the fixed pump will be modulated to observe microwave transitions from excited electronic
and observed, thus greatly simplifying the spectrum. For states and excited vibrational states, respectively.
a. Rotational energy transfer. As indicated above, to which parameters are expected to differ significantly
rotational energy transfer can be studied by use of double- between conformers, or to unique structural effects, (iv)
resonance techniques. For example, by using infrared a basis to fix particular bond distances or angles or their
pump radiation with probe radiation in the millimeter- differences in the structural analysis, (v) initial structural
or submillimeter-wave region, both rotational and vibra- parameters for prediction of spectra of new and unusual
tional state-changing collisions can be studied. Here, a molecules, (vi) calculation of quadratic force constants
CO2 laser coincidence with an infrared molecular tran- for the estimation of quartic distortion constants or the
sition provides a nonthermal population distribution in a evaluation of average structures, (vii) estimation of the
particular rotational state of the excited vibrational state. dipole components µa , µb , and µc from which transition
The infrared radiation can be introduced into the measure- intensities may be computed, and (viii) calculation of the
ment cell along with the microwave probe radiation via a cubic force field to evaluate the rotation–vibration interac-
dichroic window. The probe radiation probes a single ro- tion constants αi and subsequently equilibrium rotational
tational transition within the excited vibrational state. The constants from the effective rotational constants.
change in intensity of the rotational line, or time response, The computation of semiexperimental equilibrium ro-
is recorded after the pumping laser is pulsed. The probe tational constants from calculated cubic force constants
hence monitors the population change as the system is allows the important equilibrium structure to be obtained.
perturbed and returns to equilibrium. Results of this approach have been quite satisfactory and
In the case of studies on the symmetric top CH3 F, for ex- this can be expected to be a very useful approach where
ample, the CO2 laser line 9P(20) pumps the CH3 F energy limited vibrational data are available.
level population from J = 12, K = 2 of the ground vibra- In the case of weakly bound complexes, ab initio calcu-
tional state to the v3 = 1, J = 12, K = 2 state. The excess lations can be very useful to limit the number of possible
population in the excited vibrational state produced by the configurations that need to be considered in the structural
pump is quickly redistributed by several CH3 F CH3 F col- determination. Furthermore, the ab initio results can help
lisional processes to reestablish rotational thermal equi- to remove other ambiguities which arise and to identify
librium within v3 = 1. Also, several vibrational collision possible large-amplitude motions.
processes reestablish equilibrium among the vibrational In addition, various microwave-derived internal rota-
levels. The measured time-varying responses for the vari- tion barriers and fine structure and hyperfine structure
ous rotational transitions provide data to characterize the coupling constants all provide a wealth of data to test the
complicated internal energy transfer. Both J -changing ro- accuracy of quantum mechanical calculations.
tational collisions, where v3 = 1, J = n, K = 0 (n inte-
ger), and K -changing rotational collisions, where v3 = 1,
E. Structural Information
J = m, K = 3n (m, n integers) have different behavior
but can be accurately modeled with four parameters. Structures of many hundreds of molecules—diatomic, lin-
Many more studies on rotational energy transfer can ear, symmetric top, and asymmetric top—have been eval-
be expected in the future. Pressure broadening effects ob- uated by microwave spectroscopy. Illustrative structures
served in rotational lines due to self-broadening or foreign are given in Tables III–V and XV (see also Section VIII).
gas broadening also provide information on energy trans- Structures of some small molecules are given in Table IX.
fer processes during molecular collisions; however, the Over the last 10 years, a considerable amount of work
information is much less than the method described here. has been devoted to the study of ions, radicals, molecu-
Various pressure broadening studies have been carried out lar complexes, and other transient molecular species due
over the years. Pressure broadening is discussed further in primarily to improved experimental techniques, which are
Section XI. discussed in Section XI. Results of some of these studies
will be addressed in this section. Structures of a few un-
stable molecules along with methods for their preparation
2. Ab Initio Quantum Calculations
are given in Table X. Methods of evaluating molecular
High-quality ab initio calculations have proven to be a structures and the complications arising from effects of
helpful aid in the analysis of rotational spectra, particularly molecular vibration are discussed in Section VIII.
for predictive purposes and where there exists ambigui- By means of the very precise structural parameters that
ties. In general, ab initio calculations can provide useful can be obtained from microwave spectroscopy investiga-
information for structural determinations. These include tions, trends in bond distances or angles among a series
(i) estimates of the most stable conformers and the low- of structurally similar molecules can be readily studied.
est energy form, (ii) insight into the difference expected At present the most reliable parameters for many large
in structural parameters between conformers, (iii) a guide molecules are those derived by the substitution method
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
[r0] Haubrich, S. T., Roehrig, M. A., and Kukolich, S. G. (1990). J. Chem. Phys. 93, 121
F
87.5 1.599
F CL 1.7007 F
[r0] Koput, J., Stroh, F., and Winnewisser, M. (1990). J. Mol. Spectrosc. 140, 31
CH3
109.5
1.4323
N 1.195 C 1.717 Se
C H 1.100
[rs] Gerry, M. C. L., Stroh, F., and Winnewisser, M. (1990). J. Mol. Spectrosc. 140, 147
1.175 1.316 1.158
C N C N
[re] Le Guennec, M., Chen, W., Wlodarczak, G., Demaison, J., Eujen, R., and Burger, H. (1991).
H J. Mol. Spectrosc. 150, 493
1.5242
106.4
Ge 1.7340 F
H
H
[re] Davis, R. W., and Firth, S. (1991). J. Mol. Spectrosc. 145, 225
F F
108.3
SF 87.5
1.6295
S S ∠FS
1.8889
[r0] Brier, P. N. (1991). J. Mol. Struct. 263, 133
Feg
1.566
Br 2.2
24 F
89.3
S
C4V 1.5
91
F Fax
F
[r0] Cordonnier, M., Bogey, M., Demuynck, C., and Destombes, J. L. (1992). J. Chem. Phys. 97, 7984
H1 .
1 629 H2
157.5
52.5 1.474
Si 1 . Si 2
2 119
[ r ] Hensel, K. D., Lam, M. E., Gerry, M. C. L., and Willner, H. (1992). J. Mol. Spectrosc. 151, 184
O.
1 1650
Br 118.0 C
173.1
1.8556
N 1.2224
[ r ] Xu, Y., Gerry, M. C. L., Joo, D. L., and Clouthier, D. J. (1992). J. Chem. Phys. 97, 3931
F 1.3182
106.89 C . S
1 5870
F
[ r ] Cox, A. P., Ellis, M. C., and Perrett, T. (1992). J. Chem. Soc. Faraday Trans. 88, 2611
H 1 .10
3 110.5
1.461 1.201 1.643
C C C Cl
H
H
[ r ] Cox, A. P., Ellis, M. C., Legon, A. C., and Wallwork, A. (1993). J. Chem. Soc. Faraday Trans. 89,
F 2937
1.337
1.474 1.201 1.051
C C C H
110.6
F
F
continues
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
TABLE IX (Continued )
[ r ] Cox, A. P., Ellis, M. C., Attfield, C. J., and Ferris, A. C. (1994). J. Mol. Struct. 320, 91
O
1.213 115.7
H
HNO3 N 102.6 0.941
1.198 1.41
114.1
0 O
O
[re] Esposti, C. D., Tamassia, F., Cazzoli, G., and Kisiel, Z. (1995). J. Mol. Spectrosc. 170, 582
2.144
N 1.133
Br 114.1 O
[rs] Klesing, A., and Sutter, D. H. (1995). J. Mol. Struct. 352, 357
H
1.085
122.51
C 114.72
77 O
1.0 1
H 115.63 .610 S 1.469
[re] Nakata, M., and Kuchitsu, K. (1995). J. Mol. Struct. 352, 219
H . C C Cl
1 063 1.209 1.631
[rs] Niedenhoff, M., Yamsda, K. M. T., Winnewisser, G., and Ross, S. C. (1995). J. Mol. Struct. 352, 423
H
123.9
a Type of structure (see Section VIII) is given in brackets. Bond lengths are in angstrom units, bond angles ABC in degrees. See also references
cited.
and the scaled method, and critical evaluation of pre- results are obtained for the acetylene moiety in H2 PC CH
dictive models should make use of these parameters if depicted in Fig. 13a. Here, the C CH group is bent away
possible. from the phosphine hydrogens by 7◦ . A small nonlinear-
Numerous small structural effects have also been re- ity in the NCO, NCS, and N3 chains is found in HNCO,
vealed by microwave structure studies. In this section a HNCS, HN3 , and CIN3 . Many other examples of slightly
few of these results are pointed out to provide the reader bent configurations have been found. Some are not easily
with an indication of what can be found from a study explained by simple bonding theory.
of the rotational spectrum. Also, some of the areas of Recent microwave measurements in the laboratory have
study are mentioned briefly to indicate the wide range aided the detection of many molecules in interstellar space.
of problems being attacked by microwave spectroscopy. We mention here the linear cyanide chains,
Many detailed structures and molecular information other
than those mentioned here have been obtained, and the H 1.0631 C≡≡≡≡≡≡N,
reader should consult the Bibliography and references 1.155
cited therein for further examples. H 1.058
C≡≡≡≡≡≡C 1.378
C≡≡≡≡≡≡N,
1.205 1.159
Linear b Cooper, T. A., Firth, S., and Kroto, H. W. (1991). J. Chem. Soc. Faraday Trans. 87, 1499
1.284 1.606
F B S
Linear b Cooper, T. A., Firth, S., and Kroto, H. W. (1991). J. Chem. Soc. Faraday Trans. 87, 1
1.831 1.608
Br B S
Symmetric topc Ahmad, I. K., Ozeki, H., and Saito, S. (1991). J. Chem. Phys. 110, 912
H
114.26
1.4406 P . O
1 4763
H
H
Planar d Bailleux, S., et al. (1997). J. Chem. Phys. 106, 10016
H 122.00 122.39 H
1.0819
C . Si
1 7039 1.4671
H H
Debridgede (butterfly) Bogey, M., Bolvin, H., Cordonnier, M., Demuynck, C., Destombes, J. L., and Csaszar, A. G.
HSi SiH 104.22 (1994). J. Chem. Phys. 100, 8614
H
H
1.6680
Si
Si
2.2154
FIGURE 14 Methylcyanopolyyne structural formulas and micro- FIGURE 15 A view of η5 -C5 H5 Cr(CO2 )2 NO. Some structural pa-
wave-determined rotational constants. For details of the analy- rameters are also given. Cp denotes η5 -C5 H5 ; asterisk indicates
sis see Chen, W., Grabow, J.-U., Travers, M. J., Munrow, M. R., assumed value. For details see Kukolich, S. G., McKay, R. T.,
Novick, S. E., McCarthy, M. C., and Thaddeus, P. (1998). J. Mol. Breckenridge, S. M., Flores, L. D., Morris, G. E., Sèckafoose, S. M.,
Spectrosc. 192, 1. and Morrison, D. L. (1995). Inorg. Chem. 34, 4182.
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
the type of coupling, and various coupling cases are pos- synthesis of the many neutral molecules so far identified
sible depending on the origin of the electronic angular in interstellar space.
momentum. The stable O2 molecule in a 3 electronic Some other molecular ions that have been studied more
ground state is an important example that has been ex- recently are listed in Table XII. Some examples of ionic
tensively studied. It has a spin angular momentum of complexes are also included in the table. In some cases, the
1h and a zero electronic orbital angular momentum. O2 analysis of the rotational spectrum was aided by the use of
does not have an electric dipole moment, but does have double resonance techniques or by information obtained
a magnetic moment associated with the spin angular mo- from ab initio calculations. Most ions have been positive
ment. The corresponding magnetic dipole transitions are, species (cation). Recently, the negative ion (anion) SH−
however, much weaker in general than electric dipole has been identified in the laboratory. The ion was produced
transitions. by the low-pressure electric discharge of H2 S and argon.
Free radicals, like ions, play an important role in var- The measurement of this and other light hydrides has been
ious chemical reactions. The understanding of chemical aided by the availability of higher frequency microwave
mechanisms, whether associated with combustion, the in- sources. The first two transitions of SH− fall at 0.56 and
terstellar medium, or other areas, can be clarified by de- 1.1 THz (see Section XI).
tection of radical intermediates.
Some recent studies of free radicals are given in
3. H-Bonded Complexes
Table XI. Here, we give the ground electronic state, per-
and van der Waals Complexes
tinent comments, and structural parameters for some of
the radicals. The reader is directed to the references of the Microwave spectroscopy studies have provided consider-
table for a more detailed discussion of the experimental able information on the phenomenon of hydrogen bond-
methods and analysis of the rotational spectrum. A num- ing. Earlier studies with conventional microwave spec-
ber of interesting radicals are given in the table, such as trometers were on large H-bonded complexes. Typical
cyclic C3 H, Cn O, C3 H, Cn H, and HC3 O. Also, we note that of these studies is the bimolecule trifluoroacetic acid–
the shorter OH distance in MgOH as compared to CaOH cyclopropane carboxylic acid,
and other alkaline earth hydroxide radicals indicates the
O H O
quasilinear nature of this molecule. Y C3H5
X C
The carbon chain radical cyanopropynyl
C Y O °
O 2.62 A,
H X CF3
O H O
C C C C N
H which has recently been studied in the 1- to 4-Ghz re-
gion, where the low-J transitions of this heavy complex
has recently been detected in the laboratory, and the de- appear.
rived rotational constants (in MHz) B = 2186.430 and A number of hydrogen-bonded complexes that form lin-
D = 139 × 106 provide the basis for astronomical detec- ear, symmetric, and asymmetric rotors have been studied.
tion. Similarly, observation has recently added C5 N to the Details have been obtained on the geometry, structural pa-
radical series CN and C3 N, which have been detected in rameters, strength of the intermolecular hydrogen bonds,
interstellar space. An initial radio astronomy search for dipole moment, and electric charge distribution of the
C5 N based on (in MHz) B = 1403.080 and D = 50 × 106 complexes. This information comes from measurements
failed to find lines from this radical. of rotational constants, centrifugal distortion constants,
The observation of molecular ions such as CO+ , HCO+ electric dipole moments, and nuclear hyperfine structure.
(see Table III), HOC+ , HNN+ , NO+ , and HCS+ represents Tunneling effects and large vibrational motions for such
a relatively new area of study by microwave spectroscopy. complexes can complicate the analysis. By way of illustra-
These ions have also been detected in outer space by mi- tion, Tables III and IV give the structures of a few selected
crowave astronomers. Molecular ions represent a signif- complexes. Note the result of OC· · ·HF is consistent with
icant challenge to microwave spectroscopists, as attested the polarity, O+ C− .
to by the small number of ions so far analyzed. The major In a similar way, an ever-expanding body of knowledge
problem is in the production of a sufficient concentration is being obtained on inert-atom–molecule complexes.
of ions for detection (see Section XI). The chemistry pro- These are weakly bonded van der Waals complexes in-
duced by molecular ions is important to a number of areas volving the rare gas atoms Ne, Ar, Kr, and Xe. The forma-
such as plasmas, flames, the upper atmosphere, and the in- tion of weakly polar heteronuclear rare gas dimers allow
terstellar medium. In the latter case, they play a role in the the pure rotational spectra to be observed. In Ne· · ·Ar,
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
Electronic
Radical ground state Information/comments Reference
H2 NO 2B Spectroscopic constants, essentially planar, Mikami, H., Saito, S., and Yamamoto, S. (1991). J. Chem. Phys.
1
C2v , r0 (NH) = 1.01 (assumed), r0 (NO) = 94, 3415
1.280, HNH = 122.7
CH2 N 2B
2 Spectroscopic constants, planar, r (CH) = Yamamoto, S., and Saito, S. (1992). J. Chem. Phys. 96, 4157
1.08 (assumed), r (CN) = 1.261, HCH =
122.3
c-C3 H 2B Spectroscopic constants, C2v , planar cyclic Yamamoto, S., and Saito, S. (1994). J. Chem. Phys. 101, 5484
2
triangular structure, rs (CH) = 1.0760,
rs (CC) = 1.3771, rs (C CH) = 1.3739
HCCCO 2 A Spectroscopic constants, planar, Cooksy, A. L., Watson, J. K. G., Gottlieb, C. A., and Thaddeus,
7Å 1.1 P. (1994). J. Chem. Phys. 101, 178
1.219Å 1.38 C 92Å
1.060Å C C O
H 136.5
168 197
+
MgOH 2 Spectroscopic constants, linear, r0 (MgO) = Nuccio, B. P., Apponi, A. J., and Ziurys, L. M. (1995). J. Chem.
1.780, r0 (OH) = 0.871b Phys. 103, 9193
+
CaOH 2 Spectroscopic constants, linear, r0 (CaO) = Nuccio, B. P. et al. (1995)
1.985, r0 (OH) = 0.922
−
FeCO 3 Spectroscopic constants, linear, rs (FeC) = Kasai, Y., Obi, K., Ohshima, Y., Endoy, Y., and Kawaguchi, K.
1.7268, rs (CO) = 1.1599 (1995). J. Chem. Phys. 103, 90
−
Cn O (n = 2, 4, 6, 8) 3 Spectroscopic constants, linear, effective Ohshima, Y., Endo, Y., and Ogata, T. (1995). J. Chem. Phys. 102,
C C bondc : 1.3704, 1.2920, 1.2830, 1493
1.2802 for n = 2,4,6,8, respectively
C3 H (C1 ,C2 C3 H) 2
r Spectroscopic constants, linear,d rs (C1 C2 ) = Kanada, M., Yamamoto, S., Saito, S., and Osamura, Y. (1996).
1.3263, rs (C3 H) = 1.0171, rs (C2 C3 ) = J. Chem. Phys. 104, 2192
1.2539
CaSH 2 A Spectroscopic constants, bent, r0 (CaS) = Taleb-Bendiab, A., Scappini, F., Amano, T., and Watson, J. K. G.
2.564, r0 (SH) = 1.357, CaSH = 91.0 (1996). J. Chem. Phys. 104, 7431
FS2 2 A Spectroscopic constants, bent, r0 (FS) = Tang, J., and Saito, S. (1996). J. Chem. Phys. 104, 7437;
1.595, r0 (SS) = 1.915, FSS = 107.8 see also Zhuo, Q., et al. (1994). J. Chem. Phys. 100, 6113
H2 PO 2 A Spectroscopic constants, pyramidal, Hirao, T., Saito, S., and Ozeki, H. (1996). J. Chem. Phys. 105,
r0 (PO) = 1.4875, r0 (PH) = 1.4287, 3450
HPO = 115.5, HPH = 102.6
−
AsH 3 Spectroscopic constants, re = 1.52237 Fujiwara, H., Kobayashi, K., Ozeki, H., Saito, S., and Jaman,
A. I. (1997). J. Chem. Soc. Faraday Trans. 93, 1045
FeF 6 Spectroscopic constants, re (FeF) = 1.7803 Allen, M. D., and Ziurys, L. M. (1997). J. Chem. Phys. 106,
i
3494
+
MgBr 2 Spectroscopic constants, re (MgBr) = 2.3474 Walker, K. A., and Gerry, M. C. L. (1997). J. Chem. Phys. 107,
9835
NH2 2B
1 Spectroscopic constants, bent, re (NH) = Kobayashi, K., Ozeki, H., Saito, S., Tonooka, M., and
1.0254, HNH = 102.85 Yamamoto, S. (1997). J. Chem. Phys. 107, 9289
HSiO 2 A Spectroscopic constants, bent, r : r (SiO) = Izuha, M., Yamamoto, S., and Saito, S. (1997). J. Mol. Struct.
1.5326, r (SiH) = 1.5066 (assumed) 413, 527
HSiO = 116.8
−
PH 3 Spectroscopic constants, r = 1.43283 Klisch, E., Klien, H., Winnewisser, G., and Herbst, E. (1998).
Z. Naturforsch. 53a, 733
Cn H (n = 10, 2 Spectroscopic constants, linear carbon chain Gottlieb, C. A., McCarthy, M. C., Travers, M. J., Grabow, J.-U.,
12,13,14) and Thaddeus, P. (1998). J. Chem. Phys. 109, 5433
aBond lengths in angstrom units, bond angles ABC in degrees.
Spectroscopic constants: Rotational constants, distortion constants, angular momentum coupling fine structure terms, hyperfine coupling constants,
etc. Not all constants are reported for a given molecule. See also references cited.
b Short bond may indicate quasilinear character.
c Calculated assuming all C C bonds in the molecule are equal.
d Short C H bond is evidence for quasilinear character.
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
SH− Spectroscopic constants; method to distinguish between positively Civis, S., Walters, A., Yu Tretyakov, Yu., Bailleux, S., and Bogey,
and negatively charged ions and neutrals in plasma M. (1998). J. Chem. Phys. 108, 8369
HN+
2 Spectroscopic constants, quadrupole coupling, infrared-microwave Ho, W. C., Pursell, C. J., Weliky, D. P., Takayi, K., and Oka, T.
double resonance (1990). J. Chem. Phys. 93, 87
PO+ Spectroscopic constants, re = 1.424993 Å, ωe = 1411.5 cm−1 , Petrmichl, R. H., Peterson, K. A., and Woods, R. C. (1991).
ab initio calculations, comparison to SiF+ J. Chem. Phys. 94, 3504
XeH+ Spectroscopic constants, re = 1.6028 Å, ωe = 2270.2 cm−1 , Peterson, K. A., Petrmichl, R. H., McClain, R. L., and Woods,
quadruple coupling, Zeeman analysis, spin–rotation coupling R. C. (1991). J. Chem. Phys. 95, 2352
He· · ·Ar+ Zeeman analysis, microwave–microwave double-resonance study Carrington, A., et al. (1995). J. Chem. Phys. 102, 2379
FN+
2 Spectroscopic constants, ab initio calculations, r (F N) = 1.2461 Å, Botschwina, P., Sebald, P., Bogey, M., Demuynck, C., and
r (N N) = 1.1034 Å Destombes, J.-L. (1992). J. Mol. Spectrosc. 153, 255
FCO+ Spectroscopic constants, ab initio calculations, r (F C) = 1.2014 Å, Botschwna, P., et al. (1992)
r (C O) = 1.1151 Å
Ar· · ·H+
3 Spectroscopic constants, large-amplitude internal rotation motion, Baillex, S., et al. (1998). J. Mol. Spectrosc. 190, 130
120◦ planar rotation of the H+3 group relative to Ar atom;
r = 2.3852 Å is the distance from the center of mass of the H+
3
triangle to Ar
HNCCN+ Spectroscopic constants, linear configuration Amano, T., and Scappini, F. (1991). J. Chem. Phys. 95, 2280
a See also references cited. Spectroscopic contants: Rotational constants, distortion constants, various other interaction constants.
for example, the small induced dipole is 0.0022 D. These 4. Rotational Isomerism and Ring Conformations
and other weakly bonded complexes are usually pro-
duced by the rapid expansion of a pressurized mixture Microwave spectroscopy has also provided considerable
of the constituents through a supersonic nozzle into an information on rotational isomerism and conformations
evacuated cell of the spectrometer (see Section XI). In- of ring compounds. The existence of one or more distinct
formation like that outlined above for H-bonded com- rotational isomers for a large number of molecules has
plexes can also be attained for these complexes. These been demonstrated. Similarly, the stable conformations
studies of weakly bonded complexes provide important for numerous ring compounds have been derived. Detailed
information about intermolecular forces. Recently, the structural information has also been obtained for many
range of complexes studied has been extended to include rotamers and ring systems.
clusters such as (Ar)n HF (n = 2,3,4) and (CO2 )n HF The sensitivity of the rotational constants to the ring
(n = 2,3). conformation is illustrated in Fig. 16. The boat confor-
In the analysis of complexes, the accurate structures, mation is clearly the correct conformation. Similarly, the
barriers to internal rotation, etc., of the molecular species rotational isomers evident in Fig. 10 may be identified
involved in complex formation are important in order to by comparison of the observed and calculated separation
understand the subtle changes that may occur upon com- (B + C) between the bands of a given series.
plexation. Also, it may be noted that for a complex formed Distortion of small rings due to substitution has also
by an asymmetric-top molecule and a rare gas atom, there been observed. In going from cyclopropene,CH CHCH2 ,
are, in general, eight structures which are compatible with to 3,3-difluorocyclopropene, CH CHCF2 , for example,
the moments of inertia. This ambiguity can often be clar- the C C double bond is increased in length by 0.025 Å,
ified or reduced by use of symmetry arguments or other while the other ring bonds are shortened. On the
parameters such as the dipole moment, nuclear quadrupole other hand, axiridine, CH2 CH2 NH, and chloroaziridine,
structure, isotopic information, or ab initio calculations. CH2 CH2 NCl, are remarkable in that detailed structural
The latter has been particularly useful in clarifying struc- studies reveal that the ring structures are virtually iden-
tural ambiguities and providing insight into the most stable tical. Isoxazole,OCH CHCH N, is a good example of
form. A large number of weakly bound complexes have the tendency of C H bonds adjacent to ring heteroatoms
been studied by microwave spectroscopy. to tilt slightly toward these atoms.
Table III gives a few molecular structures. In addi- A detailed study of various isotopomers of 1,2,3-
tion, Table XIII provides the geometries of some selected trifluorobenzene showed that the bond angles at C1 and
complexes. C3 are enlarged, while the C1 C2 and C2 C3 bonds are
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
Triangular Bumgarner, R. E., and Kukolich, S. G. (1987). J. Chem. Phys. 86, 1083
Planar, T-shaped, C2V Legon, A. C., Aldrich, P. D., and Flygare, W. H. (1981). J. Chem. Phys.
75, 625
Nonplanar, cyclic Bumgarner, R. E., Pauley, D. J., and Kukolich, S. G. (1987). J. Chem.
Phys. 87, 3749
Planar, T-shaped, C2V Gutowsky, H. S., Klots, T. D., Chuang, C., Schmuttenmaer, C. A., and
Emilsson, T. (1987). J. Chem. Phys. 86, 569
C3V symmetric top, H F along C3 Gutowsky, H. S., Klots, T. D., Chuang, C., Keen, J. D., Schmuttenmaer,
axis of Ar3 group, H closest to Ar3 C. A., and Emilsson, T. (1987). J. Am. Chem. Soc. 109, 5633
Symmetric top, HCN along C3 axis of Gutowsky, H. S., Hajduk, P. J., Chuang, C., and Ruoff, R. S. (1990).
cyclic (CO2 )3 group, N closest to J. Chem. Phys. 92, 862
(CO2 )3
Linear HCN trimer Ruoff, R. S., Emilsson, T., Klots, T. D., Chuang, C., and Gutowsky,
H. S. (1988). J. Chem. Phys. 89, 138
T-shaped, C2v symmetry Lida, M., Ohshima, Y., and Endo, Y. (1991). J. Chem. Phys. 94, 6989
Ar above thiazole ring shifted toward Kretschmer, U., Stahl, W., and Dreizler, H. (1995). J. Mol. Struct.
N atom 352, 289
T-shaped, C2v symmetry Mader, H., Heineking, N., Stahl, W., Jager, W., and Xu, Y. (1996).
J. Chem. Soc. Faraday Trans. 92, 901
S· · ·Cl Cl collinear, Cl2 nearly Bloemink, H. I., Dolling, S. J., Hinds, K., and Legon, A. C. (1995).
perpendicular to H2 S plane J. Chem. Soc. Faraday Trans. 91, 2059
Water-donor, methanol-acceptor complex Stockman, P. A., Blake, G. A., Lovas, F. J., and Suenram, R. D. (1997).
J. Chem. Phys. 107, 3782
Induced dipole ∼0.01 D Jager, W., Xu, Y., and Gerry, M. C. L. (1993). J. Chem. Phys. 99, 919
Heavy atoms, approximately collinear, Yaron, D., Peterson, K. I., Zolandz, D., Klemperer, W., Lovas, F. J., and
H2 O· · ·O hydrogen bond Suenram, R. D. (1990). J. Chem. Phys. 92, 7095
Xe, Kr complexes T-shaped, Ne complex Walker, K. A., Ogata, T., Jager, W., Gerry, M. C. L., and Ozier, I.
tilted as shown (1997). J. Chem. Phys. 106, 7519
continues
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
C2v symmetry, triangular trimer Xu, Y., Jager, W., and Gerry, M. C. L. (1994). J. Chem. Phys. 100, 4171
Cs symmetry, with terminal oxygens of Gillies, J. Z., Gillies, C. W., Lovas, F. J., Matsumura, K., Suenram, R. D.,
O3 tilted toward axis of acetylene Kraka, E., and Cremer, D. (1991). J. Am. Chem. Soc. 113, 6408
a See also Novick, S. E., Leopold, K. R., and Klemperer, W. (1990). “Atomic and molecular clusters.” In “The Structures of Weakly Bond Complexes
As Elucidated by Microwave and Infrared Spectroscopy” (E. R. Bernstein, ed.), Chapter 3, Elsevier, New York.
slightly shortened compared to the opposite end of the monothioformic acid, HCOSH. In going from the trans to
benzene ring: the cis rotamer, the CSH and the HCS angles increase by
F more than 2◦ , whereas the OCS angle decreases by 3.4◦ .
The SH bond length shortens by 0.02 Å.
F 2
F
1 3 Similarly, subtle changes in structure which occur in
different isomers are illustrated for 3,3-difluoroacryloyl
fluoride in Fig. 17, where the C3 C2 C1 and C2 C1 F1 angles
Small changes in structural parameters in going from increase, while C2 C1 O decreases in going from the syn to
one rotamer to another may be illustrated by cis- and trans- the anti conformer.
Microwave spectroscopy has also provided direct ev-
idence as to the importance of intramolecular hydrogen
bonding in determining the conformations (often the more millimeter-, and submillimeter-wave regions. Active in-
stable one) of certain rotamers. For methoxyacetic acid, terplay between laboratory observations and astronomical
CH3 OCH2 COOH, for example, an intramolecular hydro- observations has helped to drive both areas.
gen bond is formed between the hydroxyl hydrogen and Laboratory measurements have been driven by con-
the ether oxygen, giving a planar five-membered ring. tinued improvements in experimental techniques such as
rapid cooling Fourier transform microwave spectroscopy
(see Section XI). This technique, coupled with improved
5. Methyl Group: Tilts and Asymmetry
species production methods, has provided capabilities to
Numerous examples of methyl group tilts have been found identify an increasing number of exotic molecular species.
where the symmetry axis of the methyl group does not Moreover, recent advances in methods to generate high-
correspond with the bond direction. The effect has been frequency microwave radiation, such as high-frequency
demonstrated from analysis of internal rotation splittings backward wave oscillators, have enabled laboratory mea-
and directly from structure determinations. In methyl ni- surements in the terahertz region (1 THz = 1000 GHz,
trate, CH3 ONO2 , for example, the CH3 groups is tilted to- equivalent to λ = 0.3 mm). This technological develop-
ward the unshared electron pair by 4.8◦ . The NO2 group is ment will enable, for example, the direct detection of light
also found to be tilted by 3.9◦ away from the methyl group. hybrids for which even the lowest rotational transitions
Asymmetric methyl groups for which the hydrogen are at high frequencies. In the case of the radical NH,
atom in-plane (plane of the heavy-atom skeleton) has a for example, with a 3 electronic ground state, the
slightly smaller C H bond length than the out-of-plane ground-state transition is split into three fine-structure
hydrogens have been reported. Examples are methyl ni- transitions at frequencies 946, 974, and 1000 GHz.
trate and thioacetaldehyde. Each of these transitions is split by quadruple (14 N) and
magnetic hyperfine (H) interactions.
The detection of molecules via their rotational spec-
6. Quasi-Linear Molecules
tra allows astrophysicists to probe interstellar clouds to
Finally, we mention HCNO, where the ground-state spec- provide information on their environment, star formation,
trum is consistent with a linear structure, but the equilib- interstellar chemistry, mechanisms for synthesis and de-
rium structure is slightly bent. It has been found that the struction of interstellar molecules, isotopic distributions,
bending potential function has a small hump well below etc.
the ground vibrational level at the linear configuration. Continued and increased advances in our knowledge of
HCNO is one of a handful of molecules, called quasi- interstellar space can be expected. The satellite FIRST (Far
linear molecules, with this characteristic behavior. Other Infrared and Submillimeter Space Telescope) is planned
examples are HNCO, HNCS, and HNCSe. In a similar to be launched by 2003 and will cover the frequency range
way, CH3 NCO, CH3 NCS, and CH3 NCSe are examples of 300–3000 GHz (λ = 1–0.1 mm). This telescope will also
quasi-symmetric-top molecules. provide information on planets and comets. These satel-
lite observations, coupled with high-frequency microwave
studies, will enable the identification of a host of other
7. Interstellar Space Applications
molecules in space.
The region between the stars contains interstellar clouds
composed primarily of dust and gas. Over 100 molecules
(neutrals, ions, and radicals), mostly carbon-containing V. CENTRIFUGAL DISTORTION EFFECTS
compounds, have been identified in interstellar molecular
clouds. The largest molecule contains 13 atoms, the carbon The rigid-rotor treatment discussed in the previous section
chain HC11 N. accounts for the general features of the rotational spec-
Interstellar spectroscopy is usually carried out with trum. These gross features are modified somewhat when
a single-dish radio telescope composed of a precision the effects of nonrigidity, nuclear coupling, and so forth
parabolic reflector with a highly sensitive microwave de- are taken into account. In this section, the effects of cen-
tector at the focal point of the reflector. Molecular identi- trifugal distortion are considered.
fications, usually via rotational emission frequencies, are The centrifugal force produced by rotation distorts the
made by comparison with precise laboratory frequency molecule from its equilibrium configuration and the bond
measurements or frequency predictions. Microwave spec- distance and angles change slightly. Hence, the rotational
troscopy plays a very important role in providing the data spectrum is no longer characterized by a set of equilibrium
for identification of new molecular species. Identifica- moments of inertia. Additional terms in the Hamilitonian
tions have been made mostly with data in the microwave, are required to account adequately for the observed
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
spectrum. If distortion is taken into account, terms in the ν = 2B(J + 1) − 4D J (J + 1)3 − 2D JK (J + 1)K 2 . (54)
angular momentum components of the fourth power, sixth
power, and so forth are introduced into the Hamiltonian. Note that neither A nor D K affects the rotational spec-
Although the centrifugal distortion constant are very small trum. The first correction term, involving D J , alters the
relative to the rotational constants, they produce signif- even spacing between successive J → J + 1 transitions,
icant effects in the rotational spectrum, particularly for while the last term also separates the superposed lines of
asymmetric tops. different K values into J + 1 closely spaced lines with the
separation increasing as K 2 . This is illustrated in Fig. 7.
A. Diatomic and Linear Rotors Typical values of D J and D JK are collected in Table IV.
D J is always positive, whereas D JK and D K may be pos-
For diatomic and linear rotors, distortion terms of the type itive or negative.
DP 4 and HP 6 are added to the rigid-rotor Hamilitonian. Higher order effects introduce additional distortion
Considering only the major P 4 effect and noting (J, terms and also distortion terms that can give rise to
M|P 4 |J, M) = h 4 J 2 (J + 1)2 , we have for the rotational splittings of certain K -levels. The effects of centrifugal
energy of a nonrigid diatomic or linear molecule distortion on the observation of forbidden K = ±3 tran-
E J = BJ (J + 1) − D J J 2 (J + 1)2 (50) sitions have already been mentioned. Induced dipole mo-
ments also allow the observation of pure rotational spectra
and the frequencies are given by of spherical tops which, because they have no permanent
ν = 2B(J + 1) − 4D J (J + 1)3 , (51) dipole moment, would otherwise have no rotational spec-
tra. For CH4 , the distortion moment is on the order of
where D J is the distortion constant, and both B and D J 5 × 10−6 D. Both J → J + 1 and J → J transitions have
are in frequency units. The physical picture of distortion been observed. The leading terms in the frequency equa-
in a diatomic molecule is quite simple. As the molecular tion for the J → J + 1 transitions are like those for a linear
bond stretches, I increases and B decreases, leading to molecule, Eq. (51); however, the molecular distortion in
a decrease in the rotational energy and a shift to lower such molecules is more complicated, and additional terms
frequency relative to the rigid rotor frequency. If P 6 are required to adequately characterize the rotational spec-
effects are considered, then a term H J J 3 (J + 1)3 must trum. Such observations have provided the rotation and
be added to Eq. (50). Distortion effects are small and distortion constants. Some examples of nonpolar molecu-
primarily important only for high J values. Some typical lar studies via microwave spectroscopy are spherical tops
values of distortion constants are listed in Table III. with Td symmetry like CH4 , SiH4 , and GeH4 and those
with D3h symmetry like BF3 and SO3 . For SO3 , a planar
B. Symmetric-Top Molecules molecule, the centrifugally induced rotational spectrum
A first-order treatment of centrifugal distortion yields for provides re = 1.4175 Å.
the distortion Hamiltonian
d = −D J P 4 − D JK P 2 PZ2 − D K PZ4 , (52) C. Asymmetric-Top Molecules
which is diagonal in the symmetric-top basis since the The evaluation of centrifugal distortion in asymmetric ro-
only nonvanishing matrix elements (in units of h) are tors is considerably more complex than for linear or sym-
(J, K , M|P 4 |J, K , M) = J 2 (J + 1)2 , (J, K , M|P 2 PZ2 |J, metric tops, and because of the nature of the spectrum, par-
K , M) = K 2 J (J + 1), and (J, K , M|PZ4 |J, K , M) = K 4 . ticularly large distortion shifts (say 1000 MHz or larger)
The energy of a nonrigid prolate symmetric top is can be observed. A first-order treatment of the P 4 distor-
tion effects gives, for the energy of a semirigid prolate
E J,K = BJ (J + 1) + (A − B)K 2 − D J J 2 (J + 1)2
asymmetric rotor (Z ↔ a),
− D JK J (J + 1)K 2 − D K K 4 . (53)
E = Er + Ed (55)
For an oblate symmetric top, the unique axis is designated
c, and the energy expression may be obtained from the E r = (1/2)(B + C)J (J + 1)
above expression by replacement of A by C. The con- + [A − (1/2)(B + C)]W (b p ) (56)
stants D J , D K , and D JK essentially represent the distor- !
tion effects of end-over-end rotation, rotation about the E d = − J J (J + 1) − JK J (J + 1)
2 2
PZ2
symmetry axis, and the interaction between these motions, ! !
− K PZ4 − 2δ J σ J (J + 1) W (b p ) − PZ2
respectively. With the selection rules J → J + 1, K → K , ! !
the rotational frequencies are found to be − 2δ K σ W (b p ) PZ2 − PZ4 , (57)
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
h m 4Be3
stants that now contain a small contribution involving the DJ = = , (59)
distortion constants. The J , J K , and so on are the quar- h f (Ie )3 ωe2
tic distortion coefficients. One may apply the above ex- with m the reduced mass; Ie and Be are, respectively,
pressions to an oblate top (Z ↔ c) by interchanging A and the equilibrium moment of inertia and rotational constant
C and setting σ = −1/bo in the above energy expression. Be = h/8π 2 Ie . Here ωe = (1/2π )( f /m)1/2 is the harmonic
To evaluate the rotational and distortion constants, dif- vibrational frequency. Thus from an analysis of the rota-
ferences between the observed and calculated rigid-rotor tional spectrum, precise values of D J can be obtained,
frequencies are analyzed via Eq. (55) by means of the least- which in turn yield, from the above expression, accurate
squares technique to determine adjustments in the original stretching force constants, or equivalently, ωe values.
rotational constants δA, δB, and δC as well as the distor- For other molecules, the details are more complicated,
tion constants J and so on. Quartic distortion constants but the principles are the same. The quadratic potential
for a few asymmetric tops are collected in Table V. function has the general form
The energy expression given in Eq. (57) can account
for a large number of asymmetric tops. However, for light 1
V = f i j Ri R j . (60)
molecules with large rotational energies, such as H2 O, 2
or when transitions from high J levels are studied, a
Infrared measurements yield the vibrational frequencies
first-order treatment does not suffice. Additional higher
associated with the various normal vibrational modes,
power terms in the angular momentum must be included
and these data, including isotopic frequency data, can be
in Eq. (57). Specifically, it is found that n + 1 distortion
used to evaluate the force constant matrix F = [ f i j ]. Since
contributions are added for each degree n in the angular
these calculations are often ill conditioned and also since
momentum. Thus seven terms are added if P 6 effects are
there are usually more force constants than vibrational
considered. The sextic distortion constants are denoted by
frequency data, both infrared data and the microwave dis-
J , JK , KJ , K , φ J , φ JK , and φ K and have been eval-
tortion constant data are often combined to help charac-
uated from a study of the rotational spectra for a large
terize the force constant matrix. Some examples are given
number of molecules. When such effects are important, a
in Table XIV. It may be noted that the P 6 or sextic dis-
first-order treatment is not sufficient. In such cases, the en-
tortion constants depend on the cubic potential constants,
ergy matrix of r + d must be set up and diagonalized to
and these data have been employed to obtain information
obtain the general effects of centrifugal distortion. Proce-
on these anharmonic potential constants.
dures for effectively including such higher order distortion
effects are discussed elsewhere. TABLE XIV Potential Constants Determined by
Combination of Infrared and Microwave Data
(mdyn/A)a
D. Information from Distortion Constants
Molecule fr f rr f αr 2 f r αr
The study of centrifugal distortion provides a number
of useful kinds of information. By including effects SO2 10.006 0.024 0.793 0.189
of centrifugal distortion, one can obtain very accurate O3 5.70 1.52 1.28 0.332
spectroscopic constants. These allow the prediction of OF2 3.950 0.806 0.724 0.137
unmeasured transition frequencies with a high degree of ClO2 7.018 −0.170 0.651 0.006
confidence over a wider range than provided by only NO2 11.043 2.140 1.109 0.481
the rigid-rotor constants. Most importantly, however, the GeF2 4.08 0.26 0.316 −0.01
centrifugal distortion constants provide information on SeO2 6.91 0.03 0.488 0.009
the vibrational potential function, particularly for small a The quadratic-valence force field potential function
molecules. This follows because the distortion constants for bent triatomic molecules XY2 is defined by
depend directly on the force constants, masses, and struc-
2V = f r δr12 + δr22 + f α δα 2 + 2 f rα (δr1 + δr2 )δα
ture of the molecule. This may be illustrated for a diatomic
molecule. The quadratic potential function is given by + 2 f rr δr1 δr2 .
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
VI. ROTATION–VIBRATION INTERACTIONS arate rotational spectrum is obtained for each vibrational
state. These excited-state lines, or satellite spectra, may
In addition to centrifugal distortion effects, other nonrigid- be shifted only a few megahertz or many hundreds of
ity effects also alter the rotational spectra. The vibrational megahertz away from the ground-state line. Because of
motions of a polyatomic molecule may be described in the Boltzmann factor e−Ev /kT , the line intensity decreases
terms of n normal modes of vibration, where n = 3N − 6 with increasing vibrational excitation, and only low-lying
(or 3N − 5 for a linear molecule) with N the number of vibrational levels give rise to lines with sufficient inten-
atoms. For linear and symmetric tops, degenerate vibra- sity to be observed. Except for diatomic molecules and
tions are present, and not all of these modes have different a few relatively simple polyatomic molecules, it is not
vibrational frequencies ωi . The modes with the same ωi possible to obtain sufficient data to determine all the αi
are usually grouped together and their number specified in Eqs. (61)–(63). Hence, it is not possible to correct the
by di . As a molecule rotates, it vibrates rapidly, even in observed ground-state rotational constants to obtain the
the ground vibrational state, and the moments of inertia equilibrium constants, for instance,
are averaged in a complicated way over the molecular vi-
Ae = A0 + αia di 2. (65)
brations. The rotational constants as well as the distortion
constants, and in fact almost all the molecular parame- Because of this, effective constants must be used to evalu-
ters derived from an analysis of rotational spectra, must ate the molecular structure, which introduces uncertainties
be considered as effective constants, that is, constants as- in the derived structural parameter. This is discussed fur-
sociated with a particular vibrational state. The depen- ther in Section VIII.
dence of the effective rotational constants on the vibra- Unless there is an accidental near-vibrational degener-
tional state v may, to a good approximation, be expressed acy, the rotational spectrum of an asymmetric top in an
by excited vibrational state is similar to that obtained in the
di
ground state, except that the spectrum is characterized
Av = Ae − αia vi + , (61) by a slightly different set of rotation and distortion con-
2
stants. Other nonrigid effects are often more important
di
v (=0, 1, 2, . . .) and J (=0, 1, 2, . . .) are, respectively, the Furthermore, the effects of vibration on Bv and Dv are
vibrational and rotational quantum numbers, ωe is the har- given by Eqs. (62) and (64). However, the rotational spec-
monic vibrational frequency, and ωe χe is the anharmonic- trum is complicated by the presence of degenerate bending
ity constant. The effective rotation and distortion constants modes of vibration. The bending mode is twofold degen-
are defined by erate since the linear molecule may bend in either of two
2 orthogonal planes. In the case of OCS, for example, there
Bv = Be − αe v + 12 + γe v + 12 + · · · , (67) are 3 · 3 − 5 = 4 vibrational modes, labeled v1 , v2 , and v3
Dv = De + βe v + 12 + · · · , (68) with the bending mode twofold degenerate, d2 = 2. With
excitation of a single degenerate bending mode v j , an an-
Hv = He + · · · . (69) gular momentum p = lh is generated along the molecular
The αe , γe , and βe are the rotation–vibration interaction axis analogous to a symmetric top with l similar to K . The
constants representing corrections for the effect of vibra- possible values of l are
tion. The selection rules for pure rotational transitions are l = v j , v j−2 , v j−4 , . . . , −v j . (74)
J → J + I, v → v, and the rotational frequencies are eas-
ily shown to be Hence, for v j = 1, l = ±1; v j = 2, l = 0, ±2; and so on. In
addition, a Coriolis interaction between rotation and vibra-
ν = 2Bv (J + 1) − 4Dv (J + 1)3 + Hv (J + 1)3 tion exists that can remove the ±l degeneracy when l = 0.
× [(J + 2)3 − J 3 ]. (70) The linear molecule behaves in an excited bending state
as if it were slightly bent; and, like a slightly asymmetric
To evaluate all of the constants, measurements of rota-
rotor, where the ±K degeneracy is lifted, the ±l degen-
tional transitions in at least three vibrational states (e.g.,
eracy is lifted. This is called l-type doubling. A detailed
v = 0, 1, and 2) must be made. Each vibrational state is
treatment for the energies including the l-type splitting of
analyzed via the above equation. For example, from the
the levels gives
data B0 , B1 , and B2 , Eq. (67) yields
E = Bv [J (J + 1) − l 2 ] − Dv [J (J + 1) − l 2 ]2
Be = 18 (15B0 − 10B1 + 3B2 ). (71)
Some spectroscopic constants obtained for a few selected ± 14 q j (v j + 1)J (J + 1), (75)
diatomic molecules are collected in Table XV. Information where q j is the coupling constant characterizing, the split-
on the vibrational constants can also be obtained from the ting for the bending mode v j . This constant is usually sig-
rotational constants, for example, nificant only for the case |l| = 1. Here Bv and Dv have
4Be3 their usual meaning. If the splitting term is omitted, it is
ωe2 = , (72) apparent that the levels are doubly degenerate since they
De
2 depend on l 2 , except when l = 0. However, J represents
αe ωe the total angular momentum quantum number including
ωe χe = Be +1 . (73)
6Be2 the vibrational angular momentum. Hence,
J = |l|, |l| + 1, |l| + 2, . . . , (76)
B. I-Type Doubling in Linear Molecules
and depending on l, certain values of J are missing. In
For linear molecules, the rotational frequencies in excited particular, for |l| = 1, J = 1 is the lowest value of J , while
nondegenerate vibrational states are specified by Eq. (51). for |l| = 2, J = 2 is the lowest value. As a result of this,
Diatomic
molecule Be (MHz) −αe (MHz) D e (kHz) ω e (cm−1 ) ω e χe (cm−1 ) r e (Å)
for example, the 0 → 1 transition is missing in an excited ν = (1/2)q j (v j + 1)J (J + 1). (78)
bending mode with |l| = 1, and the 0 → 1 and 1 → 2 tran-
With sufficiently large q j or J , these transitions can
sitions are missing for |l| = 2. With the selection rules
fall into the microwave region. For HCN, v2 = 1, q2 =
J → J + 1, l → l, the rotational frequencies including
224.48 MHz, and a series of transitions is found in the
l-type doubling are
microwave region. In practice, it is found that q j varies
ν± = 2Bv (J + 1) − 4Dv (J + 1)[(J + 1)2 − l 2 ] slightly with J , and more elaborate expressions are re-
quired to fit accurately the l-type doublet spectra.
± 12 qi (v j + 1)(J + 1). (77) To define the rotation–vibration constants in terms of
more fundamental parameters or to understand the ori-
If l-type doubling represented by the last term is negligi-
gin of various nonrigidity effects in the rotational spec-
ble, the rotational spectrum is like that for a linear molecule
trum, the general rotation–vibration Hamiltonian must be
in a nondegenerate vibrational state except for the limita-
employed. This Hamiltonian contains pure rotation and
tions on the values of J and small effects in the distortion
vibration terms as well as interaction terms between ro-
correction because of the presence of the vibrational an-
tation and vibration. Perturbation treatments to various
gular momentum quantum number l.
orders are required to characterize the different rotation–
To distinguish the vibrational states, the value of |l| is
vibration effects. Space does not permit further discus-
added as a superscript to the vibrational quantum number
sion of this; however, we mention that such a perturbation
associated with the degenerate vibration. For a triatomic
|l| treatment shows that the α constants depend on the cubic
linear molecule, the states are specified by (v1 , v2 , v3 ).
potential energy constants of the molecule.
Thus, the notation (1, 2 , 0) corresponds to a state with the
2
FIGURE 18 Moderate-resolution scan of the J = 2 → 3 transition of FC P observed with a 2800-V cm−1 Stark
modulation. Unassigned lines are labeled µ and the l -doublet Stark lobes by an asterisk. Excited-state lines from
the bending mode and the splitting of the l = ±1 lines are shown. For the first excited bending state v2 = 1, l -type
doubling gives two (0, 11 , 0) lines shifted to high frequency relative to the ground-state line (0, 0, 0). Similar, but much
weaker, lines are found for v2 = 3. For the second excited state v2 = 2, the lines (0, 20 , 0) and (0, 22 , 0) occur. [From
Krato, H. W., Nixon, J. F., and Simmons, N. P. C. (1980). J. Mol. Spectrosc. 82, 185.]
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
spectral lines into doublets for asymmetric rotors. With a the two groups. If the barrier is very low, V3 → 0, the form
very low barrier, such as found in CH3 NO2 , a quite com- of the above waveequation is that of a spatial rotor with
plex spectrum can be obtained. In cases where asymmetric a fixed axis of rotation, and solution gives for the energy
groups are connected by a single bond, internal rotation levels
can give rise to distinct rotational isomers. These rotational
isomers are often stable enough to give separate rotational E = Fm 2 , (81)
spectra even though they cannot be chemically separated. with m = 0, ±1, ±2, . . . . The internal motion is essen-
As illustrated in Fig. 10, the different rotational isomers tially free rotation about the C C bond, and the internal ro-
can be readily distinguished because of the very high sensi- tation states are specified by the quantum number m. This
tivity of the moments of inertia to the molecular geometry. is the case for molecules such as CH3 NO2 , where the bar-
rier height is 6.03 cal/mole and the rotation is effectively
A. Single Top with a Threefold Barrier free (note that at room temperature, RT ∼ = 600 cal/mole).
For CH3 CH3 or CH3 CHO, there are three equivalent Actually, for this and other molecules of similar symme-
configurations for a complete rotation of the methyl group try, there are six equivalent configurations for a complete
about the C C bond. The potential function possesses internal revolution. The internal potential thus has sixfold
three potential energy minima and maxima as illustrated symmetry, and the leading term in V (α) is a V6 term:
in Fig. 19. Since the methyl group is symmetric, the mo-
V (α) = (V6 /2)(1 − cos 6α), (82)
ments of inertia of the molecule do not depend on the
orientation of the methyl group. The effects of internal ro- where we expect V6 V3 . For the low barrier (V6 ), the
tation are transmitted to the rotational spectrum somewhat m = ±3 levels are particularly sensitive to the barrier.
indirectly. If the barrier is very high, V3 → ∞, the internal mo-
The periodic potential function that describes the inter- tion of the methyl group corresponds to simple harmonic
nal rotation is expressed by torsional oscillation in each well. The cosine function in
Eq. (79) may be expanded, giving V (α) = (9V3 /4)α 2 , and
V (α) = (V3 /2)(1 − cos 3α), (79)
the form of Eq. (80) is like that for a simple harmonic
where V3 is the threefold barrier hindering internal rota- oscillator. Solution gives for the energy
tion and α the angle of internal rotation. The eigenvalue
equation for this internal motion is given by E = 3(V3 F)1/2 v + 12 , (83)
the sublevel spacing increases as the torsional energy in- TABLE XVII Barriers for Various Isotopic Forms of
creases, and when the torsional energy is greater than the CH3 OHa
barrier height, the energy levels go over to those of a free Species V1 V2 V3
rotor.
For intermediate values of the barrier, the torsional CH3 OH 0 0 373.08
wavefunctions and energy levels are given by solution of CD3 OH 0 0 371.8
the general wave equation, Eq. (80). The effects of tun- CH2 DOH 8.80 2.42 373.22
neling are transmitted to the rotational spectrum by the CHD2 OH −9.21 −2.29 373.28
interaction between overall and internal rotation. Since CH3 OD 0 0 366.25
the coupling effect differs for the two torsional sublevels, CD3 OD 0 0 362.19
there is a set of rotational energy levels associated with CHD2 OD −8.73 −2.90 365.00
each sublevel, and the two sets of rotational energy levels a For details see Su, C. F., Liu, M., and Quade, C. R.
are characterized by the Hamiltonians v A and v E . For (1991). J. Mol. Spectrosc. 146, 264.
moderately high barriers, the forms of these Hamiltonians
are the same as for a rigid rotor, Eq. (34), and there are two isotopic forms require V1 , V2 , and V3 terms in the potential
sets of rotational constants Av A , Av E , and so on. The ef- function. As apparent from the table, the V3 term for OH
fective rotational constants contain contributions from the and OD species is independent of isotopic substitution of
effects of internal rotation, and Av A , Av E , and so on differ the methyl group, although V3 changes going from OH
from each other. Hence, there is a rigid-rotorlike spectrum to OD. The changes observed provide data to understand
associated with each sublevel v A and v E. Each rotational more clearly electrostatic and other interactions character-
transition in a torsional state v appears as a doublet rather izing hinder rotation and data to test quantum models and
than a single line. This doublet separation is a sensitive ab initio calculations.
function of the barrier height, and analysis of these split- In the application of the various theoretical approaches,
tings yields the barrier that is hindering internal rotation. the emphasis in many cases is on the derivation of an ac-
If the splitting cannot be observed in the ground torsional curate potential function with as small a number of terms
state, it can often be observed in the excited torsional as possible in the Hamiltonian. On the other hand, if the
states, where the splitting increases. Numerous molecules desire is to provide a spectral analysis to microwave ac-
with threefold barriers have been studied by microwave curacy, a rather extensive number of fitting parameters is
spectroscopy; examples are collected in Table XVI. In required in the Hamiltonian. For the classic internal ro-
addition, molecules with two equivalent methyl groups tation molecule CH3 OH, such an analysis provides the
such as CH3 OCH3 and molecules with two nonequivalent capabilities to calculate accurate spectral predictions over
tops such as CH3 OSiH3 have been analyzed by microwave a significant range of quantum states. Such capabilities are
spectroscopy. A number of molecules with twofold barri- particularly important for applications in radio astronomy.
ers have also been studied. Nitrobenzene is an example of For CH3 OH, direct diagonalization of an internal rotation
this latter case. Hamiltonian with 60 fitting parameters, 37 of these repre-
Internal rotation analysis can be quite complicated for senting interaction terms describing the coupling between
molecules with asymmetric rotors and asymmetric frames torsion and end-over-end angular momentum, gave a fit of
such as CH2 DOH. Various theoretical approaches have 470 rotational transitions with a root-mean-square devia-
been developed to handle most situations. Table XVII tion of only 0.06 MHz.
gives a summary of barriers for various isotopic forms
of CH3 OH. Depending on the molecular symmetry, some
B. Rotational Isomerism
When the groups that rotate relative to each other are asym-
TABLE XVI Some Potential Barriers Measured with Micro- metric, various rotational isomers are possible. Each iso-
wave Spectroscopy
mer corresponds to a potential energy minimum. Since
Molecule V (cal/mole) Molecule V (cal/mole) the moments of inertia depend on the internal rotation an-
gle, distinct rotational spectra are observed for different
CH3 CH2 F 3306 (CH3 )2 CO 778
isomers. For each isomer, various kinds of information
CH3 CHO 1168 (CH3 )2 C CH2 2210
including detailed molecular structures, dipole moments,
CH3 COF 1041 (CH3 )3 CH 3900
and so forth may be evaluated. Furthermore, by compar-
CH3 COOH 483 CH3 NO2 6.03
ing the relative intensities of the rotational lines associated
CH3 CH2 COOH 2360 C6 H5 CH3 13.94
with different isomers, one can obtain information on the
CH3 CF CH2 2440 FC6 H5 CH3 13.82
energy difference between the isomers and determine the
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
molecules. Since the vibrational contributions are slightly TABLE XVIII Equilibrium Structure and Various
different for different isotopic species, the structural pa- Ground-State Structures of SOa2
rameters obtained from different combinations of isotopic Structure r (Å) θ
species are slightly different. The spread in these effective
parameters is much larger than expected from the experi- Equilibrium (re ) 1.4308 119◦ 19
mental uncertainty in the moments of inertia. On the other Average (r ) 1.4349 119◦ 21
hand, equilibrium parameters obtained from different sets Effective (r0 ) 1.4336 119◦ 25
of moments of inertia are consistent with the experimen- Substitution (rs ) 1.4312 119◦ 30
tal uncertainties in the moments of inertia. Because of this Mass dependence (rm ) 1.4307 119◦ 20
general problem, different procedures have been devel- Scaled (rρ ) 1.4308 119◦ 19
oped to minimize these troublesome vibrational effects.
This has led to different definitions of structural parame-
ters depending on their method of calculation. An under- librium bond distance for a diatomic molecule X—Y is
standing of these different structural measures is important evaluated from
if comparisons are to be made of structural parameters. " #
m X + m Y e 1/2
The different structural parameters are defined as follows: re = Ib , (87)
mXmY
1. Equilibrium structure re : the bond length or angle where
for the vibrationless state, evaluated by correction for the h/8π 2
effects of vibration Ibe = (88)
Be
2. Average structure r or rz : the bond length or angle
association with the average configuration of the atoms in and the conversion factor is given by
the ground vibrational state, evaluated by partial correc- h/8π 2 = 505,376 amu Å2 MHz. (89)
tion for the effects of vibration
3. Effective structure r0 : the bond length or angle ob- With more complicated molecules, additional isotopic
tained from effective ground-state moments of inertia data are needed. For a linear molecule such as XYZ, the
4. Substitution structure rs : the bond length or angle moments of inertia for two molecular species and the ex-
derived from isotopic data using differences in ground- pression from Table II give two equations to be solved
state moments of inertia for the two bond lengths. Alternately, the coordinates of,
5. Mass-dependence structure rm : the bond length or for example, the X-atom z X , that is, the distance from the
angle derived from a large number of isotopic species by center of mass, can be evaluated from Kraitchman’s equa-
a first-order treatment of isotopic effects tion (see Section VIII.D) using isotopic data from X∗ YZ.
6. Scaled structure rρ : The bond length or angle de- Subsequently, this coordinate can be used in the moment-
rived from scaled effective moments of inertia employing of-inertia and first-moment equations for the XYZ species,
a scaling factor obtained from a set of isotopic species.
IX = m i z i2 , m i z i = 0, (90)
The r0 and rs structures are defined operationally and
do not provide well-defined physical parameters. At this
time the rρ structures provide the best near-equilibrium TABLE XIX Comparison of Various Bond
Lengths (Å) for Some Diatomic Molecules
structures. A comparison of these different structures for
SO2 is given in Table XVIII. The evaluation of these dif- Molecule re r0 r
ferent structures is reviewed briefly using diatomic and
HF 0.9170 0.9257 0.9326
linear molecules as the principal examples. The values of
DF 0.9171 0.9234 0.9284
re , r , and r0 are compared for some diatomic molecules
TF 0.9177 0.9230 0.9272
in Table XIX.
H35 Cl 1.2745 1.2837 1.2904
Some of the unique structural information which can
H37 Cl 1.2746 1.2837 1.2904
be obtained from microwave spectroscopy has been dis-
D35 Cl 1.2744 1.2813 1.2858
cussed in Section IV.E.
D37 Cl 1.2744 1.2813 1.2858
T35 Cl 1.2746 1.2800 1.2853
A. Equilibrium Structure T37 Cl 1.2746 1.2800 1.2853
127 I35 Cl 2.3209 2.3236 2.3246
By correcting B0 for the effects of vibration, as discussed 127 I37 Cl 2.3209 2.3235 2.3245
previously, one can obtain Be . From Table II the equi-
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
to give two equations in the two unknowns z Y , z Z . From TABLE XX Effective and Substitution Struc-
the coordinates the bond lengths are then given by tures of OCS
Effective structures
di j = |z i − z j |. (91)
Bond length (Å)
With equilibrium moments of inertia, the important point
is that the particular method of calculation is not important. Isotopic species used C O C S
TABLE XXI Inertial Defects for Some Planar Moleculesa where νt is the low-frequency out-of-plane vibration and
Molecule (amu ∆ Å2 ) Molecule (amu ∆ Å2 )
Ic is the out-of-plane principal moment of inertia. The first
term follows directly from theory neglecting other molec-
H2 O 0.0486 H2 S 0.0660 ular vibrations, but it overestimates the magnitude of ,
H2 Se 0.0595 SO2 0.1348 indicating the positive contributions from other vibrations
H2 CO 0.0574 ClNO2 0.2079 are not negligible. Analysis of a number of molecules
Furan 0.046 Fluorobenzene 0.033 with low-frequency out-of-plane vibrations yields the
Pyrrole 0.076 Benzonitrile 0.084 above empirical formula with α = 0.0186 and 0.00803
a for aliphate and aromatic molecules, respectively. This
= Ic − Ia − Ib .
expression closely approximates observed inertial defects
for molecules with a low-frequency out-of-plane vibration
When effective moments are used, the residual defined and is hence useful in judging the planarity of molecules
above is small but does not vanish. This arises because the with small, negative inertial defects. It also follows from
vibrational effects associated with the different principal the above relation that νt can be estimated from an ob-
moments are slightly different. Nonetheless, a small served inertial object.
inertial defect is usually good evidence of the planarity
of a particular molecule. This is illustrated in Table XXI.
As observed, a small, positive inertial defect provides D. Substitution Structure
evidence for molecular planarity. On the other hand, the
presence of a very low out-of-plane vibration can result Substitution structures involve the use of Kraitchman’s
in a negative inertial defects for planar molecules as il- equations, which provide the position of an atom in a
lustrated in Table XXII. A simple relation has been de- molecule utilizing the changes in moments of inertia from
veloped to explain negative inertial defects observed for isotopic substitution. One isotopic form is selected as the
planar molecules in the ground vibrational state, parent molecule, and Kraitchman’s equations give coor-
√ dinates of the isotopically substituted atom in the center-
= −(33.715/νt ) (amu Å2 cm−1 ) + α Ic , of-mass principal axis system of the parent molecule. For
diatomic or linear molecules, Kraitchman’s equation has
TABLE XXII Inertial Defects for Some Pla- the form
nar Molecules with Low-Frequency Out-of-Plane
" #1/2
Vibrations (νt )a M + m
|z| = Ix , (96)
Molecule ∆ (amu Å2 ) ν t cm−1 Mm
CHO CHO −0.1286 108 where Ix = Ix − Ix is the difference in the moment of
CHO O CHO −0.1909 85 inertia of the isotopically substituted molecule of mass
S −0.164 90 M + m and the parent molecule of mass M. The m is,
NO2 hence, the mass change due to isotopic substitution. This
relation can also be used to find the position of an atom lo-
−0.128 111 cated on the symmetry axis of a symmetric-top molecule.
CHO
Only absolute values of the coordinates are obtained from
−0.775 30 relations such as Eq. (96). The sign of the coordinate must
F CHCH2 be assigned from other considerations such as the arrange-
ment of the atoms and an approximate location of the
CH2 CH NO2 −0.0665 100 center of mass. In general, to evaluate the bond distance
CHO CHS −0.070 104b between two atoms, the effective moments of inertia must
−0.146 100 be obtained for the parent molecule and two singly substi-
NO
tuted species. Likewise, from the moment-of-inertia data
−0.127 57 for a parent and three singly substituted species, an inter-
NCO atomic angle can be evaluated. For a diatomic molecule,
the parent X—Y and two isotopic forms X∗ —Y and X—
−0.131
N CHO
124
Y∗ are required to give the internuclear distance
dX−Y = |z X − z Y |. (97)
a=
Ic − Ia − Ib . For an extended compilation see
Oka, T. (1995). J. Mol. Struct. 352, 225. Since differences in the effective moments of inertia
b Estimated. are employed in calculating substitution structures, the
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
zero-point vibrational effects tend to cancel, and more this method is primarily based on the observation that
consistent structural parameters can be obtained from ρ is constant for different choices of parent species to
different combinations of isotopic data. This is illustrated ca. 1/104 ([I s ]g /[I 0 ]g ∼
= [I s ]1 /[I 0 ]1 ), and the Iρ com-
in Table XX by the different rs structures for OCS. puted from the above prescription give values very close
Note that the spread in the rs structures is 10 times to I e .
smaller than the spread in the r0 structures. One of the Once the set of L scaled moments of inertia Iρ have
best approximations to the equilibrium structure is the been evaluated, the molecular structural parameters are
substitution structure, and numerous such structures have derived by means of a standard least-squares fitting of the
been determined, with typical structural uncertainties of Iρ ’s. This is found to provide the best averaging of small
±0.002 to ±0.005 Å and ±0.2◦ to ±0.5◦ . residual vibrational effects. For a linear triatomic molecule
XYZ, the four moments of inertia would be analyzed for
E. Mass-Dependence Structure the parameters dYX and dYZ (see Table II). Importantly,
the method employs a minimal set of isotopic substitution
To evaluate the mass-dependence structure, various iso- data compared to the mass-dependence method. It is, how-
topic species are employed to calculate the substitution ever, necessary to select the parent such that all isotopic
coordinates of each nonequivalent atom for a given parent. substitutions satisfy either m i > 0 or m i < 0 for all
These substitution coordinates are then used to evaluate atoms i. This minimizes residual vibrational effects. For
the moments of inertia, Eq. (5), which are called substitu- the general case, there are moments Iα0 and Iαs (α = a, b, c)
tion moments of inertia Iαs . The mass-dependence moment associated with each axis, and these are used to calculate
of inertia Iαm is calculated from the relation ρ ρ ρ
the corresponding ρa , ρb , ρc and the Ia , Ib , Ic . The mo-
ρ
Iαm = 2Iαs − Iα0 (α = a, b, c). (98) ments of inertia Iα are then analyzed by least squares for
the structural parameters. Table XXIII compares several
To first order, the Iαm moments are equal to the equilibrium
structures for OCS. Results for SO2 are summarized in
moments of inertia Iαe . The above procedure is repeated
Table XVIII. It is apparent that the rρ structures compare
for another parent species. Once a sufficient number of
most favorably with the re structures. Similar results are
I m for different parent isotopic species have been deter-
found for other molecules.
mined, the moment-of-inertia equations may be solved to
This method based on the use of a set of scaled mometns
give the rm structure. The rm structure for SO2 is given in
of inertia provides a molecular structure which is a bet-
Table XVIII. This measure of the molecular structure has
ter approximation to the re structure than the conventional
limited applicability because of the large amount of pre-
rs structure, particularly for heavy-atom molecules. For
cise isotopic moment-of-inertia data required and because
molecules which contain hydrogen atoms, additional con-
the first-order approximation I m ∼= I e is not sufficient es-
siderations apply because of larger vibrational effects.
pecially for light atoms. Thus, hydrogen bond lengths can-
The quantity ρ is no longer virtually constant, but varies
not be determined by this method.
significantly when hydrogen atom substitution species
(H ↔ D, T ) are considered, and the Iρ values do not
F. Scaled Structure give reliable structures. Corrections now have to be ap-
A method has been proposed which employs a set of mo- plied to obtain near-equilibrium structures, and an empir-
ments [I ρ ]g defined for L isotopic species and computed ical method has been developed. In particular, the mo-
from ments of inertia for the deuterated species are corrected
via
[I ρ ]g = (2ρ − 1)[I 0 ]g , g = 1, 2, . . . , L , (99)
with the scaling factor obtained from (Iρ )D ρ D
corr = (I ) + . (101)
ρ = [I s ]1 [I 0 ]1 . (100)
TABLE XXIII Structural Calculations for Car-
Here [I 0 ]1 (g = 1) is the ground-state moment of the par- bonyl Sulfide (OCS)a
ent isotopic species and [I s ]1 is the substitution moment rs r0 rm rρ re
of inertia calculated from the set of substitution coordi-
nates of the parent species. The dataset, the L isotopic CO 1.1605 1.1568 1.1587 1.1551 1.1543
species, is that needed for evaluation of a complete sub- CS 1.5596 1.5645 1.5593 1.5621 1.5628
stitution structure. For a linear XYZ molecule, four iso- a All distances in angstroms. [From Harmony, M. D.,
topic species (including the parent) are needed for the and Taylor, W. H. (1988). J. Mol. Spectrosc. 118, 163;
substitution structure. From Eqs. (99) and (98), it follows see also Berry, R. J., and Harmony, M. D. (1988). Struct.
for the parent, g = 1, that [Iρ ]1 ≡ I m . The rationale for Chem. 1, 49.]
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
The correction factor utilizes a bond elongation param- evaluation of the dipole components along the principal
eter δrD (∼
=0.003 Å). Simple correction procedures have axes.
been given for various molecule types. The (Iρ )D corr and In the general case, the molecular dipole moment fixed
the other scaled moments of the complete set are then in the molecule can have three nonvanishing components
used in a least-squares determination of the structural pa- µx , µ y , µz along the molecule-fixed principal axis system.
rameters. The results found are quite good. The scaled The Stark effect Hamiltonian is then expressed by
method represents a significant step in the improvement
of spectroscopic structure determinations of polyatomic = − µg Zg , g = x, y, z, (104)
molecules.
where Zg are the direction cosines of the x, y, z axes rel-
ative to the space-fixed Z axis, the direction along which
IX. STARK EFFECT AND ZEEMAN EFFECT the electric field is applied. For linear and symmetric-
top molecules only a dipole component along the z axis
Applied electric or magnetic fields modify the rotational is present: µz = µ and µx = µ y = 0. Usually, the above
spectrum. The most important of these is the effect of elec- interaction can be adequately treated by first- and second-
tric fields, commonly called the Stark effect. In the absence order perturbation theory. When levels that interact via
of an electric field, the rotational states are (2J + 1)-fold are degenerate or near degenerate, the frequency dis-
degenerate since the energies are independent of M, placement ν of the Stark components is linear in the
electric field and depends on M, that is, ν = B J τ M.
M = 0, ±1, ±2, . . . , ±J, (102)
Here B J τ are functions of J and other possible quantum
which specifies the possible orientations of the angular numbers. In this case, a rotational line splits into 2J + 1
momentum vector relative to a fixed direction in space. components. On the other hand, when levels interact via
When an electric field is present, the field interacts with which are widely separated, the frequency displace-
the molecular dipole moment and the rotational energy ment varies as ν = (A J τ + B J τ M 2 )2 , and only J + 1
levels split into a number of sublevels. The degeneracy in components are obtained for J = ±1 and J components
the space orientation quantum number M is thus partially for J = 0. Here J refers to the smaller J value involved
or completely lifted by the interaction. A schematic illus- in the transition.
tration is provided in Fig. 3. A rotational line, therefore, The relative intensities of the Stark components depend
splits into a number of components. The general selection on the transition type
rules for rotational transitions are
I M = P[(J + 1)2 − M 2 ], J → J + 1
M → M, M → M ± 1, (103) IM = Q M 2, J→J M → M.
I M = R[J 2 − M 2 ], J → J − 1
in addition to those already given for the different types
of rotors (Section IV). Usually, the electric field is ap- (105)
plied parallel to the electric vector of the radiation, and For asymmetric tops all three expressions apply, while
the selection rule M = 0 applies. for linear and symmetric-top molecules only the first in-
The Stark effect is usually used to modulate rotational tensity expression is needed. The P, Q, R coefficinets
lines to improve their detection. This is the basis of the are independent of M but depend on the intensity of the
Stark-modulation spectrometer (Fig. 2) discussed earlier. unsplit line. Note that the M = 0 component is forbid-
With this type of spectrometer both the Stark lines and den for a J = 0 transition, and for a second-order effect
the zero-field transitions are displayed. The Stark effect (ν ∼ M 2 ), a factor of 1/2 must be included in the inten-
pattern can be a valuable aid in the assignment of rota- sity expression for the M = 0 component since the +M
tional spectra, particularly for asymmetric tops. Specifi- and −M degeneracy is lost.
cally, by counting the number of Stark components, one The general features of the second-order effect for the
can obtain an indication of the smaller J value involved linear molecule OCS are illustrated in Fig. 3. Clearly, as
in the transition. Another particularly important applica- the electric field is increased, the field-dependent Stark
tion of this effect is in the evaluation of very accurate components move further away from the zero-field line.
electric dipole moments. These are determined by care- The three components |M | = 0, 1, 2 are as expected for a
ful measurement of the displacement of the Stark com- J = 2 → 3 transition. The component nearest the unsplit
ponents from the zero-field absorption line as a func- line requires a high field before it shows up clearly. By
tion of the applied field. The analysis of these Stark plotting ν against 2 , the effect is easily classified as a
splittings by means of the appropriate expression allows second-order effect. The appearance of a first-order Stark
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
(1, 1) (2, 2) (2, 1) (3, 3) (3, 2) (3, 1) (4, 4) (4, 3) (4, 2) (4, 1)
J =1 → 2 0.3333 — — — — — — — — —
J =2 → 3 0.0833 0.3333 0.1666 — — — — — — —
J =3 → 4 0.0333 0.1333 0.0666 0.3000 0.2000 0.1000 — — — —
J =4 → 5 0.0166 0.0666 0.0333 0.1500 0.1000 0.0500 0.2666 0.2000 0.1333 0.0666
a Multiply entry by (0.50344)µ to obtain the shift of the Stark component from the undisplaced line.
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
TABLE XXV Relative Second-Order Stark Shifts for Linear (if present). The contribution of each component has the
Moleculesa form
Value of ∆[J(J + 1) − 3M 2 ]/J(J + 1)(2J − 1)(2J + 3) (2)
for various M E g J τ M = µ2g 2 A J,τ + B J,τ M 2 , (111)
0 1 2 3 4 where A and B are called Stark coefficients and depend
on J , τ , and the inertial asymmetry κ. Though simple
J =0→1 0.5333 — — — — expressions cannot be given for these coefficients, they
J =1→2 −0.1524 0.1238 — — — may be calculated from second-order perturbation theory
J =2→3 −0.0254 −0.0071 0.0476 — — and knowledge of the direction cosine matrix elements in
J =3→4 −0.0092 −0.0056 0.0052 0.0288 — the asymmetric rotor basis. They have been tabulated for
J =4→5 −0.0044 −0.0033 −0.0001 0.0054 0.0165 low J . Once these quantities are specified, the second-
a Multiply entry by (0.50344)2 µ2 2 /2B to obtain the shift of the
order Stark energy may be calculated from Eq. (111).
Stark component from the undisplaced line.
With asymmetric rotors, the possibility of degeneracies
or near degeneracies exists, and in this case the above
second-order expression does not apply. Near degenera-
The perturbation connects adjacent J states. Evaluation cies often occur between asymmetry doublets. For J = 2
of the direction cosine matrix elements in the basis |J, M) and a near-prolate rotor, for instance, the pair of levels
and the energy level differences gives 21,2 , 21,1 or 22,1 , 22,0 interact via µa , and this interaction
(2) is very large if the levels are very close together. Ordinary
E J,M = (0.50334)2
second-order perturbation theory then fails. Thus a transi-
" # tion involving a level that can interact with a nearby level
µ2 2 J (J + 1) − 3M 2
× , does not exhibit a typical second-order effect. To a good
2B J (J + 1)(2J − 1)(2J + 3)
approximation, these levels may be separated from the
J = 0 (109) other levels, and the problem can be treated by standard
methods of quantum mechanics as a two-level system.
In addition, for J = 0 and M = 0, we have
Consider two levels ψ10 , ψ20 that are eigenfunctions of the
(2) unperturbed Hamiltonian 0 with eigenvalues E 10 and E 20 .
E J =0 = −(0.50344)2 µ2 2 /6B. (110)
The complete Hamiltonian is = 0 + , with the
perturbation. In the basis of 0 , there are no off-diagonal
The Stark splitting ν (2) depends on the difference of
elements from 0 and no diagonal elements for . The
the bracket term in Eq. (109) for the two levels involved
secular determinant thus has the form
in the transition. Table XXV gives the difference in the
0
bracket term for some J → J + 1, M → M transitions. E1 − E µ12
= 0, (112)
It follows from the table that Stark components can oc- µ12 E 20 − E
cur on both the high- and low-frequency sides of the un-
perturbed line. For the J = 2 → 3 transition, two low- where µ12 = (1| |2). The roots are
frequency Stark lobes and one high-frequency lobe are 2 1/2
predicted and observed for OCS in Fig. 3. Furthermore, E 1,2 = 12 E 10 + E 20 ± 12 E 10 + E 20 + 42 µ212 ,
the magnitude of the splitting decreases with increasing
(113)
J , and the largest high-frequency displacement occurs
for M = J . As a quantitative example, consider OCS, with E 10 > E 20 ; the plus sign is used for E 1 and the negative
where 2B = 12,162.97, µ = 0.715 D, and a large field sign for E 2 . When (E 10 − E 20 )2 > 42 µ212 , the above can be
= 2800 V/cm. For the J , M = (2, 2) → (3, 2) transition, expanded to give a second-order effect:
we find ν (2) = 4 MHz, which is easily observable but
quite small compared with a first-order effect. µ212 2
E 1 = E 10 + + ···, (114)
E 10 − E 20
B. Asymmetric-Top Molecules
µ212 2
For an asymmetric rotor, the analysis basically follows E 2 = E 20 − + ···. (115)
E 10 − E 20
the same procedure; however, the details require more
space than possible in the scope of this presentation. The Note, however, that the second-order effect could be rather
Stark effect is usually second order, and the Stark energy large if the energy denominator is small. If the perturbation
contains contributions from all three dipole components is large, (E 10 − E 20 )2 < 4µ212 2 , then
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
Molecule µa µb µc µ(D)
H2 CO — — — 2.339
HFCO 0.58 1.91 — 1.99
HCOOH 1.391 0.26 — 1.415
CH3 CFO 2.83 0.88 — 2.96 FIGURE 23 Principal inertial axes of HCOF. The direction of µb
trans-CH3 CH2 CH2 F 1.970 0.566 — 2.050 may be selected from chemical arguments, but not that of µa .
gauche-CH3 CH2 CH2 F 1.137 1.450 0.472 1.902 However, for µa to increase on deuteration, which results in a
CH CHF 1.280 0.629 — 1.427 rotation of the b axis toward the D atom, the direction must be as
shown. The arrowhead indicates the negative end of the dipole
(CH3 )2 PH — 1.15 0.44 1.23
moment components.
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
the Zeeman effect. Except for molecules with nonzero along the spin axis). This nonspherical nuclear charge can
electronic angular momentum and consequently a perma- interact with the surrounding unsymmetrical molecular
nent magnetic dipole moment, the effect is small, but ob- charge distribution arising from all the other charges in
servable nonetheless with kilogauss magnetic fields. The the molecule. This latter distribution is measured by the
Zeeman effect Hamiltonian for the interaction of a mag- electric field gradient ∂ 2 V /∂ 2 z 2 , with V the electrostatic
netic dipole with an applied field H is given by potential. In such cases the nuclear spin I is coupled to
the overall rotation J, and a nuclear quadrupole hyperfine
H = −µ · H. (118)
structure results.√The total angular moment F = J + I has
For molecules in singlet ground electronic states, the the magnitude h F(F + 1), and a new quantum number
magnetic moment µ is generated by the molecular ro- F is now required to characterize the energy levels, or
tation (rotation of charges). With the assumption that the hyperfine states,
molecular magnetic moment is proportional to the angular
momentum, it can be shown that the Zeeman splitting of F = J + I, J + I − 1, J + I − 2, . . . , |J − I |. (120)
the rotational levels is given by
Each rotational level is hence split into a number of differ-
E H = −g J,τ β I HM, (119) ent levels labeled by the values of F. When J > I , there
are in general (2I + 1) values of F. The projection of F
where g J,τ is the rotational g-factor and depends on the
along an axis fixed in space, h M F , is specified by the
inner quantum numbers K and τ for symmetric and asym-
quantum number
metric tops, respectively. Here β I is the nuclear magneton,
H is the applied magnetic field, and M is the orienta- M F = F, F − 1, F − 2, . . . , −F. (121)
tion quantum number. The selection rules are M = 0 or
±1 depending on whether the magnetic field is applied This quantum number becomes important when an ex-
parallel or perpendicular to the electric radiation vector. ternal field is applied to a molecule with a quadrupolar
Analysis of the Zeeman splittings provides values of the nucleus. Since the rotational states are split by the nuclear
molecular rotational g-factors. These g-factors, for exam- interaction, a given rotational transition splits into a num-
ple, when measured for two isotopic species, can be used ber of components and a group of closely spaced lines
to determine the vector direction of the electric dipole is observed. The selection rules are as before, with the
moment. additional rules
Besides the molecular magnetic moments generated by
rotation, smaller magnetic moments are induced by the F → F, F → F ± 1, I → I. (122)
external field. These additional effects are expressed in
This splitting, in general, increases for some of the more
terms of a magnetic susceptibility tensor χ, and analysis
common nuclei in the order 14 N(I = 1) < 37 Cl(I = 3/2) <
of these effects yields the elements of the magnetic sus- 35
Cl(I = 3/2) < 81 Br(I = 3/2) < 79 Br(I = 3/2) < 127 I(I =
ceptibility tensor. Considerable progress has been made in
5/2). Many common nuclei have I = 0 (e.g., 12 C, 16 O,
the study of magnetic properties of molecules and in de- 32
S) or I = 1/2 (e.g., 1 H, 13 C, 15 N, 19 F) and hence do
riving information on the electronic structure of molecules
not give rise to a quadrupole coupling interaction. The
from these studies. However, their description is outside
quadrupole splittings decrease with increasing J , often
the scope of this presentation.
becoming unresolvable at sufficiently high J . Nuclear
quadrupole hyperfine structure in the rotational spectrum
indicates the presence of a quadrupolar nucleus such as
X. NUCLEAR QUADRUPOLE
Cl; it can be useful in the assignment of a spectrum and it
HYPERFINE STRUCTURE
provides information on chemical bonding.
The effects of nuclear coupling can give rise to hyperfine
structure in the rotational spectrum. The most important A. Linear and Symmetric-Top Molecules
type of nuclear interaction occurs when a nucleus with a
The nuclear quadrupole energy for a linear molecule with
nonzero nuclear quadrupole moment (I > 1/2) is present
a single coupling nucleus in the absence of external fields
in the molecule. The nuclear quadrupole moment Q mea-
is given by
sures the deviation of the nuclear charge distribution from
a spherical distribution. Generally, Q becomes larger as E Q = −χ Y (J, I, F), (123)
the nucleus becomes heavier. A positive Q indicates a
prolatelike distribution (elongated along the spin axis) and where χ = eQq is the nuclear quadrupole coupling con-
a negative Q indicates an oblatelike distribution (flattened stant in frequency units with q = ∂ 2 V /∂z 2 , the electric
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
field gradient relative to the principal axis system, and weaker F → F components (3/2 → 3/2, 5/2 → 5/2, and
where z is along the molecular axis and Q is consid- 7/2 → 7/2), and two even weaker F → F − 1 compo-
ered known and is characteristic of the coupling nucleus. nents (5/2 → 3/2 and 7/2 → 5/2). These considerations
Also, enable the identification of the hyperfine components
(3/4)C(C + 1) − I (I + 1)J (J + 1) of a given rotational transition. Early applications of
Y (J, I, F) = microwave spectroscopy to the study of hyperfine
2(2J − 1)(2J + 3)I (2I − 1)
structure used the appearance of the fine structure, that is,
(124) the relative spacings and intensities of the components, to
C = F(F + 1) − J (J + 1) − I (I + 1) (125) determine unknown nuclear spins. For example, the spin
of 33 S was found to be 3/2 by this method.
is a function of the various quantum numbers and the spin For a symmetric top, with the coupling atom on the
I . It has been tabulated for various I and J . It may be symmetry axis,
observed that Y (J, I, F) is undefined for I = 0 and 1/2, " #
which is consistent with the requirement I t for a nu- 3K 2
EQ = χ − 1 Y (J, I, F), (127)
clear quadrupole interaction. The E Q must be added to J (J + 1)
the rigid-rotor energy E r to give the total energy. Apply- where χ is the coupling constant with reference to the
ing the selection rules J = 0, F = 0, ±1, one obtains molecular axis of symmetry. From the above expression
the rotational frequencies including effects of quadrupole it follows that each J , K level splits into a number of sub-
coupling: levels of different F. The selection rules are J = ± 1,
ν = νr − χ [Y (J + 1, I, F ) − Y (J, I, F)], (126) K = 0, I = 0, F = 0, ±1. When K = 0, the hyper-
fine pattern is like that for a linear molecule. For other K
where νr is the unperturbed rotational frequency and values, a similar pattern is obtained. However, when dif-
F = F, F ± 1. The coupling constant can be evaluated ferent K lines are separated by less than the quadrupole
from the splitting between any two hyperfine components. splitting, the individual patterns for each K overlap, and
To evaluate the rotational constant, the rigid-rotor fre- a quite complex overall structure can be obtained.
quency νr is required, and this may be evaluated by cor- For coupling atoms off the symmetry axis we have a
recting the hyperfine components ν with the known χ via more complicated problem. An example would be HCCl3 .
the above frequency expression. The hyperfine structure for molecules with two or more
To understand the appearance of the hyperfine pattern, a coupling nuclei is more complex but has been treated the-
knowledge of the relative intensities of the components is oretically and observed experimentally.
required. The explicit expressions require too much space
to give here. However, we may point out that for any class B. Asymmetric-Top Molecules
of rotor when J > I , there are 2I + 1 components for
F → F + 1, 2I components for F → F, and 2I − 1 com- For an asymmetric prolate rotor with a single coupling
ponents for F → F − 1. Furthermore, the most intense nucleus we have
components are those where F = J . An approximate ( ! ! )
E Q = 3 Pa2 − J (J + 1) χaa − σ Pa2 − W (b p ) ηχaa
intensity rule is that for J → J + 1 transitions, the inten-
sities of the F → F + 1 components are proportional to Y (J, I, F)
× , (128)
F, while the F → F components are considerably weaker J (J + 1)
and the F → F − 1 components even weaker. For the
where σ = −1/b p , and W (b p ) is Wang’s reduced energy
J → J transition, the intensity of the F → F component
and Pa2 the average PZ2 in the asymmetric rotor basis.
is proportional to F, while the F → F ± 1 components
The asymmetry parameter
are considerably weaker. In fact, the intensities of the
components for F = J decrease rapidly with increasing χbb − χcc
η= (129)
J . Also, the function Y (J, I, F) may be positive or nega- χaa
tive, but for the maximum and minimum values of F, the measures the departure of the field gradient from cylin-
function is positive. Moreover, the strongest component drical symmetry about a. The coupling constants χaa =
is usually not significantly displaced from the rigid-rotor eQqaa , χbb = eQqbb , and χcc = eQqcc are relative to the
position. Consider, for example, the 2 → 3 transition principal axis system; qaa = ∂ 2 V /∂a 2 and so on are the
with I = 3/2. For J = 2, F = 1/2, 3/2, 5/2, 7/2 and corresponding field gradients. Since Laplace’s relation
for J = 3, F = 3/2, 5/2, 7/2, 9/2. We expect four F → holds,
F + 1 components (1/2 → 3/2, 3/2 → 5/2, 5/2 → 7/2,
and 7/2 → 9/2 in order of increasing intensity), three χaa + χbb + χcc = 0, (130)
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
hand, for linear and symmetric rotors, the field gradient is splittings are possible in singlet electronic ground states
symmetric (χx x = χ yy ) about the bond axis z, which cor- and have been observed. This arises from the interaction
responds to a principal axis, and from Eq. (130) it follows ( = −µ I · H) of the nuclear spin magnetic moment µ I
that χx = χ y = − 12 x z = − 12 χ . and a weak magnetic field H generated by rotation of the
molecule. Hence, H is proportional to the angular momen-
C. Quadrupole Coupling Constants tum. High-resolution techniques are required to observe
and the Chemical Bond the small magnetic splittings required for evaluation of the
Considerable information on chemical bonding has been nuclear magnetic coupling constants. The theory for com-
obtained from the quadrupole coupling constants. The ba- bined nuclear quadrupole and magnetic hyperfine interac-
sic approach is to relate the nuclear coupling in molecules tions has been developed. For CH2 O, the observed hyper-
to the known coupling in free atoms. The field gradient fine structure arises solely from the magnetic moments of
in atoms is primarily determined by the p electrons, and the two protons, since neither hydrogen nor oxygen has a
in a molecule any factor that affects the p electron dis- quadrupole moment. On the other hand, for CHDO there
tribution gives rise to a change in the atomic coupling is a quadrupole interaction from D which has a quadrupole
constant. Relations between the charge distribution near moment as well as a magnetic interaction from the nuclear
the coupling nucleus and the coupling constants are avail- spins of H and D.
able. These analyses have provided additional insights into
the concepts of ionic character, hybridization, π -bonding, XI. ADVANCED EXPERIMENTAL METHODS
and electronegativities. Table XXVII illustrates the depen-
dence of the field gradient q on the electronic environment, Continued developments in experimental and theoretical
showing eQq for the Cl nucleus for a series of molecules. techniques have ensured that microwave spectroscopy re-
The value for FCl is larger than for atomic Cl, which indi- mains a very active field. Since its inception shortly after
cates a positive charge on the Cl atom, F− Cl+ . At the other World War II, numerous experimental improvements and
extreme, for the highly ionic bond in K+ Cl− , where the Cl technological advances have been made. These and vari-
nucleus is surrounded by a symmetric charge distribution, ous theoretical improvements are discussed and well doc-
we find q ≡ 0. On the other hand, the coupling constant umented in other articles and books (see Bibliography).
for BrCl is near that for atomic chlorine, which indicates Here we highlight a few of the more recent developments.
essentially a covalent bond with little ionic character. As
one proceeds from ICl to AlCl, the bond becomes less A. Spectroscopy of Isotope-Substituted
covalent and more ionic in character. Nonpolar Molecules
It was often observed that microwave spectroscopy could
D. Magnetic Interactions not be applied to nonpolar molecules. This observation
must be somewhat clarified. Improvements in detection
Although we shall not treat magnetic interactions, the
sensitivity have opened the way to measurements on
reader should be aware that small magnetic hyperfine
very slightly polar molecules with electric dipole mo-
ments approaching 10−5 D. Moreover, mechanisms for
TABLE XXVII Comparison of eQq (in MHz) for a the generation of such small moments have now been rec-
Chlorine (35 Cl) Nucleus in Various Molecules
ognized and exploited.
Molecule eQq Small dipole moments can be induced in nonpolar
molecules by three mechanisms. First, small dipole mo-
ClF −145.9
ments can be vibrationally induced when, for example,
Cl (atomic) −109.74
symmetric or spherical tops are in excited, degenerate vi-
BrCl −102.4
brational states. Second, for nonlinear molecules without
ICl −85.8
a center of symmetry, centrifugal distortion can distort the
ClCN −83.4
molecule, hence generating a small dipole moment. This
HC CCl −79.7
mechanism, as discussed in Section IV. B, has now allowed
CH3 Cl −74.8
the observation of rotational transitions in the ground vi-
SiH3 Cl −40.0
brational state of spherical tops. Third, nonsymmetric iso-
AgCl −37.3
topic substitution of a nonpolar molecule can give rise to a
TICl −15.8
small molecular dipole moment by vibrational averaging
AlCl −8.8
effects. This case is the most general and is applicable to
KCl 0
a large number of nonpolar molecules.
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
17 O C O,a Gripp, J., Mäder, H., Dreizler, H., and Teffo, J. L. (1995). J. Mol. Spectrosc. 172, 430
µ = 0.0006 D
18 O C Oa ,
µ = 0.0011 D
D Bottcher, O., Heineking, N., and Sutter, D. H. (1990). J. Mol. Spectrosc. 139, 236
, a µ 0.0115 D
D
Deuterium substitution often provides dipole moments for such cells minimizes radical decomposition. A typical
of sufficient magnitude for ready detection, typically cell is illustrated in Fig. 26. This type of cell, or a variant of
0.01 D. Such isotopic substitution allows the direct mea- this cell, can be used to study molecular ions, semistable
surement by microwave spectroscopy of many funda- molecules, and molecules at high temperature.
mental molecules, such as CH4 , CH3 CH3 , CH2 CH2 , Although molecular ions have been detected and as-
and CH CH. In addition to new structural information, signed (see Section IV.E.2), their number is still small,
other advantages are possible. Internal rotation informa- primarily because of the difficultly in producing signi-
tion is difficult to extract from symmetric-top spectra. By ficant concentrations. A new technique which increases
converting a symmetric top to an asymmetric top (say their density by about two orders of magnitude employs
(CH3 CH3 → CH3 CHD2 ), effects such as internal rotation a longitudinal magnetic field along the axis of a glow dis-
can be more readily studied. Heavy-atom substitution also charge tube. The addition of the magnetic field increases
produces a small dipole moment; however, the result of the length of the ion-rich negative glow and the concen-
this type of substitution will be much smaller than for tration of the molecular ions. The signal enhancement by
H-atom substitution. For example, in 16 O12 C17 O, the ob- application of a magnetic field for HCO+ is illustrated in
served dipole was estimated to be only 6 × 10−4 D. Other Fig. 27. The inset depicts the glass pipe absorption cell
examples of small induced dipole moments from isotopic (5 ft long, 1.5 in. inside diameter). The glass pipe is con-
substitution are given in Table XXVIII. nected to short transition sections that flare to 4 in. The
transition sections house cylindrical electrodes of the same
B. Unstable and Semistable Species inside diameter as the glass pipe to maximize microwave
transmission. The solenoid provides a field of up to 300 G.
Many molecular species studied by microwave spec-
By employment of high-temperature cells, numerous
troscopy are unstable to various degrees, and special
molecules have been studied that at room temperature
preparation techniques, absorption cells, and instrumenta-
would not have sufficient vapor pressure to give an ob-
tion methods have been developed for their investigation.
servable microwave spectrum. Various alkali halides, for
These techniques and methods have been applied to the
study of radicals, ions, and semistable molecules. The low
operating pressure of typical microwave studies helps in
minimizing decomposition from wall collisions and inter-
molecular collisions.
Radicals (see also Section IV.E.2) are very short-lived,
reactive species. They are often produced as products of
an RF electric discharge. Alternatively, the products of
an electric discharge are allowed to react with another
substance to produce the desired radical. In these produc-
FIGURE 26 Details of a typical free space cell for the study of
tion methods, a continuous flow of radicals is supplied to
radicals or other unstable species. The reactive species pass into
the microwave absorption cell. The use of glass absorp- the cell and are subjected to microwave radiation, and the resulting
tion cells with Teflon windows is particularly useful for absorption is detected. The molecular species are continuously
such studies. The large volume-to-surface ratio possible replaced by the pumping system.
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
FIGURE 27 Apparatus for enhanced production of positive molecular ions. A 13-mA discharge in a 1:1 mixture of
CO and H2 is used to produce HCO+ . Enhancement of the line at 267.5 GHz is apparent when the field is turned on.
[From De Lucia, F. C., Herbst, E., Plummer, G. M., and Blake, G. A. (1983). J. Chem. Phys. 78, 2312.]
example, have been studied this way. In situ production in or inside the absorption cell. As general examples, we
hot absorption cells has been useful for many cases where cite the production of thioketene, CH2 C S, by pyrol-
simple vaporization is not applicable because of dissoci- ysis of the trimer [(CH3 )2 CS]3 at 1000◦ C, preparation of
ation or decomposition. Here, reactions are often made bromoketene, BrHCCO, by pyrolysis of Br2 HCCOCl with
to occur within the hot cell by flowing appropriate con- zinc metal at 300◦ C, the synthesis of ClBSe by passing
stituents into the cell. Likewise, pyrolysis and thermolysis Cl2 Se2 over boron at 1100◦ C, and of XNCO (X = Br, I)
have been developed as useful techniques for producing by passing the halide vapor, X2 , over warm, dry AgNCO.
numerous semistable molecular species. Figure 28 shows an example of a high-temperature reac-
The lifetimes of semistable molecules are usually on tion flow system. The molecule or molecules to be reacted
the order of 1 sec; hence, the molecular species are not are passed through a small quartz tube heated to an appro-
isolable, but they live long enough to flow through con- priate temperature. The thermolysis products flow directly
ventional absorption cells at moderate flow rates. Such into and through the absorption cell of a microwave spec-
molecular species are not generally accessible by stan- trometer. The flow rate is adjusted to maintain a suitable
dard chemical methods and their properties and chemistry pressure (ca. 0.1–100 µm Hg). The low operating pres-
are often relatively unknown. sures characteristic of microwave spectroscopy help to
By coupling a reaction flow system with a microwave reduce unwanted polymerization or decomposition with
spectrometer, semistable molecules can be detected and the metal walls, etc., and the flow technique continually
readily identified. A simple production method would in- replaces the sample with fresh reaction products.
volve merely heating an appropriate precursor and passing
the resulting decomposition products into an absorption
C. Low-Temperature Spectroscopy
cell. Alternatively, a compound may be heated (or not)
and transported with (or without) an inert carrier gas into At very low temperatures, the number of thermally pop-
a reaction zone where reaction with another (heated) sub- ulated states is significantly reduced and complex spectra
stance takes place. The reaction zone may be just before can be greatly simplified. The simplification of complex
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
FIGURE 28 Illustration of a high-temperature reaction flow system coupled to a microwave spectrometer for producing
short-lived molecules. The gas is made to flow through the Stark cell by means of vacuum system and the flow rate is
adjusted by a needle valve. The 8-mm-inside-diameter quartz tube is heated in a 10-cm-long furnace. The distance
between the cell and furnace is also about 10 cm. [From Kroto, H. W. (1982)]. Chem. Soc. Rev. 11, 435.]
spectra contributes immensely to the task of assigning the cell, where it undergoes a series of collisions with the cold
spectra. Moreover, detection is improved because of an in- background gas as it randomly walks toward the cell wall
crease in the absorption coefficient with decreasing tem- and freezes out. The spectroscopic gas may be heated by
perature. Temperatures near 1 K have been obtained in means of the small oven around the injector tube to tem-
microwave studies. Very low temperature spectroscopy peratures up to ca. 1300 K. Since the spectroscopic gas
provides the opportunity to study weakly bound species, is at low concentration, it does not affect the temperature
discussed in Section IV.E.3, as well as other processes. of the inert buffer gas. The collisions cool the molecule’s
Some of the advantages and applications are summarized various degrees of freedom, and typically some 10,000
in Table XXIX.
Very low temperature conditions can be achieved by
use of free expansion jets. Numerous weakly bound com-
plexes have been produced by this method (see Sec-
tions IV.E.3 and XI.D). Another method is a collisional
cooling technique. This technique is particularly versa-
tile; the system is shown in Fig. 29. The system employs
a variable-temperature injector to produce an enhanced
population in excited vibrational states and liquid nitro-
gen temperature trapping. The cell is maintained at 77 K
and contains a buffer gas of helium or nitrogen, which has
a significant vapor pressure at 77 K. A small amount of
the gas to be studied is injected into the cooled absorption
Improved resolution
Increased sensitivity
Simplification of complex spectra
FIGURE 29 Schematic illustration of a collisionally cooled cell.
More direct correspondence between observable parameters and
The 4-in.-diameter cell is 3 ft long with the center 1-ft section made
theoretical parameters
of copper and the end sections of stainless steel. The dashed-
Enhance production and lifetime of transient species line region is enclosed in a 77 K medium. Windows for passing
Energy transfer microwave radiation are 12 -in. Teflon. The injector is a copper tube
Reaction dynamics at low temperature (0.04-in. inside diameter) surrounded by a small oven at the inlet
Pressure-broadening processes where hν > kT to the cell. The inset shows a random collision scenario. [After
Goyette, T. M., Ebenstein, W. L., and De Lucia, F. C. (1990). J. Mol.
Simulate low-temperature and low-pressure environments (outer space)
Spectrosc. 140, 311.]
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
collisions occur before the molecule collides with the cell surements are made at various pressures. The linewidth
wall. It is found, that in less than 100 collisions, the trans- ν is extracted from the spectral line shape. These data
lational and rotational temperature is cooled to that of the provide γ from Eq. (131) and n from the temperature de-
buffer gas. On the other hand, the relaxation of the vi- pendence of Eq. (132).
brational degrees of freedom is slower. The vibrational As an example, typical of pressure broadening stud-
temperature is found to depend on the ejection tempera- ies, we consider the 22,0 → 31,3 transition of water at
ture, the number of collisions with the background gas, 183 GHz with H2 as a collision partner in the tempera-
and the vibrational relaxation cross section. It is there- ture range 80–600 K and pressure range 0.05–1.0 Torr.
fore possible to attain separate translational/rotational Above 150 K, the temperature dependence is given by the
and vibrational temperatures by control of the injec- above relation with γH2 (300 K) = 3.20 ± 0.08 MHz/Torr
tion temperature and the pressure and temperature of the and n = 0.95 ±0.07. However, below 150 K, there is a sig-
buffer gas. nificant deviation from the power law. This result indicates
that H2 as a collision partner is considerably more com-
plex, and H2 does not act as a classical hard sphere. This
Collisional Broadening is also evident from the value of n since for a hard-sphere
To extract molecular concentration information and to model n = 0.50. On the other hand, for He as a collision
model the earth’s atmosphere accurate knowledge of pres- partner, n over the whole temperature range is found to be
sure and temperature effects on spectral line shapes is 0.49 ± 0.02, indicating that He behaves essentially as a
needed. Such pressure broadening studies are necessary classical hard sphere in collisions with H2 O.
to develop models of the chemistry and physics of the Similarly, the proper interpretation of radio astronom-
atmosphere and to gain insight into problems associated ical spectral lines from dense interstellar clouds requires
with air pollution, the greenhouse effect, ozone hole, etc. collisional information involving ions at low temperatures.
Spectroscopic remote sensing of planetary and interstellar By combining the considerations reflected in Figs. 27 and
atmospheres also requires such line broadening informa- 29, it is possible to obtain the pressure broadening of
tion. Line shape information provides direct information molecular ions at very low temperatures. This has been
on the environment of the molecule, viz., temperature, demonstrated for the ion HCO+ and the collision partner
pressure, collision partner, etc. Principal collision partners H2 . Likewise, extension of direct time-resolved measure-
for minor atmospheric species are N2 and O2 . Pressure ments, as discussed in Section IV.D, has been recently
broadening studies have been carried out at temperatures carried out incorporating the collisional cooling technique
characteristic of the earth’s atmosphere. However, for discussed here.
planetary species, the dominant collision partners are He
and H2 . Furthermore, the atmospheres are characterized D. Fourier-Transform Microwave Spectroscopy
by low pressures and temperatures, which are difficult to
simulate with conventional approaches. At the low tem- The technique of Fourier-transform microwave spec-
peratures, with conventional techniques, the vapor pres- troscopy (FTMS) has been applied to the study of a num-
sures would be vanishingly small due to condensation. ber of weakly bonded complexes, the observation of weak
On the other hand, the collisional cooling technique dis- isotopic species, and the resolution of hyperfine struc-
cussed here provides an ideal laboratory method to sim- ture. It is characterized by higher resolution and sensi-
ulate planetary conditions and similar low-temperature, tivity than conventional Stark-modulated spectros-
low-pressure conditions. copy. The superior resolution is demonstrated in Figs. 24
In linewidth studies, the width is measured versus pres- and 25.
sure and temperature. The linewidth varies with pressure In this method, a short, intense microwave pulse is ap-
at a given temperature as plied to the sample. This pulsed microwave excitation of
the sample produces a transient emission signal which is
ν = γP + ν0 , (131) detected by a transient signal averager. The time response
of the system is hence observed. Both waveguide-based
where P is the pressure and γ is the pressure broaden- sample cells and cavity-based cells have been employed.
ing coefficient. Here ν denotes the total linewidth. The To produce significant transient emission, a high power
temperature dependence of γ is taken as source (order of watts) is required for the waveguide-
γ (T ) = γ0 (T0 /T )n , (132) based system. Lower power sources are applicable to the
cavity systems because of the very narrow bandwidth of
where γ0 is the coefficient at the reference temperature T0 such systems. The transient emission is usually averaged
and n is a constant. At each temperature, linewidth mea- over many cycles to improve the signal-to-noise ratio. This
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
time-domain signal is converted to the frequency domain signal) computed from the time-domain signal to give
by calculating the Fourier transform. Earlier waveguide the spectrum. Because the cavity is tuned to ν and has
MWFT spectrometers developed by Drizler and cowork- a very narrow bandwidth (∼1 MHz), a high radiation den-
ers used square cross-section sample cells. Improved ver- sity is obtained. This produces appreciable polarization
sions have used circular waveguide sample cells and of the complex when rotational transitions lie inside the
bridge-type arrangements where one arm of the bridge frequency bandwidth. Switch 2 serves to protect the de-
is the sample cell and the other arm the reference cell. tector, and its opening is delayed to allow the microwave
In the study of complexes, a Fabry-Perot cavity is em- pulse to dissipate within the cavity before detection of the
ployed as a sample cell along with a pulsed supersonic weak, longer duration, coherent resonant rotational emis-
nozzle source. The weakly bonded complex is produced sion. Gas pulses are repeated by rapid evacuation of the
at very low temperatures (a few degrees Kelvin), by ex- cavity at a rate of ∼1 per second. A timing scheme is set
panding (adiabatically) as a short pulse a mixture of the up to properly sequence the various events and to allow
reactive species in an inert buffer gas. Weak complex for- alternate collection of the background which is subtracted
mation occurs undoubtedly via three-body collisions, for from the molecular emission. Various cycles are averaged
example, to produce the desired spectrum.
Though the resolution is exceptionally higher, because
Xe + Xe + HCl → Xe· · ·HCl + Xe, of the narrow bandwidth, only small elements of the spec-
and the low temperatures obtained help stabilize the tra can be covered at a time. Also, the low gas temperature
complex. allows primarily the ground vibrational state to be studied.
A simplified diagram is given in Fig. 30 of a pulsed As the technique has become more mature, various
nozzle FTMS which has been developed by Flygare and molecular species have been studied, and various tech-
coworkers. The cavity consists of two spherical concave niques have been introduced to increase the versatility of
metal mirrors. Each mirror has a centrally located small the FT spectrometers. For the cavity-pulsed FTMW spec-
hole (iris) to couple the radiation into and out of the cavity trometer, in addition to studies of weakly bound clusters,
with a waveguide. Before exiting, radiation reflects back various studies have been carried out on reactive short-
and forth between the mirrors, giving the cavity a long lived unstable molecules, as well as stable molecules with
effective path length and high sensitivity. The cavity res- low concentrations. In the case of large, stable molecules,
onant frequency, determined by the mirror dimensions, is for example, the spectral density can be very high be-
varied by very small movement of one of the mirrors. In cause of the small rotational constants and hence closely
operation, a valve is opened and a short pulse (a few mil- spaced energy levels. This, coupled with vibrational satel-
liseconds) of gas is introduced into the cavity and shortly lites from low-frequency vibrations and/or large nuclear
thereafter a pulse (5 µsec) of microwave radiation ν from hyperfine structure, makes the assignment of such spec-
the BWO source is produced by opening and closing pin tra difficult. The use of a supersonic expansion greatly
diode switch 1. Switch 2 is then opened and the resulting simplifies the spectrum by forcing the molecules into the
coherent emission is detected. The signal is processed and lowest rotational states (low J ) and the ground vibrational
digitized, and the Fourier transform (frequency-domain state. This, coupled with the high sensitivity of the tech-
nique, allows the assignment of large molecules or low-
concentration molecules.
The rapid expansion process provides for molecular
formation in a very short time (∼10 µsec) and leaves
molecules in an essentially frozen state where additional
chemical reactions via collisions cannot occur. This pro-
vides the ability to study unstable molecular clusters,
short-lived transient species, etc.
Different strategies can be used to produce the desired
molecular species to be studied. For example:
r A high-temperature (ca. 1000 K) capillary tube can be is in the microwave region and the pump transition is in
used to pyrolyze a molecule in a buffer gas followed by the millimeter-wave region. The pump radiation is intro-
gas expansion duced into the cavity by means of an oversized microwave
r A dc glow discharge of an appropriate gas mixture can horn mounted inside the vacuum chamber. The interaction
be used to produce the desired molecule. Alternatively, region for the millimeter waves with the microwave radi-
the electric discharge can be used to excite vibrational ation is a small cylindrical region above the horn.
states in one component of a gas mixture and these The inset in Fig. 31 depicts a typical DR experiment.
excited molecules form a vibrationally excited The double arrow line represent a microwave signal tran-
complex, for example, Ar· · ·HCl (10◦ 0 state). sition and the single arrow a millimeter-wave pump tran-
r Laser ablation of solids can be used to provide metal sition. This is only one of four possible double resonance
reactants. connections for this three-level system. In an actual ex-
periment, the 20,2 –30,3 microwave transition is monitored
In laser ablation, a metal is vaporized by a high-powered while the millimeter-wave pump transition is slowly swept
laser (Nd:YAG, 532 nm) and the metal vapor is allowed through a given spectral region. A change in the signal
to react with other species present in a buffer gas. The transition (DR effect) confirms the energy level structure
product is stabilized by expansion through a nozzle into and provides an accurate measure of the pump transition
the cavity. This method has been used to produce and study 20,2 –31,3 . This DR experiment illustrates how spectral as-
such species as InX (X = F, Cl, Br), AlCl, YCl, YBr, and signments may be confirmed. Also, note that the high-
others. frequency pump transition, which lies above the frequency
Besides the perpendicular configuration shown in region of the cavity MWFT spectrometer, can be indirectly
Fig. 31, it is possible to use a parallel configuration where measured with this DR experiment.
the nozzle is mounted in one of the mirrors and the molec-
ular beam from the expansion nozzle travels parallel to
the axis of the cavity. This configuration, in fact, provides E. Terahertz Region
improved resolution.
As apparent from Eq. (1), line intensity increases with
increased frequency and hence the terahertz region offers
1. Pulsed-Nozzle MWFT
significant advantages for the detection of weak signals.
Double-Resonance Spectroscopy
The major methods to obtain high frequencies are (i)
With the configuration of Fig. 31, it is possible to con- frequency multiplication of lower frequency sources and
veniently perform microwave–millimeter wave double- (ii) high-frequency tunable backward-wave oscillators
resonance (DR) experiments. Here, the signal transition (BWOs). BWOs have been developed that provide radi-
ation to 1.3 THz. These sources coupled with broadband
scanning techniques provide capabilities similar to the
lower microwave and millimeter-wave frequency regions.
Significant improvements in multiplier design employing
a planar Schottky diode also allow efficient conversion
of fundamental frequencies to over a terahertz. Here, a
high-power, lower frequency BWO operating in the range
300–600 GHz is used as the primary source. The region is
hence characterized by high sensitivity, high resolution,
high measurement accuracy, and broadband tuning
capabilities.
FIGURE 31 A schematic representation of a pulsed cavity FTMW SEE ALSO THE FOLLOWING ARTICLES
spectrometer with both molecular beam and cavity axis parallel.
For DR experiments, a microwave horn is used to inject millimeter- ANALYTICAL CHEMISTRY • COLLISION-INDUCED SPEC-
wave radiation into a small interaction zone of the molecular ex- TROSCOPY • HYDROGEN BOND • INFRARED SPEC-
pansion. The inset shows the energy level scheme for a DR ex-
TROSCOPY • INTERSTELLAR MATTER • ION KINETICS AND
periment. Here the signal frequency was fixed at the 20,2 → 30,3
transition and this transition was monitored as the pump radiation ENERGETICS • MICROWAVE COMMUNICATIONS • MIL-
was scanned. For details of the DR experiment see Jager, W., and LIMETER ASTRONOMY • QUANTUM CHEMISTRY • TIME
Gerry, M. C. L. (1995). J. Chem. Phys. 102, 3587. AND FREQUENCY
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN009N-447 July 19, 2001 23:3
I. Introduction
II. The Nucleus as a Probe of Molecular Structure;
Internal Interactions and the Effects of Motion
III. NMR Spectra of Liquids
IV. NMR in Solids
V. Pulse and Fourier Transform NMR
VI. Two-Dimensional NMR
VII. Multiple Quantum NMR
VIII. Medical Imaging
701
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
carcinogenic molecules. The manufacturer can use it to ments of their magnetic moments with respect to the static
study the morphology of polymer blends for constructing magnetic field [i.e., the nuclei develop discrete energies,
more durable battery casings. or states, the energy differences of which are in the radio-
frequency (millions of cycles per second, or megahertz)
I. INTRODUCTION region]. One may relate a frequency ν in cycles per sec-
ond, to an energy difference E by the relation E = hν,
Because some of the terms used to describe the phe- where h is Planck’s constant, h = 6.627 × 10−27 erg sec.
nomenon of nuclear magnetic resonance are a bit unfa- When such nuclei are exposed to radio-frequency radia-
miliar, a more familiar physical example illustrating these tion at the appropriate resonant driving frequency match-
terms is presented. ing the energy differences of nuclear states in a magnetic
To illustrate the meaning of “resonance,” recall that a field, then in much the same manner that the suspen-
car driven at a certain speed, generally about 80 km/hr, will sion system of a car will efficiently absorb energy from a
sometimes begin to vibrate due to a wheel out of balance. vibrating wheel, the nuclei undergo resonant absorption of
This vibration is an example of the phenomenon of “res- the radio-frequency energy from a driving source, which is
onance.” In a “resonant energy exchange,” at a particular a resonant alternating current circuit of the type used in ra-
driving frequency (in this case the angular frequency of the dio broadcasting. Such excitation disturbs the nuclei from
wheel), a system with a “resonant state” matching the en- their initial equilibrium states in the static field. Changing
ergy of the driving frequency will begin to absorb energy the driving frequency or the external magnetic field will re-
from an external source in a very efficient manner. The move the resonant condition, and the nuclei will cease ab-
system with the resonant state in this case is the spring sorbing energy. Following, and during, a resonant absorp-
suspension system of the car. When the out-of-balance tion of energy, the nuclei can relax from their excited states
wheel, which is vibrating at all speeds, reaches a vibration back to states of energy intermediate between their initial
frequency matching a “resonant frequency” of the car’s states before resonant excitation and their excited states.
suspension system, then there is a “resonant exchange” Two common relaxation processes for nuclei are termed
between the wheel and the remainder of the car, and the “spin–lattice” or longitudinal relaxation, and “spin–spin”
entire car begins to vibrate as the energy in the vibrating or transverse relaxation. In longitudinal relaxation, the nu-
wheel becomes efficiently transferred to the remainder of clei lose their energy to the molecular framework in which
the car. The energy transferred in a resonant exchange can they reside, and relax to their initial state, which may be
be stored, and if not released quickly enough, can serve to thought of as alignment of their magnetic moments along
destroy the system. In a car, that might mean a tire blow- the static magnetic field. In transverse relaxation, the nu-
ing out from repeated resonant bouncing. In a molecule, clei relax to a state in which the net magnetic moment
it could mean thermal decomposition. Alternatively, the perpendicular to the static field is created by many nuclei
system might relax from its high energy state by releasing having their spins pointing “in phase” in the same direc-
energy back to the driving source in an oscillatory fashion, tion in the transverse plane, and the nuclei lose their phase
or by releasing energy to its surroundings in some mono- coherence, and thus their net nuclear magnetism perpen-
tonically decreasing manner, perhaps in an exponentially dicular to the static field. In the simplest cases (e.g., that
damped fashion. For example, the out-of-balance wheel on of nuclei in molecules in many liquids), these relaxation
a car might simply exchange energy with the suspension processes are described by simple exponential forms. The
system of the car and oscillate indefinitely, if the speed of time constant for longitudinal relaxation is designated by
the car is maintained such that resonant exchange can take the symbol T1 . After a resonant excitation, the recovery
place. Another possibility is that the spring suspension of the nuclear magnetization parallel to the static field is
system of the car, receiving the vibrations of the out-of- proportional to (1 − e−t/T1 ). Note that this expression is
balance wheel, might eventually be shaken apart and fail. zero at time = t = 0 (i.e., immediately following the exci-
A driver of a car with a wheel badly out of balance might tation), and smoothly reaches unity at times longer than
automatically change the speed of the car to move away 5T1 . Similarly, the time constant for relaxation of magne-
from the resonant situation, and thus allow the vibration tization transverse to the magnetic field is called T2 , and
to damp away, or relax, by energy interchange between transverse relaxation is exponential in T2 .
the suspension system of the car and the surroundings; The phenomenon of nuclei absorbing resonant radio-
the rapid vibration might have heated the springs, and the frequency energy in a static magnetic field is called nu-
heat could have been dissipated in the passing air. clear magnetic resonance (NMR), and this phenomenon is
Nuclei possessing a magnetic moment, and placed in a always accompanied by nuclear relaxation. The details of
static magnetic field of from 5 to 10 T (for comparison, resonant absorption of radio-frequency energy by nuclei
the earth’s field is about .00005 T, or one-half Gauss) may in a magnetic field leads to the “spectrum” of absorption
be thought of as developing discrete, or quantized align- lines, and the relaxation processes are responsible for line
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
intensities and the widths of the observed spectral lines. ber” I , which is proportional to its magnetic moment M,
The study of spectra is called “spectroscopy,” hence the the proportionality factor being the gyromagnetic ratio
term NMR spectroscopy. It is seen, therefore, that at the γ ; M = γ I . The values of γ , and therefore the resonant
very least, an NMR experiment requires (1) a source of frequencies for NMR at fixed field differ for each nuclide.
radio-frequency (rf) oscillation at some radial frequency For example, 1 H and 13 C have spin quantum numbers
ω (rad sec−1 ) tuned to energy splittings between nuclear I = 1/2, whereas 6 Li and 174 Lu are spin 1. 27 Al is spin
magnetic states, (2) a magnet developing a static field B, 5/2. This spin quantum number imparts a special char-
to produce the split nuclear magnetic states, (3) nuclei acter to the nucleus’ ability to detect its local molecular
with magnetic moments M placed in a resonant radio- architecture, as will be seen in Section II. The NMR ab-
frequency (RF) circuit to absorb the rf energy at frequency sorption spectra, examples of which are shown in Sections
ω, and (4) some means of detecting this energy absorp- III and IV, are generally represented on an intensity (ordi-
tion. An NMR spectrometer is basically a high quality FM nate) vs frequency (abscissa) plot, and appear as a series
radio station and accompanying FM receiver. The carrier of peaks of various widths and shapes that are a reflection
is in the (video) megahertz region, generally between 5 of the local molecular environment of the nuclei under
and 700 MHz. The information content generally comes observation. This is to say that the local environments of
through in the (audio) kilohertz region, but the sounds nuclei in matter supply effective fields, Beff , which may
that are produced by resonating nuclei, when sent over be used to infer that environment.
an audio speaker, are generally fairly monotonic, and are We now inquire in more detail as to why NMR has this
not nearly as pleasant as those designed by a Mozart. The remarkable capability, and why this resonant spectroscopy
magnet is an expensive addendum to make a portion of is such a powerful tool, relative to other spectroscopies,
the experiment possible. such as ultraviolet and infrared spectroscopies.
The fundamental relation between the experimenter-
supplied parameters, B and ω, and the nuclear moment M
is II. THE NUCLEUS AS A PROBE OF
MOLECULAR STRUCTURE;
ω ∝ BM .
INTERNAL INTERACTIONS AND
The resonant frequency of absorption of energy of mag- THE EFFECTS OF MOTION
netic nuclei in a magnetic field is proportional to both the
strength of the field, and to the magnetic moment of the nu- A nucleus residing in a molecule, either in a solid or a liq-
cleus. The resonant condition for NMR may be achieved uid sample, has access to quite an intimate view of its local
by varying either B or the driving frequency. As alluded to molecular architecture. This nucleus senses the locations
previously, the local electronic and nuclear environment and types of its nearest neighbors, and in a diffuse man-
about a nucleus in a molecule, along with the external mag- ner, the bulk matter around it. In addition, this nucleus
netic field created by a magnet, contributes to the effective is sensitive to motion of its environment. The nucleus,
value of B. Thus the resonant NMR frequency is a finger- when properly interrogated with resonant excitations, can
print of the local electronic environment of the nucleus, give detailed information about its local molecular envi-
but depends upon the external magnetic field, which is ronment when that environment is motionless. In addition,
at the control of the experimenter. The magnetic moment the alteration of this information caused by molecular mo-
of a nucleus is a quantity fixed by nature, and is not an tion is used to infer details of such motion. It is this type
experimental variable. Table I lists all of the known mag- of information which, when properly interpreted as indi-
netic nuclides, their resonant frequencies in the absence of cated in the introduction, can lead to the wide variety of
interactions associated with the atomic or molecular envi- applications described there.
ronment at a field in which 1 H resonates at 100 MHz, and The sensitivity of the nucleus to its environment and
relavent added material, which will become more mean- to motion are all the result of the arrangements of molec-
ingful as further information is developed. Note that from a ular framework electrons and nuclei about the nucleus
quick glance at Table I, it is possible to infer that the physi- in question. The effects of this molecular framework
cian, the materials scientist, the chemist, the physicist, the upon the effective magnetic fields, and thus upon the
polymer chemist, the solid state scientist, the geologist, resonance frequencies of nuclei in matter are generally
and the engineer all have problems that may be attacked separated into four contributions, termed interactions:
with the help of NMR, since workers in all of these spe- these are designated (1) “shielding,” (2) “dipolar cou-
cialties deal with systems containing one or more of the pling,” (3) electric field gradients, or “quadrupolar cou-
nuclei listed. pling,” and (4) “scalar coupling.” These interactions are
Table I indicates that each magnetic nucleus has a num- all anisotropic. This means that they are directionally de-
ber of fingerprints. One is its “nuclear spin quantum num- pendent on the relative orientations of the static magnetic
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
1H 1
2 99.985 5.68 × 103 — 26.7510 100.0000
2H 1 0.015 8.2 × 10−3 2.73 × 10−3 4.1064 15.351
3H 1
2 — — — 28.5335 160.663
3 He 1
2 0.00014 3.26 ×10−3 — −20.378 76.178
6 Li 1 7.42 3.58 −8 × 10−4 3.9366 14.716
7 Li 3
2 92.58 1.54 × 103 −4.5 × 10−2 10.3964 38.864
9 Be 3
2 100 78.8 5.2 × 10−2 −3.759 14.052
10 B 3 19.58 22.1 7.4 × 10−2 2.8740 10.744
11 B 3
2 80.42 7.54 × 102 3.55 × 10−2 8.5794 32.072
13 C 1
2 1.108 1.00 — 6.7283 25.145
14 N 1 99.63 5.69 1.6 × 10−2 1.9331 7.226
15 N 1
2 0.37 2.19 × 10−2 — −2.7116 10.137
17 O 5
2 0.037 6.11 × 10−2 −2.6 × 10−2 −3.6264 13.556
19 F 1
2 100 4.73 × 103 — 25.181 94.094
21 Ne 3
2 0.257 3.59 × 10−2 9 × 10−2 −2.1118 7.894
23 Na 3
2 100 5.25 × 102 0.12 7.0761 26.452
25 Mg 5
2 10.13 1.54 0.22 −1.6375 6.122
27 Al 5
2 100 1.17 × 103 0.149 6.9704 26.057
29 Si 1
2 4.70 2.09 — −5.3146 19.867
31 P 1
2 100 3.77 × 102 — 10.8289 40.481
33 S 3
2 0.76 9.73 × 10−2 −5.5 × 10−2 2.0534 7.676
35 Cl 3
2 75.53 20.2 −8.0 × 10−2 2.6210 9.798
37 Cl 3
2 24.47 3.8 −6.32 × 10−2 2.1817 8.156
39 K 3
2 93.1 2.69 5.5 × 10−2 1.2483 4.666
40 K 4 0.012 3.52 × 10−3 (−)c −1.552 5.801
41 K 3
2 6.88 3.28 × 10−2 6.7 × 10−2 0.6851 2.561
43 Ca 7
2 0.145 5.27 × 10−2 −0.05 −1.8001 6.729
45 Sc 7
2 100 1.71 × 103 −0.22 6.4982 24.292
47 Ti 5
2 7.28 0.864 0.29 ±1.5084 5.639
49 Ti 7
2 5.51 1.18 0.24 ±1.5080 5.638
50 V 6 0.24 0.755 ±0.21 2.6491 9.970
51 V 7
2 99.76 2.15 × 103 −5.2 × 10−2 7.0362 26.303
53 Cr 3
2 9.55 0.49 ±3 × 10−2 −1.5120 5.652
55 Mn 5
2 100 9.94 × 102 0.55 6.6195 24.745
57 Fe 1
2 2.19 4.2 × 10−3 — 0.8661 3.238
59 Co 7
2 100 1.57 × 103 0.40 6.3472 23.727
61 Ni 3
2 1.19 0.24 0.16 −2.3904 8.936
63 Cu 3
2 69.09 3.65 × 102 −0.211 7.0965 26.528
65 Cu 3
2 30.91 2.01 × 102 −0.195 7.6018 28.417
67 Zn 5
2 4.11 0.665 0.15 1.6737 6.257
69 Ga 3
2 60.4 2.37 × 102 0.178 6.420 24.001
71 Ga 3
2 39.6 3.19 × 102 0.112 8.158 30.497
73 Ge 9
2 7.76 0.617 −0.2 −9.331 3.488
75 As 3
2 100 1.43 × 102 0.3 4.5804 17.123
77 Se 1
2 7.58 2.98 — 5.1018 19.072
79 Br 3
2 50.54 2.26 × 102 0.33 6.7023 25.054
(continues)
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
TABLE I (Continued )
Resonance
Atomic Natural Quadrupole Gyromagnetic frequency
weight/ abundance Sensitivity moment ratio (107 rad (1 H TMS
element Spin (%) (versus 13 C) (10−28 M2 ) T−1 sec−1 ) 100 MHz)
81 Br 3
2 49.46 2.77 × 102 0.28 7.2246 27.007
83 Kr 9
2 11.55 1.23 0.15 −1.029 3.848
85 Rb 5
2 72.15 43 0.25 2.5828 9.655
87 Rb 3
2 27.85 2.77 × 102 0.12 8.7532 32.721
87 Sr 9
2 7.02 1.07 0.36 −1.1593 4.334
89 Y 1
2 100 0.668 — −1.3108 4.900
91 Zr 5
2 11.23 6.04 −0.21 −2.4868 9.296
93 Nb 9
2 100 2.740 × 103 −0.2 6.5476 24.476
95 Mo 5
2 15.72 2.88 ±0.12 1.7433 6.517
97 Mo 5
2 9.46 1.84 ±1.1 −1.7799 6.654
99 Tc 9
2 100 1.562 × 103d −0.19d 6.0211 22.508
99 Ru 5
2 12.72 0.83 7.6 × 10−2 −1.2343 4.614
101 Ru 5
2 17.07 1.56 0.44 −1.3834 5.171
103 Rh 1
2 100 0.177 — −0.8520 3.185
105 Pd 3
2 22.23 1.41 0.8 −0.756 4.576
107 Ag 1
2 51.82 0.195 — −1.0828 4.048
109 Ag 1
2 48.18 0.276 — −1.2448 4.654
111 Cd 1
2 12.75 6.73 — −5.6714 21.201
113 Cd 1
2 12.26 7.6 — −5.9328 22.178
113 In 9
2 4.28 83.8 1.14 5.8493 21.866
115 In 9
2 95.72 1.89 × 103 0.83 5.8618 21.913
115 Sn 1
2 0.35 0.695 — −8.792 32.86
117 Sn 1
2 7.61 19.54 — −9.5319 35.632
119 Sn 1
2 8.58 25.2 — −9.9756 37.291
121 Sb 5
2 57.25 5.20 × 102 −0.53 6.4016 23.931
123 Sb 7
2 42.75 1.11 × 102 −0.68 3.4668 12.959
123 Te 1
2 0.89 0.89 — −7.0006 26.170
125 Te 1
2 7.0 12.5 — −8.4398 31.550
127 I 5
2 100 5.3 × 102 −0.79 5.3525 20.009
129 Xe 1
2 26.44 31.8 — −7.4003 27.658
131 Xe 3
2 21.18 3.31 −0.12 2.1939 8.200
133 Cs 7
2 100 2.69 × 102 −3 × 10−3 3.5087 13.116
135 Ba 3
2 6.59 1.83 0.18 2.6575 9.934
137 Ba 3
2 11.32 4.41 0.28 2.9728 11.113
138 La 5 0.09 0.43 −0.47 3.5295 13.194
139 La 7
2 99.91 3.36 × 102 0.21 3.7787 14.126
141 Pr 5
2 100 1.66 × 103 −5.9 × 10−2 7.836 29.291
143 Nd 7
2 12.17 2.31 −0.48 1.455 5.438
145 Nd 7
2 8.3 0.37 −0.25 0.895 3.346
147 Sm 7
2 14.97 1.26 −0.21 1.104 4.128
149 Sm 7
2 13.83 0.59 6 × 10−2 0.880 3.289
151 Eu 5
2 47.82 4.83 × 102 1.16 6.634 24.801
153 Eu 5
2 52.18 45.3 2.9 2.930 10.952
155 Gd 3
2 14.73 0.23 1.6 1.022 3.820
157 Gd 3
2 15.68 0.48 2 1.277 4.775
159 Tb 3
2 100 3.31 × 102 1.3 6.067 22.679
(continues)
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
TABLE I (Continued )
Resonance
Atomic Natural Quadrupole Gyromagnetic frequency
weight/ abundance Sensitivity moment ratio (107 rad (1 H TMS
element Spin (%) (versus 13 C) (10−28 M2 ) T−1 sec−1 ) 100 MHz)
161 Dy 5
2 18.88 0.45 1.4 0.881 3.295
163 Dy 5
2 24.97 1.59 1.6 1.226 4.584
165 Ho 7
2 100 1.03 × 103 2.82 5.487 20.513
167 Er 7
2 22.94 0.66 2.83 0.773 2.890
169 Tm 1
2 100 3.21 — −2.21 8.272
171 Yb 1
2 14.27 4.05 — 4.72 17.612
173 Yb 5
2 16.08 1.14 (—)b 1.31 4.852
174 Lu 1 — — — — —
175 Lu 7
2 97.41 1.56 × 102 5.68 3.05 11.407
176 Lu 7 2.59 5.14 8.1 2.10 7.872
177 Hf 7
2 18.50 0.88 4.5 0.95 4.008
179 Hf 9
2 13.75 0.27 5.1 −0.609 2.518
181 Ta 7
2 99.988 2.04 × 102 3 3.2073 11.990
183 W 1
2 14.28 5.89 × 10−2 — 1.1145 4.166
185 Re 5
2 37.07 2.8 × 102 2.8 6.0255 22.525
187 Re 5
2 62.93 4.90 × 102 2.6 6.0862 22.752
187 Os 1
2 1.64 1.14 × 10−3 — 0.6105 2.282
189 Os 3
2 16.1 2.13 0.8 2.0773 7.765
191 Ir 3
2 37.3 2.3 × 10−2 1.5 0.539 1.718
193 Ir 3
2 62.7 5.0 × 10−2 1.4 0.391 1.871
195 Pt 1
2 33.8 19.1 — 5.7412 21.462
197 Au 3
2 100 6.0 × 10−2 0.58 0.357 1.729
199 Hg 1
2 16.84 5.42 — 4.7912 17.911
201 Hg 3
2 13.22 1.08 0.5 −1.7686 6.612
203 Tl 1
2 29.50 2.89 × 102 — 15.3078 57.224
205 Tl 1
2 70.50 7.69 × 102 — 15.4584 57.787
207 Pb 1
2 22.6 11.8 — 5.5797 20.858
209 Bi 9
2 100 7.77 × 102 −0.4 4.2986 16.069
209 Po 1
2 — — — — —
235 U 7
2 0.72 4.9 × 10−3 4.1 0.479 1.791
a Most values taken from Brevard, C., and Grager, P. (1981). “Handbook of High Resolution Multinuclear
(1978). “NMR and the Periodic Table,” Academic Press, London, pp. 5–7; Pople, J. A., Schneider, W. G.,
and Bernstein, H. J. (1959). “High-Resolution Nuclear Magnetic Resonance,” McGraw-Hill, New York,
pp. 480–485; Harris, R. K., private communication.
c Poorly known or unknown.
d Franklin, K. J., Lock, C. J. L., Sayer, B. G., and Schrobilgen, G. J. (1982). J. Am. Chem. Soc. 104,
5303–5306.
field, and the three-dimensional coordinate system ori- the only contribution present. The observed spectrum of
enting the particular interactions. A nucleus that expe- many nuclei is effectively due to a single one of the above
riences all of the above four effects of the molecular contributions, so it makes sense to discuss them one at a
framework will, in general, have a set of resonance fre- time.
quencies that are a reflection of all of these contribu-
tions. The physical origins of each of these are now
A. Shielding; Chemical Shifts
discussed in turn, and for simplicity, the effect on the res-
onance frequency due to each of the four contributions NMR was discovered in 1945 independently by Bloch
from the molecular framework is discussed as if that were and co-workers and by Purcell and co-workers. It was
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
not widely used by chemists, however, until the discov- a spherical cloud, as it would be, for example, for a nucleus
ery, five years later by Procter and Yu, that a given nu- in an inert gas such as helium. When this cloud is exposed
clear type in different chemical environments (e.g., 14 N to a static magnetic field, it responds in accord with the
in the ammonium ion and in the nitrate ion in an aqueous laws of quantum mechanics by effectively developing a
solution of ammonium nitrate, NH4 NO3 ) would absorb field that algebraically adds to the static field. In other
radio-frequency irradiation at a frequency specific to the words, the nucleus is “shielded” from the external field
chemical environment; nitrogen in the ammonium ion ab- by the molecular electronic cloud about it. This shielding
sorbed at a frequency different than nitrogen in the nitrate interaction causes the magnetic field that is seen by the
ion. Subsequent to this discovery, NMR was rapidly as- nuclei in the molecule to be different than the static field
similated into the chemical laboratory for routine deter- due to the magnet in the NMR experiment. As mentioned
mination of structures of molecules in liquids. previously, the basic relation in the NMR experiment is
For many nuclei in liquids, the frequency differences of ω = MB. This relation, taking into account the fact that the
a given nucleus, associated with differences in chemical effective magnetic field at the nucleus is provided both by
functionality of the atom in which the nucleus resides, or the static, external field B0 supplied by the experimenter
shifts, known as chemical shifts from some standard, are and by the local magnetic field due to the environment of
of order of parts per million (ppm) of the applied field. For the nucleus, now becomes ω ∝ MBeff . Now, however, the
example, a proton in a methyl group and a proton in an OH product MBeff depends upon the direction of the external
group in methyl alcohol, CH3 OH, will have a frequency magnetic field relative to the nonspherical electronic
difference of roughly 7 ppm. Thus, at a resonant frequency environment producing the shielding shift. We, therefore,
of 300 MHz for the proton in the methyl group, the proton digress a moment to consider a physical picture of an
in the OH group will absorb at 300 × 106 (1–7 × 10−6 ) anisotropic shielding environment that will be useful in
Hz, or the difference in resonance frequencies between our discussion of all of the four effects of the molecular
1
H in CH3− and 1 H in OH will be 1,800 Hz; this is the framework upon resonance frequencies of nuclei in NMR
desired information in the audio region of frequencies. experiments.
Currently, the differences in resolution of absorption A useful pictorial representation of an anisotropic
lines of nuclei in different environments (to be clearly shielding environment is provided by an ellipsoid with
distinguished from quantitative detection, i.e., amount) three unequal axes, shown in Fig. 1. The relation between
available using the highest field magnets now produced, the frequency observed in an NMR experiment, ωobs , and
which currently are 14 T, and liquid samples, is equivalent this ellipsoid is as follows: If the ellipsoid represents the
to being able to distinguish two meter sticks standing half anisotropic shielding, or chemical shift interaction, the
a meter apart on the moon, when observing from the earth. observed angular resonance frequency when the external
One of the most powerful fingerprints of nuclei available magnetic field B0 is parallel to the x axis of the ellipsoid
to the practicing chemist, the chemical shift permits both would be given by the simple equation
a quantitative and qualitative analysis of the molecules
ωobs = γBeff = γB0 (1 − σx x ).
containing the nucleus under investigation, since individ-
ual chemical functionalities such as hydrogen in CH3 , Here, σx x is the magnitude of the x axis of the shielding
and hydrogen in OH can readily be distinguished, and ellipsoid. Clearly, γB0 represents the NMR frequency in
the intensity of the NMR lines corresponding to hydrogen
in these two different environments is proportional to the
number of hydrogens in that environment. For example,
the NMR spectrum of protons in pure methyl alcohol
would consist of two lines about 7 ppm apart, with an
intensity ratio I (CH3 )/I (OH) = 3:1. Further, protons
in all methyl groups resonate in a small frequency
range compared to the difference between protons in
CH3 and protons in OH. A similar statement applies to
protons in all hydroxyl groups, with some understandable
exceptions, so one talks of the “methyl group range of
absorption,” etc.
The origin of this “chemical shift” is that a nucleus in a
FIGURE 1 Representation of the anisotropy of an internal inter-
molecule (including infinite molecules such as metals) is
action as an ellipsoid. The principal axes of the ellipsoid represent
surrounded by an electron charge cloud that is a reflection resonant frequencies for absorption. The three angles orienting
of the chemical bonding about the nucleus, and that is in this ellipsoid with an arbitrary coordinate system represent the
general some complicated shape. This is to say that it is not other three independent pieces of information.
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
the absence of shielding, and the term −γB0 σx x is the fre- sociated with the fact that there are other magnetic nuclei
quency shift due to the anisotropic shielding along the x present. These nuclei act as little magnets, or magnetic
axis of the shielding ellipsoid. The observed resonance dipoles, and provide an additional local field to the nu-
frequencies with the external field parallel to the y and z cleus in question. This field represents a classical “through
axes of the shielding ellipsoid would be correspondingly space” interaction, and has no relation to the electronic
different if the values of σ yy , and σzz differ from σx x . With charge cloud present in molecules. The shape of this “dipo-
the convention σx x > σ yy > σzz , and B0 in some direction lar field” due to a nuclear magnet is of the form of the
other than parallel to one of these axes, the observed reso- pattern that iron filings take when spread around a bar
nance frequency for a given nucleus will lie between that magnet. Recall that this pattern varies both with direction,
with the field parallel to the x, and that with the field par- and distance from the magnet. Thus the effect of this field
allel to the z axes of the shielding ellipsoid. The spectral upon a neighboring nucleus will depend on where this
width (i.e., the range of resonance frequencies associated neighbor is located within the dipolar field. The shift in
with the anisotropic shielding interaction) for a sample in NMR frequency due to this dipolar field will similarly de-
which a nucleus is described by a shielding ellipsoid σ, pend on where the nucleus feeling the field is with respect
and all orientations of σ are present for fixed B0 , will be to the nucleus producing the field. The observed shift in
angular resonance frequency ωdip of a nucleus in the
ω = γB0 (σzz − σx x ).
presence of another nuclear magnetic dipole has spatial
Thus, a powdered sample of solid benzene, C6 H6 (solid), dependence
in which all protons are chemically identical in that they
ωdip = const · (1 − 3 cos2 θ)/r 3 ,
are all aromatic protons on a single benzene ring, will have
a proton NMR spectrum associated with the shielding in- where θ is the angle between the line connecting the inter-
teraction that is a powder average of the individual lines acting nuclei and the external magnetic field, and r is the
associated with the specific orientations for each benzene internuclear distance. The constant is proportional to the
molecule. This spectrum has been found to be roughly 6 magnitudes of the magnetic moments of the interacting nu-
ppm wide and looks roughly as shown in Fig. 4e, Sec- clei. We thus see that the resonance frequency shift due to
tion IV. What happens now when the solid sample is con- the dipolar interaction, if it could be measured without the
verted to a liquid? The benzene molecules are free to rotate interference of other interactions, provides a measure of
isotropically in solution. This means that the shielding el- nuclear geometries. It is notable that an interacting pair of
lipsoids characterizing the NMR lines associated with the dipolar nuclei have their resonance frequencies shifted by
shielding interaction are rotating isotropically. When the the inverse cube of their internuclear distance, so this shift
rotation frequency is faster than the spectral width char- is very sensitive to distance. Note that the above spatial
acterized by the difference (σzz − σx x ), then the observed dependence of the dipolar interaction between two nuclei
shielding frequency is characterized by the isotropic av- does not contain the azimuthical angle φ. This statement
erage σiso = (σx x + σ yy + σzz )/3, and the observed NMR translates into the fact that the ellipsoid describing the
spectrum associated with the shielding interaction is a sin- anisotropy of the dipolar interaction between two mag-
gle line at angular frequency ωiso = γ B0 (1 − σiso ). netic nuclei has two axes that are equal (i.e., there will
In exactly the same manner that the anisotropic shield- be a plane in which all resonance frequencies due to the
ing is represented by an ellipsoid with three unequal axes, pairwise dipolar interaction will be the same). This result
the anisotropy of the other three interactions may be so enforces a particular symmetry upon the NMR spectrum
represented. A complete specification of an anisotropic associated with a powdered sample of interacting dipolar
interaction would include the lengths of the three axes pairs, which is illustrated in Section IV, Fig. 3a. It is a fact
of the interaction ellipsoid, and the three angles that ori- that the isotropic value of the dipolar interaction, which
ent this ellipsoid with respect to some coordinate sys- is Diso = (Dx x + D yy + Dzz )/3 is zero. This fact will be
tem, such as the molecular framework in which the nu- subsequently important when the effects of motion on the
cleus resides, or a fixed coordinate system within the dipolar interaction are considered.
laboratory. Therefore, six independent pieces of infor- A large number of interacting dipoles in a powdered
mation completely specify an anisotropic interaction in sample would yield many resonance frequencies that
general. would reflect the powder average of the angular distri-
bution, (1 − 3 cos2 θ ), and the sum of all pairs i, j, with
internuclear distance ri j , so the spectrum of nuclei in such
B. Dipolar Interaction
a sample due to dipolar interactions alone would be a
In addition to the shielding field, mentioned previously, a broad, featureless spread just reflecting average geome-
nucleus in a molecule will experience a magnetic field as- tries and distances, as shown in Fig. 4c.
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
C. Electric Field Gradient; Quadrupolar Nuclei interaction for nuclei in the first two rows of the periodic
table.
The electric field gradient is simply the change in electric
To summarize the discussion, a nucleus that is not
field with direction due to the local distribution of nu-
moving has an NMR spectrum that is characterized by
clear and electronic charges at a particular point in space
a number of anisotropic interactions. These interactions
in which the nucleus in question is located. For exam-
may each be thought of as ellipsoids, with principal axes
ple, a sodium ion in a sodium chloride crystal would see
yielding NMR frequencies associated with the axis of
an electric field and an electric field gradient associated
quantization being along that axis. On isotropic rotation
with the presence of all neighboring Na+ and Cl− ions. In
with a rotational frequency fast compared to the spectral
this special case, the electric field gradient is zero because
width associated with the anistropy of any interaction in
the crystal symmetry is cubic. If this nucleus is magnetic,
question, the observed spectrum is a single line at the fre-
but has spherical nuclear charge symmetry (spin 12 , e.g.,
13 quency specified by the istropic value of the interaction.
C), then it is unaffected by a field gradient. If the nuclear
For example, the two major interactions of protons in solid
charge symmetry is not spherical (spin greater than 12 , e.g.,
benzene are shielding and dipolar. The shielding interac-
27
Al with spin 52 ), it can orient in an electric field gradient,
tion has a spectral width of about 6 ppm, or of 1,800 Hz at
which is to say that its nuclear energy levels that determine
a proton resonant frequency of 300 MHz, and the dipolar
the NMR spectrum are sensitive to the field gradient. The
interaction has a spectral width of roughly 20 kHz. On
spectrum associated with the central, 12 – 12 transition of
27 melting, the benzene molecules in the liquid are isotrop-
Al in an electric field gradient that has axial symmetry
ically rotating at frequencies much faster than 20 kHz,
for a sample of a powdered solid is shown in Fig. 3f, Sec-
with the result that the observed proton NMR spectrum is
tion IV. Thus, the NMR spectrum of a quadrupolar nucleus
associated with the isotropic values of the shielding and
associated with the presence of a nonzero electric field gra-
dipolar interactions. Because the isotropic value of the
dient is a measure of both local nuclear, and electron-cloud
dipolar interaction is zero, the observed spectrum does
geometries. The ellipsoid characterizing the spatial sym-
not reflect the dipolar interaction at all. It is simply shifted
metry of the electric field gradient is in general completely
from some reference by the isotropic value of the chemi-
asymmetric (i.e., E x x
= E yy
= E zz ). While in general the
cal shift only. The scalar coupling then acts to further split
isotropic value of the electric field gradient ellipsoid is not
the lines in the observed spectrum, as will be illustrated
zero, to a first approximation it may be taken to be so. This
later. If the nucleus in question is a quadrupolar nucleus
fact will be important in considering the effects of motion
(e.g., 23 Na, with spin I = 32 , then in addition to the effects
on the NMR of quadrupolar nuclei.
of shielding and dipolar interactions of, for example, Na
in NaNO3 , on the NMR frequency of Na, there will be an
effect of the local electric field gradient. However, since
D. Scalar Coupling
the isotropic value of the electric field gradient is almost
At this point we have seen that nuclear resonance frequen- zero, sufficiently rapid isotropic rotation of the molecu-
cies can be sensitive to just the total electronic distribution lar environment about the sodium nucleus, such as would
(shielding), or just to the local distribution of magnetic be experienced in an aqueous solution of sodium nitrate,
nuclei about the nucleus in question (dipolar interaction), will result in a sodium NMR spectrum that again reflects
or to the total distribution of nuclei and electron charge primarily the isotropic values of the shielding and scalar
(interaction of a quadrupolar nucleus in an electric field coupling.
gradient). The fourth interaction to which all nuclei re-
spond has a physical origin slightly different than any of
the previous. It is a type of dipolar interaction, but trans- III. NMR SPECTRA OF LIQUIDS
mitted from one nucleus to the other through the electronic
charge distribution in a molecule. More specifically, it is In Section II we have seen that although the nucleus in
transmitted by just that portion of the electronic cloud a molecule has an NMR spectrum that is a reflection
that touches both of the interacting nuclei. It is therefore of the entire molecular framework and the anistropy of
a measure of a portion of the total electronic charge cloud this framework about the resonating nucleus, in the liq-
in molecules. Because it is a dipolar interaction, its spatial uid state the resonance frequencies are simply a reflection
dependence is exactly the same as for the classical dipo- of isotropic shielding and scalar coupling. This fact sim-
lar interaction between two nuclei, with powder spectrum plifies the observed spectra of liquids relative to those
shown in Fig. 4a, Section IV. However, its magnitude, of solids. In addition, NMR has the capability of yield-
which depends on different physical factors, is quite dif- ing both a quantitative and a qualitative analysis simulta-
ferent, and in general smaller than the classical dipolar neously. The reason for this remarkable fact is that it is
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
possible to uniformly excite all of the nuclei in a liquid (bottom), its carbon framework structure, and the time de-
sample in an NMR experiment. The result is that the total cay of the magnetization under a pulse NMR experiment
nuclear magnetization observed in an NMR experiment is (see Section IV) used to obtain the spectrum in the Fourier
proportional to the number of the nuclei in each chemical transform mode. We shall have more to say about pulse
environment in all molecules under observation. Such is and Fourier transform NMR in a bit; for the moment, just
not the case for spectroscopies between the infrared and concentrate on the spectrum and the structure. This struc-
ultraviolet regions of radiation, which monitor rotational, ture has four methyne bridgehead carbons, H C, and six
vibrational, and electronic energy levels in molecules. methylene carbons, CH2 . A quantitative analysis of the
In these spectroscopies, different chemical functionalities areas of the two peaks shows that the areas are in the
within a given molecule can have vastly different transi- ratios 4:6, or 2:3, the same ratio as for the numbers of
tion probabilities (i.e., extinction coefficients). This is the carbons with the chemical identification indicated.
reason such spectroscopies, while diagnostic qualitative The 13 C NMR spectrum shown for adamantane in
tools, are not convenient quantitative tools without exten- Fig. 2a has been deliberately broadened to illustrate just
sive calibration procedures. the chemical shift spectrum in the absence of scalar cou-
In addition, for NMR of liquid samples, again unlike pling. Nevertheless, one sees that the two peaks are 10
visible absorption spectroscopies, the spectra of molecules ppm apart, with widths of roughly 5 ppm. This is quite
are especially simple and easy to interpret. For example, a fantastic resolution, being able to see differences in ab-
the chemical shift NMR spectrum of 13 C in adamantane, sorptionfrequencies in the parts per million. With a magnet
(CH)4 (CH2 )6 , consists of two peaks roughly 10 ppm apart. having a field of 14 T, carefully made uniform in order to
Figure 2a indicates the 13 C NMR spectrum of adamantane limit the NMR linewidths to the minimum possible value
FIGURE 2 (a) The NMR of 13 C in adamantane. Top, time decay; center, adamantane; bottom, NMR spectrum. The
3 : 2 ratio of the areas of the peaks represents the quantitative ratio of methylene to bridgehead methyne carbons. (b)
Spin–spin splitting at 13 C by 1 H in adamantane. Each proton acts as an extra field seen by the carbon, with direction
along, and against the external field. H splits C-doublet. The two methylene protons split the methylene carbon signal
into a triplet with intensity ratio 1:2:1. [High-resolution spectrum of 13 C in adamantane courtesy of Dr. Serge Lacelle.]
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
of the lifetime of the states involved in the transition, the further example of the use of NMR to elucidate the struc-
attainable resolution has been 0.001 ppm, or a part per ture of molecules in solution, the proton NMR of phenyl
billion. This resolution was the origin of the analogy of epoxide Novolak resin is shown in Fig. 3. The interpreta-
being able to see two meter sticks half a meter apart on the tion of the spectrum is given in the legend of the figure.
moon. If a mixture of adamantane and methanol were to Now we consider the effect of scalar coupling on
be the sample under investigation, with a resolution simi- the NMR spectrum of adamantane. We recall that the
lar to that exhibited for adamantane alone, the liquid-state isotropic value of the scalar coupling is nonzero, so this
NMR of 13 C in this sample would reveal three peaks, one value is an observable in the NMR spectrum of a nucleus
for each chemically different carbon in the sample, and the in a molecular liquid. This interaction may be thought of
intensities and peak locations would lead one to correctly as resulting from nuclei other than the nucleus under con-
identify the sample as a mixture of the two chemicals, sideration giving rise to small magnetic fields transmitted
with relative amounts indicated by the peak intensities. via the bonding electron glue to the nucleus in question.
Chemists, and more recently biochemists, have been uti- These small magnetic fields act further to split the ener-
lizing this capability of liquid-state NMR now for almost gies of the nucleus being considered, giving rise to further
thirty years. No chemical laboratory is complete without splitting of the spectrum that is quite diagnostic of the type
an NMR machine for structural studies. Almost every pa- and number of nuclei causing the splitting. For example,
per in the scientific literature reporting a new compound let the “nucleus in question” be the bridgehead methyne
containing C and H that can be put in solution will have an carbon, CH, in adamantane. Let the nucleus giving rise
NMR analysis as a routine part of the manuscript. As one to the splitting of this carbon be the attached hydrogen,
FIGURE 3 Proton NMR of phenyl epoxide Novolak resin. The monomer is shown above the spectrum. The line
increasing in value running from right to left is the integrated value of the peaks in the spectrum, thus providing a
quantitative analysis of the different types of hydrogen present in the polymer. The aromatic protons peaks at 6.69
and 6.83 ppm indicate that the rings are para substituted. The bridging methylene (CH2 ) protons overlap with the
epoxide band at 4.0 ppm.
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
with spin 12 , which means that the proton acts as a small and of the scalar coupling. It has been previously noted
magnet lining either with or against the external magnetic that at least six pieces of information characterize each
field. Then, as indicated in Fig. 2b, the methyne carbon anisotropic interaction responsible for nuclear resonance
resonance will be split into a doublet, with frequencies f 1 frequencies. Thus, with four interactions and six pieces
and f 2 , centered about the original unsplit signal shown of information each, there are 24 parameters that could
in Fig. 2a. Indicated at the right-hand side of Fig. 2b is in principle contribute to each observed NMR spectrum.
the energy-level diagram of the methyne carbon, split For a nucleus in a solid, such as 27 Al in an aluminosilicate
by its lone proton, and the transitions, which are vertical catalyst, all interactions will be present and will contribute
lines with a wiggle in the center. These transitions are at to the shape and width of the spectrum. The resultant spec-
frequencies f 1 and f 2 , and are seen in the high-resolution trum will be quite broad and featureless compared to the
spectrum of adamantane shown at the bottom of Fig. 2b. relatively sharp NMR absorption lines observed for alu-
At the left side of Fig. 2b are shown three peaks cor- minum in a liquid solution. In other words, resolution,
responding to the methylene carbon CH2 in adamantane, and thus information, has been lost. While there are in
split by its two attached protons. The manner in which principle 24 pieces of information available, they are not
these two protons split the carbon energy levels is indi- separable from each other in this case, and in fact lead
cated at the center of Fig. 2b. One proton supplies the to a loss of information because they obscure each other.
same type of splitting as shown for the methyne carbon, However, in many cases, there is one major contributor
and the second proton further splits that level as indicated. to the NMR spectrum of a nucleus in a solid. For exam-
The transitions associated with this splitting are indicated ple, the major contribution to the proton NMR of hydro-
by the “bathroom tile” diagram, at frequencies f 3 , f 4 , and gen in gypsum, CaSO4 ·2H2 O, is the dipolar interaction
f 5 . Note that there are two equivalent transitions at f 4 , between two localized protons. The major contribution
so the intensity of the center line will be twice that of to the NMR of deuterium in perdeuterated polyethylene
the two satellites. This experimental result is found in the would be the interaction of the local electric field gradi-
high-resolution spectrum, shown on the lower left side of ent with the quadrupole moment of the deuterium. Thus
Fig. 2b. the physical information inherent in the spectrum due to
Scalar coupling, or “spin–spin splitting” as it is com- single contribution (internuclear distances from dipolar
monly termed, is thus a diagnostic tool that adds consider- interactions, etc.) may be extractable.
ably to the power of the NMR spectroscopist to diagnose It is therefore useful to examine the powder spectra as-
details of molecular structure. sociated with each of the interactions in a solid. These
Structure determination of molecules in solution is by spectra are shown in Fig. 3. These “powder patterns” have
far the most common use of NMR at the present time. The quite characteristic forms for particular symmetries of the
amount of space devoted to this important application of interactions, and from these forms the types of structural
the technique in this discussion is thus not commensurate information mentioned in the introduction can be derived.
with its present use, but perhaps not out of line in view of For example, the powder pattern associated with the dipo-
future applications in other fields such as materials science lar interaction of two localized spin- 12 systems, with gyro-
and medicine. magnetic ratios γ1 and γ2 , is shown in Fig. 4a. The splitting
in frequency ν/Hz between the two sharp horns of this
spectrum is simply related to the internuclear vector ri j by
IV. NMR IN SOLIDS the formula
ν = 1.5hγ1 γ2 2πri3j .
In Section II, we discuss the effects of local arrangements
of the electron cloud and other nuclei about a resonating Thus, as stated in Section II, the frequency splitting of
nucleus in a molecule that are responsible for the effec- this famous “Pake doublet powder pattern” is proportional
tive magnetic field seen by this nucleus, and thus for its to the inverse cube of the internuclear separation and is
NMR absorption spectrum. Each interaction is discussed quite sensitive to this separation. For example, the two
in turn, as if it were the only interaction present, the jus- isolated protons at a distance of 0.248 nm (2.48 Å) in the
tification being that many timesthe NMR spectrum of a trichloroacetic acid dimer, (Cl3 CO2 H)2 , in the solid state,
given nucleus appears to be the result of one or two in- yield a Pake doublet with “horns” 11,771 Hz apart, inde-
teractions. A case in point is the NMR spectrum of a nu- pendent of the magnetic field. A change in the interproton
cleus in a liquid, discussed in Section III. In this case, all distance by 0.01 Å results in a frequency change of 80 Hz
information relating to anisotropy of the local environ- in the separation of the horns, and easily detectable change
ment effectively disappears, and the resulting spectrum is by NMR and an almost impossible change to detect us-
simply due to the isotropic portions of the chemical shift ing X-ray diffraction. The separations of localized pairs
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
1 1
FIGURE 4 NMR powder patterns of nuclei. (a–c) Dipolar coupling. (d,e) Shielding anisotropy. (f) Central 2–2 tran-
sition of a weak quadrupolar nucleus. (g) Scalar coupling anisotropy. (h) Lifetime broadening.
of protons in solids may be obtained from NMR much action so as to remove it from the observed result. This
more easily than from X-ray diffraction. What does one is the quite narrow line inside the broad line. At the bot-
do, however, when a given nucleus is exposed to a num- tom is an enlarged version of the narrow line, where the
ber of anisotropic interactions, all of which are roughly abscissa has been changed from kHz to ppm. In the spec-
the same magnitude, such that the solid state spectrum is trum shown, the resonant frequency was 56.6 MHz, so
featureless and less informative than the information from one ppm is 56.6 Hz. The full width of half height (fwhh)
just the chemical shift plus scalar coupling obtained in a is about 3 ppm, corresponding to about 170 Hz. Thus a
measurement of NMR in the liquid state? spectrum 20,000-Hz wide has been reduced to a spectrum
Fortunately, the introduction of transient techniques in about 200-Hz wide by manipulating the nuclear spins in a
NMR has led to the ability to remove, or attenuate some, manner now becoming available via transient techniques
and in favorable cases all, of the above intractions in a in NMR.
selective manner, such that all but the interaction desired In a similar manner, each of the other broadening inter-
to be seen remains. These techniques are outside the lim- actions can be manipulated to remove or scale their values
itations of the present writing, but may be viewed in de- such that the resulting spectrum reveals features associated
tail in some of the references supplied. An example of the with predominantly one interaction. This manipulation of
removal of broadening due to proton–proton dipolar inter- spin and real-space operators is a small branch of science
actions, without removal of shielding anisotropy, in a pow- in itself, and the subject of a number of texts referenced at
dered solid of high-density linear polyethylene (in which the end of this article. An interesting feature of these ma-
there is only one chemical species of proton, the methy- nipulations is the physical rotation of the samples at speeds
lene proton), is given in Fig. 5. Here, the powder spec- about that of dentists drills: about 180,000 rpm. A favorite
trum without (a) and with (b) transient techniques used to angle of orientation of the rotation axis with respect to the
separate proton–proton dipolar coupling from shielding static field is the so called “magic angle,” which is that
anisotropy are shown. In part (a) of Fig. 4 is shown the along the diagonal of a cube with respect to a cube edge:
powder spectrum of protons in polyethylene under a stan- 54.77◦ . If a sample is rotated at this angle in the static
dard experiment in which the spectrum is a result of both field at speeds large compared to the shielding anisotropy
shielding anisotropy and dipolar coupling. In this case, in cycles per second, the resulting broad powder spec-
the dipolar coupling dominates the spectrum, and the line trum associated with the shielding anisotropy is reduced
width is roughly 20 kHz. Also at the top the narrow line to its isotropic value (i.e., to a sharp line comparable in
is indicated resulting from manipulating the dipolar inter- width to spectra observed in liquids). Thus, for a system
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
equilibrium in a static magnetic field), then creating for ing”). The moment will in general process about this axis
a time tp another field (the rf field), perpendicular to the (center, Fig. 7), giving rise to an oscillating signal detected
static field. As indicated in the introduction, the basic re- by the experimenter (bottom, Fig. 7). This oscillation will
sponse of a nucleus in a magnetic field is to precess about in general be damped, with a time constant T2∗ such that
the field with a precession frequency ω = γ B. Therefore, the envelope of the oscillation is of the form exp[−t/T2∗ ].
during a pulse with spectral width ν = 12 tp , all nuclei The term T2∗ is called the transverse, or spin–spin, relax-
within this spectral width may be thought of as simply ation time. Its value offers an insight into motions of the
precessing about the B1 magnetic field of the pulse with sample in the zero frequency and 2ω0 frequency range.
angular precession frequency ω1 = γ B1 . If the pulse is The time constant characterizing the return of the ensem-
left on for a time tp , then the precession angle p is ble of nuclear spins back to the direction of the static field
given by is called the spin-lattice, or longitudinal relaxation time,
T1 . Its value gives information about motion in the fre-
p = γ B1 tp = ω1 tp /rad.
quency range of the precession frequency of the spins in
If tp ω1 set to π/2 radians, the nuclear magnetization will B0 , which is ω0 = γ B0 .
precess to a position perpendicular to its original orienta- Pulse experiments can be performed that characterize
tion. At this point in time, it is then free to process around other time constants, the description of which is beyond
the static field B0 . In accord with classical magnetism, a the scope of the present treatment.
rotating magnet creates a voltage in a coil arranged with In the previous discussion, we have concentrated on
its axis perpendicular to the axis of rotation of the magnet. “one-dimensional” data acquisition; intensity versus fre-
This oscillating voltage is the nuclear induction signal that quency. There are multidimensional techniques available,
is observed as the time decay and in turn is transformed which we now introduce.
into the spectrum. A classical picture of the process just
described is given in Fig. 7. At the top, the pulse field
rotates the magnetization to the transverse plane. The ex- VI. TWO-DIMENSIONAL NMR
perimenter views this magnetization by gazing at a fixed
axis in this plane (this process is known as “phase detect- In a one-dimensional NMR experiment, data are taken
as a function of a single time parameter, and the relation
between these data and the frequency spectrum is the pre-
viously discussed Fourier transform relation. Over the past
few years, a number of experiments have been developed
in which the time intervals in the NMR experiments are
divided into regions, a region t1 , followed by another re-
gion, t2 . The time domain signal, then, is a function of both
of these times; S(t) ≡ S(t1 , t2 ). An immediate result of this
statement is that the frequency domain signal, S(ω1 , ω2 ),
now becomes a three-dimensional contour plot, as shown
in Fig. 8.
Figure 8 is a two-dimensional plot in which chem-
ical shifts of the three different carbons in n-hexane,
CH3 CH2 CH2 CH2 CH2 CH3 , are plotted on the
“ω2 ” axis (going into the plane of the paper), and the
chemical shifts-plus-spin–spin couplings are plotted on
the “ω1 ” axis (parallel to the plane of the paper). The “ω1 ”
plot is what one would obtain in a 1-D NMR experiment
in which both chemical shifts and scalar (J) couplings are
simultaneously present. The “ω2 ” plot is what one would
obtain in a 1-D experiment in which the scalar couplings
of the protons to the carbons are averaged to zero by what
is called “decoupling,” accomplished by irradiating the
proton frequencies while the carbon signal is observed.
FIGURE 7 Classical picture of a pulse NMR experiment. Relation
between precessing moment (top and center) and the observed Clearly, there is less information on the ω1 and the ω2
transverse component of the magnetization as a function of time axes than in the 2-D plot shown in the plane, where it
(bottom). is obvious which chemically shifted carbons are attached
P1: GNH/FEE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN010C-493 July 19, 2001 20:30
FIGURE 13 (a) Scan through upper trunk perpendicular to spine. (b) Scan through upper trunk parallel to the spinal
cord. [Courtesy of Dr. John Schenck, General Electric Company.]
of characteristic fingerprints that can be used to probe their ation times available to the nucleus. The field of imaging is
environments. We have talked about fingerprints associ- just beginning to use these fingerprints for enhanced reso-
ated with shielding, dipolar interactions, quadrupolar in- lution. For example, there now exist “T1 images”, and “T2
teractions, and their reflections as seen in the various relax- images” that use the fact that transverse and longitudinal
relaxation times of nuclei in a given tissue are characteris-
tic of that tissue. The full range of interactions of nuclear
behavior has yet to be exploited for imaging.
Two recent developments which illustrate the burgeon-
ing power of NMR imaging (MRI) in medical, biologi-
cal, and materials science are (1) noninvasive diagnosis
of cancer by so-called chemical-shift imaging; (2) imag-
ing of live silk-butterfly pupae growing inside the cocoon;
and (3) imaging with the imaged body (live and human, or
inanimate, and a rubber band) outside of the magnetic sys-
tem, via the so-called NMR “Mouse.” Examples of each
type of image is shown in Fig. 14.
ACKNOWLEDGMENT
BIBLIOGRAPHY Harris, R. K., and Mann, B. E. (1976). “NMR and the Periodic Table,”
Academic Press, New York.
Hwang, S. J., King, T. S., and Gerstein, B. C. (1991). “Probing inter-
Becker, E. D. (1980). “High Resolution NMR—Theory and Applica- mediates in the reaction of ethylene over supported Ru: Utility and
tions,” 2nd ed. Academic Press, New York. limitations of the use of multiple quantum coherence,” Catalysis Let-
Ernst, R. R., Bodenhausen, G., and Waukon, A. (1987). “Principles of ters 8, 367–373.
Nuclear Magnetic Resonance in One and Two Dimensions,” Oxford Macomber, R. S. (1988). “NMR Spectroscopy: Essential Theory
Univ. Press, Oxford. and Practice,” Harcourt Brace Jovanovich College Outline Series,
Farrar, T. C., and Becker, E. D. (1971). “Pulse and Fourier Transform Harcourt Brace Jovanovich, San Diego.
NMR,” Academic Press, New York. Mehring, M. (1981). “High Resolution NMR of Solids,” Springer,
Fraissard, J. P., and Resing, H. A. (eds.). (1980). Magnetic resonance Heidelberg.
in colloid and interface science, NATO Adv. Study Inst. Ser, Ser C: Mountford, C., Doran, S., Lean, C. L., and Russell, P. (1980). “Cancer
Mathematical and Physical Science 61. pathology in the year 2000,” Biophys. Chem. 68, 127–135.
Fukushima, F., and Roeder, S. B. W. (1981). “Experimental Pulse Petrakis, L., and Fraissard, J. P. (eds.). (1984). “Magnetic resonance: In-
NMR: A Nuts and Bolts Approach,” Addison-Wesley, Reading, troduction, advanced topics, and applications to fossil energy,” NATO
MA. Adv. Study Inst. Ser., Ser. C, Mathematical and Physical Sciences
Fyfe, C. A. (1983). “Solid State NMR for Chemists,” C. F. C. Press, 24.
Guelph, Ontario. Roeder, S. B. W. (1981). “Experimental Pulse NMR: A Nuts and Bolts
Gerstein, B. C., and Dybowski, C. R. (1985). “Transient Techniques Approach,” Addison-Wesley, Reading, MA.
in NMR of Solids: Introduction to Theory and Practice,” Academic Slichter, C. P. (1978). “Principles of Magnetic Resonance,” 2nd ed.,
Press, Orlando. Springer, Heidelberg.
P1: LLL/LOS/GJM P2: GLM Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
Organic Chemistry,
Compound Detection
Raphael lkan
Bernard Crammer
Hebrew University of Jerusalem
459
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
Droplet countercurrent chromatography (DCCC) in which both of the frequency axes corresponds to
Separation technique based on liquid–liquid partition homonuclear shifts in a single spectrum. Also known
chromatography. as COSY.
Flash chromatography Also known as rapid column Infrared spectroscopy Technique in which many func-
chromatography. tional groups and atoms are characterized by their vi-
Fourier transform Technique in which a short power- brations and deformations in the 4000–200 cm−1 range.
ful radio-frequency pulse (microseconds) excites either Ion-exchange chromatography (IEC) Technique in
all the 1 H nuclei or 13 C nuclei simultaneously. Each which the stationary phase consists of a rigid matrix
nucleus shows a free induction decay (FID) which is (polymer), the surface of which carries a net positive
an exponentially decaying sine wave with a frequency (cationic) or negative (anionic) charge to give an ion
equal to the difference between the applied frequency exchange site R+ or Y− , respectively, which will attract
and the resonance frequency for that nucleus. and hold the counterions. Ion exchangers are divided
Gas–liquid chromatography (GLC) Technique in into anion and cation exchangers.
which the organic sample is carried through a column Lab on-a-chip This technique uses a planar device on
by a carrier gas (mobile phase) and the separation of which a number of chemical processes are being per-
the organic compounds occurs in the stationary phase formed in order to go from reactants to products or
(the column packing). The compounds are estimated from a sample to analysis.
by means of a detector. Liquid–solid (absorption) chromatography (LSC)
Gas–solid chromatography Technique in which the Separation of compounds (from a mixture) by a liq-
moving phase is a mixture of gases and the station- uid mobile phase and a solid stationary phase which
ary phase is a solid phase. The carrier gas such as ni- reversibly absorbs the solute molecules.
trogen or helium replaces the solvent in column chro- Mass spectrometry Technique in which a vaporized
matography. The solid may be finely powdered Celite sample of a substance is bombarded with a beam of
or kieselguhr. The technique is suitable for organic sub- electrons, and the relative abundance of the resulting
stances that are volatile without decomposition up to positively charged molecular fragments is determined.
about 300◦ C. The relative abundance versus mass-to-charge ratio that
Gel permeation chromatography (GPC) Technique is produced from the substance by the mass spectrom-
that separates substances according to their molecu- eter is called the mass spectrum of the substance.
lar size and shape. Three classes of stationary phases Matrix-assisted laser desorption-ionization mass spec-
are used: aerogels (porous glass), xerogels (crosslinked trometry (MALDI–MS) MALDI–MS technique en-
dextran), and xerogel-aerogels (polystyrene). ables ionization a broad range of materials, with high
1
H NMR “polarimetry” Technique using chiral shift sensitivity, broad mass range of materials, fast sample
reagents in NMR spectroscopy for the determination preparation, and the absence of fragmentation.
of enantiomeric purity of natural semiochemicals. Matrix isolation (MI) Technique used with Fourier
Heteronuclear shift correlation (HETCOR) 2D-NMR transform infrared spectrometer (FT-IR) to detect spec-
technique where a two-dimensional map is generated in troscopically transient chemical species at ambient
which one frequency axis corresponds to 13 C chemical temperatures.
shift or other nonproton nuclei, and the other frequency Nuclear magnetic resonance spectroscopy Measures
axis corresponds to the 1 H chemical shift. the absorption of light energy in the radio-frequency
High-performance thin-layer chromatography portion of the electromagnetic spectrum. 1 H NMR
(HPTLC) Technique enabling the separation of very spectroscopy furnishes indirect information about
complex mixtures of organic compounds. The plates the carbon skeleton of organic molecules. In 13 C
are prepared from optimized thin adsorbent layers. NMR peaks corresponding to all carbon atoms are
HPTLC offers greater separation efficiency through recorded.
smaller plate heights than the conventional TLC Paper chromatography (PC) Technique in which the
plates, shorter analysis time, and detection limits in fixed phase is a sheet of filter paper. The sample is
the nanogram and picogram range. placed near the edge of the paper as a small spot. The
High-pressure liquid chromatography (HPLC) Tech- edge is then dipped in the developing solvent (mobile
nique consisting of a stationary phase (a solid surface, phase). The solvent rises up the paper by capillary ac-
a liquid, an ion-exchange resin, or a porous polymer), tion taking the substance along. The positions of the
held in a glass or metal column with the liquid mobile spots are observed by visible or UV light or by spray-
phase being forced through under pressure. ing with a chromogen.
Homonuclear shift correlation (HOMCOR) 2D-NMR Raman spectroscopy (RS) Technique which involves
technique where a two-dimensional map is generated the analysis of the intensity of Raman scattering of
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
monochromatic light as a function of frequency of the compounds. In modern technology this involves the use of
scattered light. The RS technique provides complemen- chromatography (paper, thin-layer, gas–liquid, high-pre-
tary information to infrared spectroscopy. Generally ssure liquid); spectroscopy (infrared, Raman ultraviolet
strong vibrations in the Raman absorption are weak in and visible, nuclear magnetic resonance); mass spec-
the infrared absorption. trometry; and reaction chromatography (chemical reac-
Reaction GLC Certain chemical reactions of organic tions on thin-layer plates or gas chromatographic columns
compounds (such as reduction, oxidation, dehydration) which can be carried out prior to, during, or immedi-
that take place in a gas–liquid chromatographic col- ately after the chromatographic separation). Pyrolysis and
umn. The products are detected by means of the con- X-ray crystallography of organic compounds furnish im-
ventional detectors. portant structural information on the partial structures or
Spectroscopy Instrumental method of assigning struc- on the whole molecule, respectively. The combination
tural features and functional groups to organic or in- (and computerization) of chemical, chromatographic, and
organic molecules; such features are displayed by in- spectroscopic techniques has become a more efficient tool
tensities and patterns of spectroscopic signals. for the detection and identification of organic compounds
Supercritical fluid chromatography (SFC) Technique than any of these techniques.
in which the mobile phase (fluid) is maintained at tem-
peratures somewhat above its critical point. The mo-
bile phases used in SFC are gases such as freon, ethy- I. INTRODUCTION
lene, or carbon dioxide. It has superior solution prop-
erties and enhances the chromatographic separation It was only about 40 years ago that chemists had the tedious
of higher molecular weight compounds. The column task of identifying and characterizing unknown organic
packings used in SFC are the same as those used in compounds especially in the area of natural products. This
HPLC. may involve degradation of the molecule followed by syn-
Supersonic jet spectroscopy (SJ) Method which mea- thesis involving many steps. For example Woodward elu-
sures with high discriminating power utilizing laser- cidated the structure of strychnine in 1947 and seven-years
induced or lamp-induced fluorescence spectroscopy. later successfully synthesized this compound.
Thermal chromatography Volatilization of organic The advent of computers and Fourier transform com-
compounds at high temperatures and their separation pletely revolutionized the detection and identification of
by chromatographic techniques such as TLC. organic compounds. Modern automated instruments al-
Thin-layer chromatography (TLC) Chromatographic low very small samples in the nanogram (10−9 g) range to
technique in which the mobile phase is a liquid and be characterized in a very short time. The application of
the stationary phase is a thin-layer (usually 0.25 mm Fourier transform nuclear magnetic resonance (FTNMR)
thick) of an adsorbent (silica gel, alumina, cellulose) and Fourier transform infrared (FTIR) allows recovery of
spread homogeneously on a flat plate (usually a glass the sample in contrast to mass spectrometric (MS) deter-
plate) of various dimensions. mination which is a destructive but quite often necessary
Two-dimensional NMR spectroscopy (2D-NMR) Two- technique.
dimensional NMR refers to spectral data that are col- Modern methods especially in the separation of com-
lected as a function of two time scales—evolution and plex organic mixtures utilizing gas–liquid chromato-
detection. graphy (GLC), high-pressure liquid chromatography
Vacuumless spectrochemistry in the vacuum ultravi- (HPLC), and droplet counter-current (DCC) chromatog-
olet (VUV) Technique in which the emission wave- raphy can separate samples rapidly and efficiently in the
length region is below 200 nm. “Far ultraviolet” is more picogram range which until fairly recently has been im-
appropriate, since it includes access to this wavelength possible. Coupling the chromatographic instruments to
region by gas-filled optical techniques. spectrometers enables a partially automated analysis in
X-ray crystallography Method for determining the even less time. The following coupling of chromato-
molecular structure of crystalline compounds which graphic instruments has been performed: GC–MS, GC–
provides information on the positions of the indi- FTIR, GC–MI–FTIR, GC–UV–VIS, HPLC–MS, HPLC–
vidual atoms of a molecule, their interatomic dis- FTIR, HPLC–FTNMR and MS–MS. (Fig. 1).
tances, bond angles, and other features of molecular These semi-automated systems of analyzing and char-
geometry. acterizing small samples are vital to the natural product
organic chemist and biochemist for detection of highly
active substances in extremely low concentration in living
COMPOUND DETECTION in organic chemistry refers organisms. A typical example is in the field of pheromones
to the methods of separation and identification of organic which includes insect sex attractants which differ quite
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
FIGURE 1 Chromatographic and spectroscopic techniques for detection and identification of organic compounds.
GC, gas chromatography; GLC, gas–liquid chromatography; GSC, gas–solid chromatography; TLC, thin layer chro-
matography; HPTLC, high-performance thin layer chromatography; PC, paper chromatography; LSC, liquid–solid
chromatography; FC, flash chromatography; SFC, supercritical fluid chromatography; LLC, liquid–liquid chromatog-
raphy; DCCC, droplet counter current chromatography; PBC, bonded phase chromatography; HPLC, high pressure
liquid chromatography; IEC, ion exchange chromatography; EC, exclusion chromatography; GPC, gel permeation
chromatography; GFC, gel filtration chromatography; IR, infrared; UV, ultraviolet; NMR, nuclear magnetic resonance;
MS, mass spectroscopy; FT, fourier transform; T-MS, Tandem mass spectroscopy; MI-FTIR, matrix isolation fourier
transform infrared.
markedly in many insects. The concentration has often phy (GSC). Liquid chromatography is divided into two
been found in the 10−9 –10−12 g range. main types: column and planar chromatography methods,
namely, thin-layer chromatography (TLC) and paper chro-
II. CHROMATOGRAPHIC METHODS matography (PC). Column chromatography is subdivided
further into five major column types which constitute
There is an old Dutch word for chemistry, Scheikunde, HPLC. They are (1) liquid–solid chromatography (LSC),
which literally means the art of separation. Indeed, sep- (2) liquid–liquid or partition chromatography (LLC), (3)
aration methods from the chemical substance ultimately bonded-phase chromatography (BPC), (4) ion exchange
depends on separative operations. chromatography (IEC), and (5) exclusion chromatography
Chromatographic methods occupy a rather unique posi- (EC). The latter form includes gel permeation chromatog-
tion among modern methods in the field of detection, sepa- raphy (GPC) and gel filtration chromatography (GFC).
ration, and identification of organic compounds. The latest GLC and HPLC are the most widely used techniques
chromatographic methods provide simple techniques for for separation of organic compounds. However, they are
separation, purification, and structure determination of or- distinctly different techniques. For GC analysis the sam-
ganic compounds including the most complicated natural ple must be volatile and stable when the column is heated.
and synthetic macromolecules in biological and geologi- Samples for liquid chromatography must be soluble in
cal systems and their metabolic or breakdown products. mobile phase and differentially retarded by the station-
Chromatography permits the separation and partial de- ary phase. An active mobile phase increases the versatil-
scription of substances whose presence is unknown or un- ity of the liquid chromatographic technique. By contrast
suspected. As an exploratory method, chromatography is the mobile phase in GC is an inert gas which does not
indispensable in all sciences dealing with chemical sub- influence the separation. A variety of stationary phases
stances and their reactions. Chromatography is, in fact, (both polar and nonpolar) can be used. The versatility of
a physical method of separation in which mixtures are liquid chromatography depends on the liquid phase and
separated by distribution between two phases: a mobile solid absorbent. Gases, liquids, and solids can be ana-
phase and a stationary phase. The mobile phase can be a lyzed by GC. The normal range of molecular weights is
gas (as in gas chromatography) or a liquid (as in liquid from 2 to 500, although derivatives of carbohydrates rang-
chromatography). ing up to 1800 have been analyzed. Liquid chromatog-
Chromatographic methods can be classified according raphy is applicable to liquids and solids that are either
to the nature of the stationary phase. Gas chromatog- ionic or covalently bonded. Molecular weights of 32 to
raphy is divided into GLC and gas–solid chromatogra- 1,000,000 have been analyzed by LC. The flame ionization
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
the use of 10- to 100-fold higher flow rates and lower new methods such as droplet countercurrent chromatogra-
cycle times without significant losses in either resolution phy (DCCC), rotation locular countercurrent chromatog-
or capacity compared with the conventional soft gels and raphy (RLCC), and coil planet centrifugation. These
HPLC columns. methods have the advantage of being more rapid and less
solvent consuming than traditional CCC. Furthermore, the
advent of commercially available, compact apparatus has
III. FLASH CHROMATOGRAPHY led to a widespread acceptance of these new liquid–liquid
techniques as standard laboratory procedures for the sep-
The traditional method for preparative separations of sim- aration of natural products. The detection of compounds
ple mixtures of organic compounds by means of col- that are eluted from a DCCC can be performed by three
umn chromatography is giving way to flash chromatogra- methods: (1) UV detection for suitable UV-active sub-
phy (FC) or low-pressure liquid chromatography (LPLC) stances, (2) monitoring of the fractions that are collected
which was initially carried out by Still in 1978. FC is a by TLC, and (3) weighing of fractions after evaporation of
simple, cheaper, and faster technique for routine purifi- solvent. The majority of DCCC separations involve polar
cation of mixtures, allowing preparative separations from compounds, especially glycosides, which are often dif-
10 mg to more than 25 g. The resolution associated with ficult to purify. Chloroform–methanol–water systems of
FC on a standard 5 in. high column of 40 µm silica gel varying compositions remain the most widely used, in
is as great as 200 theoretical plates. This amount of sepa- view of the good formation of droplets and the convenient
rating power effectively means that compounds having Rf viscosity of this combination. The most notable develop-
values on analytical TLC as close as 0.1 may be reliably ments in the application of DCCC have occurred in the
separated with sample recovery of at least 90%. FC is ba- field of polyphenols, in particular in the separation of tan-
sically an air-pressure-driven hybrid of medium-pressure nins. DCCC has also been applied in the separation of
and short-column chromatography which has been opti- natural products such as alkaloids, triterpene glycosides,
mized for rapid separations. Elution of the components steroid glycosides, basic steroid saponins, and glycosides
is extremely rapid, usually taking about 5 to 10 min. As of flavonoids. Rotation locular countercurrent chromatog-
with other forms of chromatography, sample size is one raphy (RLCC) relies on the percolation of one layer of a
of the most important variables. The amount of sample two-phase solvent system through compartments (loculi)
that can be separated on a given column is proportional to that contain the second layer. During passage of the mo-
its cross-sectional area and the degree of separation of the bile phase, the loculi (connected into tubes) are constantly
components as indicated by TLC. Still et al. successfully rotated, to increase contact between the two phases. Ba-
separated a 1-g mixture of epimeric alcohols I and II with sically, RLCC has the same advantages as DCCC. As in
only a 65 mg mixed fraction in 7 min on a 40-mm diameter DCCC, the apparatus can be run in either ascending or de-
column using 500 mL of 5% ethyl acetate/petroleum ether. scending solvent modes but the formation of droplets is not
a necessary condition of RLCC. Consequently, a broader
range of solvent system is possible, and a system con-
taining ethyl acetate (often incompatible with DCCC) has
been used, for example, in the separation of flavonoids—
Reversed-phase systems with nonpolar bonded C8 and an important application of this method has been the
C18 silica are used in the separation of sugars. A mixture separation of enantiomers of (±)-norephedrine on an in-
of 200 mg of fructose and sucrose were completely strument consisting of 16 columns and each column con-
separated in the 3/1 acetonitrile/water mobile phase in- taining 37 loculi. The stationary phase was sodium hex-
corporating a silylamine bonded phase with silica in less afluorophosphate solution at pH 4, and the mobile phase
than 1 h. was (R, R)-di-nor-5-yl tartrate in 1,2-dichloroethane. Pre-
This method has been used for the separation of various sumably, the enantiomers of (t)-norephedrine form dif-
products of organic synthesis and small biomolecules. ferent diastereotopic complexes with the tartrate ester,
and these complexes are then partitioned differently be-
tween the two solvent phases. Separations by RLCC of
IV. DROPLET COUNTERCURRENT a range of natural products, including flavones, xanthone
CHROMATOGRAPHY glycosides, and antitumor antibiotics have been reported.
RLCC provides a useful complementary method to DCCC
The technique of countercurrent chromatography (CCC) in instances in which suitable solvent systems are not
has seen a rapid expansion following the introduction of available.
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
Range of application Selective Universal Very selective Very selective Universal Very selective
Minimum detectable quantity, g. 10−9 10−6 10−12 10−12 10−9 10−11
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
however, most of the sample co-extractives are transpar- chiral phases vary by the size of cyclodextrin macro-
ent and no longer interfere with the detection of aflatoxins. molecules and by substituents used to derivatize them.
Although much work is perfomed with detectors that op- Both factors greatly affect a column’s ability to resolve
erate in the UV region of the spectrum a good deal is also the enantiomers of different chiral compounds.
practiced in the visible region.
G. Enantiomeric Analysis of Amino
C. Fluorescence Detection Acids by HPLC
Fluorimetry is well known for its very high selectivity Most of the protein amino acids, except glycine, have at
and sensitivity to very small quantities of some samples least one asymmetric (chiral) carbon atom and can exist
occurring in biological fluids while being completely in- as two isomers, designated D and L enantiomers. When
sensitive to many other materials such as durgs, vitamins, the D and L enantiomers each react with a chiral (optically
and steroids. GC detectors such as electrical conductivity active) molecule of, for example, the L configuration, they
and electron capture are also used in LC. form the diastereoisomeric compounds DL and LL .
Amino acids of L configuration are more abundant in
nature; the D-enantiomers are usually found in bacte-
D. Radioactivity Detectors
rial cell walls, antibiotic compounds, and rare biological
There are many applications in the studies of the molecules. The DL-form (the racemate) are found in geo-
metabolism of drugs, pesticides, etc., in which radioactive logical specimens such as fossil shells and bones, or they
samples are employed to enable the compounds of interest are formed by abiotic synthesis.
to be detected at very low concentrations. The important Methods for resolving amino acids into their respec-
current trends in HPLC are the use of super critical fluids tive enantiomers are of importance in the preparation of
as eluants and the coupling of HPLC with NMR, MS, or peptides, drugs, and food additives as well as the extreme
FTIR. complex amino acid mixtures in meteorites in which the
An equally important trend is a reduction in column ready determination of optical purity is essential.
size and a concomitant increase in analytical speed. Short Two main approaches using HPLC are (1) covalent
columns can reduce analysis time, and efficiency may be bonding of chiral ligands (which can complex copper (II)
as high as 5000 theoretical plates since the particle size of ions) to solid supports (such as polystyrene and polyacry-
the packing material is 3 µm. The analysis time is about lamide) and resolution of amino acids by eluting with a
1 minute. A standard HPLC column has a diameter of mobile phase containing copper (II) ions; (2) introduc-
4.6 mm as compared to 2, 1, or even 0.5 mm for the new tion of chirality into the mobile phase. Metal ions such
columns. The small size of these columns allows the use as Cu(II), Zn(II), Co(II), and Mg(II), in conjunction
of more expensive solvents; deuterated solvents might be with chiral ligands are added to the mobile phase. Thus,
used when it is necessary to analyze the collected samples a Cu(II)–L proline complex as the chiral additive can be
by NMR. The use of ultramicrobore (50-µm diameter) operated in conventional cation-exchange resin.
columns allows feeding of the eluted peaks directly into a A good enantiomeric resolution of α-amino acids was
mass spectrometer of FTIR instrument. recently achieved by using chiral complexes of copper (II)
with N,N-di-n-propyl-L-alanine (DPA) as the additive in
the mobile phase. Actually, the mixture of amino acids is
E. Preparative Liquid Chromatography
separated into four groups by conventional ion-exchange
Most preparative HPLC columns are capable of separating chromatography and then resolved by means of the chiral
as much as 10 g of sample at a time. Industry often utilizes DPA reagent (Fig. 5).
columns 13 to 22 cm in diameter and 53 cm long having The relationship between temperature and enantiomeric
flow rates of 3 to 20 L/min. They can separate from 1 to separation may not always follow the same rules as con-
10 kg of sample per hour. ventional chromatography. Although the common liquid
stationary phases used in GC do not possess adequate
selectivity for enantiomeric separation, the addition of
F. Chiral Separations Using Capillary GC
derivatized cyclodextrin macromolecules to common sta-
Chiral capillary chromatography has increasing utility in tionary phases often creates the ability to separate volatile
flavor, fragrance, pharmaceutical, and industrial chemical enantiomers. Figure 6 shows that linalool and linalyl ac-
applications. The proper selection of a column is essential etate have symmetrical peak shapes and excellent chiral
since chiral stationary phases illustrate different capabil- separation on a Rt βDEXse at 25 mg per component on
ities and applications. The compositions of commercial column.
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
FIGURE 6
acids as been achieved with diastereomers, as well as with the identification of organic compounds than either of the
chiral stationary phases. two individual methods alone.
analysis time, reduction of sample weight, and also in- technique is usually integrated with a gas chromatograph,
creases the accuracy of determination. a mass spectrometer, or both. This arrangement enables
pyrolysis products to be analyzed immediately, so that
transfer losses and secondary degradations are minimized.
7. Combination of TLC and Vapor-Phase
The resulting chromatogram or mass spectrum may pro-
Chromatography
vide qualitative information concerning the composition
The combination of TLC and GLC enables a more com- or identity of the sample, quantitative data on its constitu-
plete separation and identification of complex mixtures of tion, or it may enable mechanistic and kinetic studies of
organic compounds. In most cases the mixture is subjected thermal fragmentation processes.
to preliminary separation by TLC followed by a complete The combined Py/GC/MS technique requires a minute
separation of the preseparated compounds of the mixture amount of sample, and the analysis time is very short.
by GC. The TLC–GLC technique has been applied in the Furthermore, such systems are now totally automated and
analysis of citrus and other essential oils and for determi- computerized.
nation of steroids in urine as well as unsaturated hydroxy Analytical pyrolysis has been used successfully in many
acid and glycerides which are separated on a thin layer of disciplines such as polymer chemistry, organic geochem-
silica impregnated with AgNO3 and then subjected to GC. istry, soil chemistry, forensic sciences, food science, en-
The fractions from the chromatoplates can also be altered vironmental studies, microbiology, and extraterrestrial
prior to running a vapor-phase chromatographic separa- studies involving meteorites and lunar samples. A large
tion. Thus, acids are converted to methyl esters, alcohols number of organic substances found in nature are unsuit-
and sterols to acetates, etc., prior to GLC. An application of able for direct analysis by modern techniques such as col-
the combination of TLC with GLC is the direct application umn chromatography and mass spectrometry. This may be
of the compounds to the thin-layer plate as they emerge due to their complex structure and polar and nonvolatile
from the exit tube of the GC. This technique was used for character.
the analysis of fatty acids, steroids, essential oils found A significant step in the evolution of analytical py-
in coffee essence, and many alkaloids. The association of rolysis was the combination of pyrolysis with a sophis-
TLC techniques with GC–MS was described for various ticated physiochemical technique for the efficient sepa-
separations and identifications of organic compounds. ration and/or identification of the fragments. In 1959, a
combined Py/GC system of polymers was introduced. Im-
provements in the analysis conditions were obtained by us-
C. Pyrolysis Gas Chromatography (Py/GC)
ing the high-resolution capabilities of capillary columns.
Prolysis is the thermal fission of naturally occurring In 1970, a continued interest in the analysis of extrater-
and synthetic polymers producing a range of smaller restrial samples such as lunar rocks and the Allende me-
molecules. Analysis of these products enables a profile of terorite (found in Mexico) was reported, using the mod-
the original compound to be reconstructed. The pyrolytic ified Py/GC/MS technique. Applying a pyrolysis/mass
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
spectrometry (Py/MS) technique such as the Curie point The thermal reactions that accompany the smoking of
system offers considerable advantages: rapid analysis, au- tobacco and cannabis furnished important structural in-
tomation of the system, and direct chemical information. formation on tobacco alkaloids. The pyrolysis of tobacco
Laser microprobe mass analysis (LAMMA) on milligram alkaloids yielded products such as quinoline and isoquino-
amount of sample has been reported. An alternative ap- line and nicotinonitrile which are probably derived from
proach is the detection of very small (picogram) amounts nornicotine and mysomine. Pyrolysis of natural polyenes
of organic matter by direct ionization due to laser irradia- such as β-carotene yielded ionene and small amounts of
tion. Thus, digitonin (MW 1228) and sucrose (MW 342) toluene, m-xylene, and 2,6-dimethylnaphthalene.
were readily detected. Although the identification and quantitation of drugs,
New techniques such as Py/GC/MS/DS (pyrolysis/gas e.g., narcotics and alcohol blood levels, are of forensic
chromatography/mass spectrometry–data system) and interest, the major impact of analytical pyrolysis in foren-
Py/MS/DS have been applied for detection and structural sic science is in the identification of complex natural and
elucidation of complex organic compounds. synthetic substances such as blood, skin, hair, wood, soil,
fibers, plastics, and waxes, which might be associated with
a suspect and found in the scene of the crime.
1. Pyrolysis of Synthetic Polymers
The thermal degradation of synthetic polymers has proven
to be an extremely important analytical technique for re- 4. Pyrolysis of Organic Geopolymers
vealing composition, structure, and stability profiles. It
Pyrolysis methods are used for detection and identification
actually causes unzipping of the polymer chain to yield
of natural organic polymeric products in the geosphere
sequential monomer units. Thus, thermal depolymeriza-
(such as humic substances and kerogens). The impetus
tion of rubber (at 700◦ C and atmospheric pressure) yielded
for much of this work has been provided by the US space
isoprene and dipentene, whereas polystyrene yielded a se-
exploration program, when the efforts in taxonomy, bio-
ries of monomeric and dimeric hydrocarbons.
logical molecules, and organic geochemistry which have
resulted in Py/GC/MS studies, were undertaken on the sur-
2. Pyrolysis of Biological Molecules face of the planet Mars. The pyrolysis technique is also
being used in the study of environmental pollutants (indus-
The rapid extension of analytical pyrolysis into diverse
trial processes, power stations, and fires). The use of the
fields such as taxonomy and soil chemistry has been
Py/GC/MS technique furnished important information on
largely due to progress in the pyrolysis of biological
the structure of humic substances, melanoidins and coals.
molecules. Such work has shown that the various classes
The Rock–Eval method uses a special pyrolysis device
of molecules with the possible exception of nucleic acids
(coupled with GC) in which a small sample of sedimentary
give highly characteristic pyrograms and has enabled the
rock (containing organic matter) is progressively heated
detection of the origin of fragments of complex samples
to 550◦ C. During the assay the hydrocarbons already
such as bacteria and soil constituents. Interest in the py-
present in the rock (S1 ) are first volatilized; pyrolysis of
rolysis of biological molecules is also due to the commer-
kerogen then results in the generation of hydrocarbons,
cial importance of thermal degradation processes, such
hydrocarbon-like compounds (S2 ), and oxygen contain-
as flame retardation, use of biomass such as algae for
ing volatiles, i.e., CO2 (S3 ) and water. Thus, S1 + S2 , ex-
production of oil and gas (energy), and thermal transfor-
pressed in kilograms of hydrocarbons per ton of rock, is
mations of organic constituents of food during cooking.
an evaluation of genetic potential (abundance and type of
Thus, the sensitivity and specificity of Py/GC/MS has en-
organic matter) of the rock.
abled octapeptides in infusion fluids and two closely re-
lated hormones, lypressin and felypressin to be detected
in nanogram amounts in aqueous solutions.
5. Extraterrestrial Matter
Coupled pyrolysis techniques have been used for the anal-
3. Pyrolysis of Drugs and Natural Products
ysis of extraterrestrial organic matter to provide evidence
Py/GC, Py/GC/MS, and HPLC/MS are routinely used for on molecular evolution within the solar system. Samples
the analysis of pure and formulated drugs or for the de- analyzed so far included meteorites collected on the sur-
tection of active principles and metabolites in body fluids. face of the earth, lunar rocks transported to earth and cer-
The Py/GC method is used for detection of sulfonamides, tain soils analyzed on the surface of that planet. The fully
barbiturates, and alkaloids such as morphine; heroine, automated Py/GC/MS/DS analysis of Martian soils re-
phenacetin and caffeine (often used in forensic science). vealed only the presence of water and CO.
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
VII. SUPERCRITICAL FLUID ons, ethylene, pentane, hexane, isopropanol, and carbon
CHROMATOGRAPHY dioxide. It has been observed that the solubility of vari-
ous solutes increases in supercritical phases. The column
In supercritical fluid chromatography (SFC), the mobile packings that are used in SFC are essentially the same as
phase is maintained at a temperature somewhat above its those used in HPLC.
critical point. Since the physical properties of a substance The instrumentation in SFC is quite similar to that of
in the supercritical state near the critical point are interme- modern high-resolution liquid chromatography. A number
diate between those of liquids and gases at ambient con- of modifications are made for LC to be suitable for opera-
ditions, it is designated as a fluid. For chromatographic tion with a supercritical fluid. The separation column can
purposes such a fluid has more desirable transport prop- be coupled to online detectors other than UV, including
erties than a liquid. SFC is superior to LC in separating MS, FTIR, FID, and other GC detectors.
efficiency and speed. In comparison to a gas, a fluid shows Carbon dioxide offers many advantages; it is inexpen-
about a 1000-fold increase in solution capabilities. The sive, available in high purity, and innocuous. Its near-
resultant enhancement in the migration rate of solutes is ambient critical temperature makes it attractive for use
especially valuable in the analysis of higher-molecular- with thermally labile compounds. The UV absorbance
weight compounds. Furthermore, some ionic solutes are of CO2 is minimal, thus allowing spectrometric detection
soluble in a supercritical fluid. This suggests that SFC down to 190 nm.
may be applicable to the analysis of compounds such as The advantages of SFC are high resolution per unit time,
the phospholipids, which cannot be volatalized for GC orthogonal column selectivity compared to GC and HPLC,
without decomposition. ease of fraction collection, and analysis of thermally labile
The number of compounds that can be analyzed by SFC molecules. It is likely that 20–40% of the solutes presently
is potentially enormous. Out of the 106 known compounds separated by HPLC are amenable to SFC separation with
which are currently more or less well characterized, only supercritical CO2 . It is possible that most separations car-
about 15% can be volatalized without decomposition. ried out by normal-phase HPLC may be handled with good
Compounds such as proteins, synthetic and natural poly- advantage by SFC.
mers, lipids, carbohydrates, vitamins, synthetic drugs, and Recent developments in SFC have not been inspired by
metal organic compounds may well be analyzed by SFC. any revolutionary new ideas. There has been progress in
When a liquid and its vapor in equilibrium with each two directions in the area of SFC. One direction parallels
other are heated in a confined space, the intensive prop- the development of capillary GC, the second direction is
erties of the two coexisting phases become increasingly similar to the development in HPLC. Because capillary
similar until, at the critical temperature, the two phases co- SFC is a high efficiency separation method, it is most
alesce into a fluid and acquire the same properties. When useful for the separation of complex mixtures. In combi-
this substance is heated beyond the critical temperature, a nation with universal (FID) detection, it can be used as a
supercritical phase is obtained, the substance is then called “fingerprint” technique.
supercritical fluid. Packed column SFC with particles of 3 to 10 µm can
For comparative purposes some physical properties of be used with mobile phases of relatively high densities.
a gas, a liquid, and a supercritical fluid are shown in At least 20,000 theoretical plates can be achieved within
Table III. The data in Table III show that the viscosity a few minutes on a routine basis.
of a supercritical fluid is comparable to that of a gas and A further advantage of SFC with respect to LC and
its diffusibility is between that of a gas and a liquid. GC is the separation of nonvolatile components and to in-
Any compound which is thermally stable to somewhat troduce them to a mass spectrometer in such a way that
beyond its critical point can theoretically be used as the both molecular weight information and reproducibly frag-
mobile phase. The mobile phases used in SFC include fre- mented spectra can be obtained.
FTIR is a powerful and highly specific detection tech- lose, alumina, polyamides and ion-exchangers, consider-
nique. The combination of SFC with FTIR has tremendous able versatility is available in the type of substances that
scope because of the possibility of separating and identi- can be separated.
fying compounds which are not amenable to GC–FTIR TLC is a microanalytical procedure and provides for
analysis. separations and at least tentative identification of sub-
SFC–FTIR is a very useful technique in characterizing stances in the milligram microgram, nonogram, and even
chemical additives in polymers. Many of these additives, picogram (pg) range. Adsorption TLC is very sensitive to
which include slip agents, plasticizers, UV absorbers, light differences in configuration that affect the free energy of
stabilizers and antioxidants are thermally labile or have a adsorption onto the layer surface and is, therefore, well
high molecular mass which makes them difficult to ana- suited to the separation of structural isomers. Quantitative
lyze by other methods. estimation of the separated compounds is carried out in
The biological or pharmacological activity and effec- situ by densitometric estimation of the TLC plates.
tiveness of chiral molecules depend largely on their con- TLC has recently been transformed by the introduction
figuration. Often only one of the enantiomers is pharma- of new and better plates, accurate and precise autospotters,
cologically beneficially active, while the other enantiomer new development techniques, and rapid scanning densit-
may not only be inactive but also toxic. Most of the chiral ometers into a sophisticated, instrumentalized, and quan-
stationary phases for TLC, GC, and LC of the chiral- titative technique.
cavity type act by allowing selective occlusion or interca- With respect to the TLC plates, there are two major
lation of one enantiomer into chiral cavities in the phase advances. The first was the introduction to high perfor-
matrix. mance TLC (HPLC) plates produced from silica gel of
between 5 and 10 µm compared to the 12 to 25 µm
commonly used for standard TLC plates. This resulted
B. Superficial Fluid Chromatography (SCF)
in more rapid and efficient chromatographic separation
and SCF/MS for Analysis of Complex
and reduced bond broadening, thus producing higher sen-
Hydrocarbon Mixtures
sitivity. Second were a large number of bond-phases, of
The physical properties of supercritical fluid mobile the type originally produced for HPLC, such as C2 , C8 ,
phases allow chromatographic separations of many com- C12 , C18 , aminopropyl, cyanopropyl, diphenyl, and the so
pounds that are generally separated by LC. The density of called chiral phase which are used in TLC.
a supercritical fluid is typically 100 to 1000 times greater The chiral plates are used extensively for the enan-
than that of a gas, which imparts “liquid-like” solvating tiomeric separation of racemic amino acids and their
properties. However, the viscosities and diffusivities of su- derivatives by means of a ligand exchange mechanism.
percritical fluids remain similar to those of gases. Conse- Detection and quantitation of analysis on the TLC plates
quently, high-solvating power mobile phases that behave have progressed considerably in recent years. A range of
similarly to gases define the advantages of SFC. These microprocessor-controlled scanning densitometers is now
properties also allow successful interfacing to MS. The available, capable of providing UV and fluoresence de-
immediate application of SFC/MS in complex hydrocar- tection at the nanogram level. Chromatographic processes
bon analysis is for mixtures containing nonpolar to mod- have been reported to be filmed on video.
erately polar compounds that are thermally labile or of
relatively high molecular weight.
A. High-Performance Thin-Layer
Chromatography
VIII. THIN-LAYER CHROMATOGRAPHY High-performance TLC enables us to carry out the most
complicated separations. The HPTLC plates are prepared
Thin-layer chromatography (TLC) is a subdivision of liq- from optimized (e.g., particle size and particle size distri-
uid chromatography (LC) in which the mobile phase is a bution) adsorbent layers and extremely even surfaces. The
liquid and the stationary phase is situated as a thin layer HPTLC plates offer greater separating efficiency (plate
on the surface of a flat plate. TLC is sometimes grouped number and resolution) through smaller plate heights than
with paper chromatography under the term planar liquid the conventional TLC plates. Shorter analysis time, de-
chromatography because of the flat geometry of the paper tection limits in the nanogram range with UV adsorption
or layer stationary phases. TLC is a simple, rapid, ver- detection and in the picogram range with fluorometric de-
satile, sensitive, inexpensive analytical technique for the tection are additional advantages.
separation of organic substances. Since numerous station- The HPTLC plates may typically deliver some 4000
ary phases (sorbents) are available, such as silica, cellu- theoretical plates over a distance of 3 cm in 10 min. This
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
compares to typical values for conventional TLC plates of detection reagent iodine can be used as a 1% alcoholic so-
some 2000 theoretical plates over 12 cm in 25 min. lution spray, but more frequently, the plate is simply placed
in a closed container containing a few iodine crystals. The
iodine vapor forms weak charge-transfer complexes with
B. Impregnated Layers
most organic compounds which show up as brown spots
The range of applications of both TLC and HPTLC is con- on a pale yellow background within a few minutes. Sen-
siderably expanded by means of different impregnation sitivities in the 0.1–0.5-µg range are often obtained with
agents such as acids, bases, or salts added to layers in var- iodine.
ious concentrations. Stable hydrophilic stationary phases Charring reagents (H2 SO4 ) are suitable for glass-
are formed by treatment with agents such as formamide, backed layers with inorganic (e.g., gypsum) binders only.
DMF, ethylene glycol, and various buffers. Lipophilic sta- Many charring reagents produce colored zones when heat-
tionary phases for reversed-phase TLC are obtained by im- ing is carried out at relatively low temperature; they form
pregnation with liquid paraffin, undecane, and mineral and black zones at higher temperatures.
silicone oils. Impregnation with specific reagents aids the Spraying of a chromatogram with a 5% solution of
separation of certain types of compounds such as AgNO3 phosphomolybdic acid followed by a brief heating at
for compounds with double bonds, boric acid or sodium ar- 110◦ C gives dark blue spots against a yellow background
senite for vicinal dihydroxy isomers; sodium bisulfite for with a large variety of organic compounds.
carbonyl compounds, and trinitrobenzene or picric acid A solution of Rhodamine B produces violet spots on a
for polynuclear aromatic compounds. pink background. Antimony trichloride or pentachloride
solution in carbon tetrachloride produce spots of different
characteristic colors with many organic compounds.
C. Detection and Visualization
Over 300 spray reagents are known to react more or
Following development, chromatograms are removed less specifically with different functional groups to reveal
from the chamber and are air- or oven-dried to remove natural products and organic or biochemicals as colored or
the mobile phase, zones are detected by various means. fluorescent zones. Table IV contains a selection of specific
Colored substances may be viewed in daylight without detection reagents. Methods for the quantitation of thin-
any treatment. Detection of colorless substances is sim- layer chromatograms can be divided into two categories.
plest if compounds show self-absorption in the short- In the first, solutes are assayed directly on the layer, either
wave ultraviolet (UV) region (254 nm) or if they can by visual comparison, area measurement or densitometry.
be excited to produce fluorescence by short-wave and/or In the second, solutes are eluted from the sorbent before
by long-wave (365 nm) UV radiation. Otherwise, detec- being examined further.
tion can be achieved by means of chromogenic reagents
(producing colored zones), fluorogenic reagents (pro-
D. Radiochemical Techniques
ducing fluorescent zones), or by biological enzymatic
methods. Radioactive isotopes are widely used as tracers or labels
Enzymatic reactions can be monitored on the plate, for substances separated by TLC for following the causes
and the end products can be detected. Biological test of chemical and biochemical reactions, determining the
procedures are used in the specific detection of biolog- distribution of substances in a reaction mixture, elucidat-
ically active compounds. Thus, detection of hemolyz- ing metabolic pathways of drugs, pesticides, pollutants,
ing compounds such as saponins is achieved by casting and natural substances in human, animal and plant tis-
a blood–gelatin suspension on the layer and observing sues, and assessing the purity of isotopes. The most widely
hemolytic zones that are transparent and nearly color- used labelled substances in TLRC contain 3 H (tritium)
less on the turbid red gelatin layer background. An- and 14 C.
other means of detection is the use of Geiger or flow The detection of radioactive substances on TLC plates
counters or other specialized means to locate radioactive is carried out by liquid scintillation, film registration or
solutes. autoradiography, and by direct scanning. A thermomicro
Detection reagents may be impregnated into the procedure for rapid extraction and direct application in
layer prior to sample application and development. TLC is the thermomicro application of substances (TAS)
Chromogenic reagents are of two types: (1) general method.
reagents that react with a wide variety of different com- The TAS method is a procedure for the isolation and
pound types and can totally characterize an unknown sam- separation of many substances from solid materials and
ple, and (2) specific reagents that indicate the presence of their direct transfer to the starting line on a TLC plate. This
a particular compound or functional group. The universal is performed by application of heat, either by distillation
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
or sublimation. The emerging vapors are deposited as a layers according to the degree of unsaturation. The sep-
spot on the TLC-plate which is then chromatographed in arated fractions are eluted and the products subjected to
the usual way. The spot can also be scraped off, extracted GLC separation. Furthermore, the resolved methyl esters
and analyzed by GLC. may be treated to reductive ozonolysis and the result-
This technique is useful in the fields of drugs, phy- ing fragments analyzed by GC, thus furnishing important
tochemistry, food additives, and other natural and syn- structural information.
thetic organic compounds. Many organic constituents of
drug-containing plants were detected by this technique. 2. GC–TLC
Examples are constituents of essential oils, purines (e.g.,
caffeine), narcotics from drugs or tobacco, marijuana con- A recent application of the combination of TLC with GLC
stituents, mescaline from Mexican narcotic fungi, organic has been the direct application of the compounds to the
contituents of fossil fuels, and oil-shales. thin-layer plate as they emerge from the exit tube of the
gas chromatographic apparatus.
E. Coupled TLC Techniques It should be pointed out that GLC separates accord-
ing to the relative volatility, and TLC separates according
1. TLC–GC to the functional groups present. A coupling device was
There are a number of ways in which TLC can be com- constructed in which the thin layer plate is moved logarith-
bined to advantage with GLC. The spots obtained from mically with time, while the gas chromatogram operates
TLC may be eluted, concentrated, and then subjected to under isothermal conditions.
GLC analysis. This method has been used in the analyses
of lipids, steroids, alcohols, fatty acids, esters, glycerides,
3. In Situ Chemical Reactions on TL Plates
hydrocarbons, essential oils, and many other natural and
synthetic organic compounds. Methyl esters of fatty acids In order to detect and identify minute quantities of organic
are first separated on silver-nitrate-impregnated silica gel compounds the sample is spotted on a plate which is then
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
covered with a reagent. After a very short time, (mostly a It should be pointed out that typical measurements in a
few seconds or minutes) the plate is developed in a suitable laboratory take several hours to perform, but many of the
solvent system whereupon the reaction products are sepa- desired applications of on-chip analytical systems require
rated and identified by color or by spectroscopic methods. data in seconds, or at most a few minutes. Fast analysis is
A great deal of information can be gained from the in therefore an important aspect of miniaturization.
situ reaction technique with the expenditure of very small
amounts of material. The following in situ reactions on TL
IX. FIELD-FLOW FRACTIONATION (FFF)
plates have been carried out so far: acetylation, dehydra-
tion, formation of derivatives: (acetates, dinitrobenzoates,
FFF is a chromatographic-like separation technique which
DNPS, methyl esters, phenylisocyanates, and semicar-
is designed for fractionation of macromolecules, colloids,
bazones), diazotization, esterification, halogenation, cat-
and particles, primarily on an analytical scale. While the
alytic hydrogenation, acid and alkaline hydrolysis, iso-
chromatographic separation is achieved by differential
merization, nitration, oxidation, photochemical reactions,
distribution of species between one stationary phase and
reduction, and Diels–Alder reactions.
one moving phase. FFF uses only one moving phase.
However, there are velocity difference within the moving
F. Lab on-a-Chip phase, and separation is achieved by differential distribu-
tion of species between the regions of different velocity.
A Lab on-a-chip is a planar device on which or in which
FFF has broad application.
a number of chemical processes are being performed in
Sedimentation FFF has been used for the fraction-
order to go from reactants to products or from a sample
ation of polystyrene, latex beads, emulsions, artificial
to analysis. There is a distinction between microfluidic
blood, viruses, and aqueous colloids, liposomes, albumin
devices and microarray devices and DNA analysis.
spheres, and DNA. Thermal FFF has been applied to dif-
DNA microchip arrays contain from several hundred to
ferent types of synthetic polymers. A recent development
several thousand immobilized DNA reagents. They pro-
in FFF has been the increase of separation speed to allow
vide a systematic way to survey DNA and RNA variation
fractionation on a minute scale instead of hours.
and may well provide a standard tool for molecular bi-
ology research and clinical diagnostics. The peripherals
such as MS and fluorescence detectors sit off-chip and are X. ION CHROMATOGRAPHY (IC)
typically benchtop instrument sized.
At present, the typical commercially produced mi- lon chromatography (IC) usually contains an ion ex-
crochip used for miniaturized chemical systems consists changer or ‘pseudo ion exchanger’. The ion exchangers
of a 2- or 3-cm-square sliver of silicon, glass, quartz, or are derived from cross-linked organic polymers by affix-
plastic that is 50 µm. The chip is covered with a plate ing to the polymer ionogenic groups that are the source or
to contain the samples and reagents. At present it is pos- the vital ion exchange processes. The IC is used for the
sible to inject volumes as low as 1 pL onto chips with determination for both inorganic or organic compounds
miniaturized chemical systems. and even species as ‘non-ionic’ as carbohydrates.
For its operation, a microfluidic chemical device re- The determination of ions in environments as diverse
quires a number of microcomponents, such as microfilter, as brine, milk, diesel exhaust, plating baths, urine and
microreactors, and microseparation columns, depending foodstuffs is a measure of its versatility. Improvements in
on the application. It also needs some means of driv- resins and other stationary phases and in the sensitivity
ing the fluids around the chip. Recently, a microtech- and automation of IC techniques should help to sustain
nology for producing “bioprocessors-on-a-chip” was de- the growth in this new area of analytical technology.
veloped. Such chips can be described as microfactories
that can within minutes diagnose infections or diseases
by separating bacteria, fatal cells, or cancer cells from XI. CAPILLARY
blood. This technology relies on dielectrophoresis for sep- ELECTROCHROMATOGRAPHY (CEC)
arating bioparticles, such as cells, viruses, bacteria, pro-
teins, and DNA with a high degree of selectivity and CEC may be regarded as a combination of high-per-
sensitivity. formance capillary electrophoresis (HPCE) and HPLC—a
With improvement in layout and packaging of mi- hybrid technique combining the best of both worlds.
crochips, it might even be feasible to integrate the de- CEC is a derivative of HPCE, wherein a packed bed
tection and computer circuiting on the chip. capillary is used. The essential difference between HPCE
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
and CEC is in the nature of the separation. If the separation A. Infrared Spectroscopy
is achieved on the basis of differential partitioning between
Infrared (IR) spectroscopy is probably the quickest and
two phases, then it is chromatography. If it uses both, it is
cheapest of the spectroscopic techniques in determining
referred to as electrochromatography.
the functional groups of the sample. The samples can be
Compared with HPLC, CEC uses electroosmotic flow
solids, liquids, or gases and can be measured in solution or
(EOF) rather than high pressure to force the mobile phase
as neat liquids mulled with KBr or mineral oil. Compari-
through the capillary. The advantage of EOF for CEC
son of IR spectra of substances of known structure has led
is that the flow profile is flat and thus precludes band
to many correlations between wavelength (or frequency)
broadening by trans-channel of radial diffusion. The re-
of IR absorption and features of molecular structure. Cer-
sult is that the number of plates is at least double that
tain structural features can easily be established. For ex-
of HPLC. Another advantage of CEC is that the pack-
ample, in an organic compound that contains only C, H
ing particles are smaller than those of today’s HPLC
and O, the oxygen can only be present as C O, O H, or
systems.
C O C or a combination of these, such as the ester or
Since there is no back pressure resisting the pump’s
carboxylic acid group.
ability to deliver the solvent at a given flow rate, as in
The presence or absence of absorption in the car-
HPLC, capillaries can be made longer. Thus, it is prac-
bonyl region (1730–1670 cm−1 ) or hydroxy region (3700–
tical to utilize packings as small as 1 µm. These can
3300 cm−1 ) can serve to eliminate or establish some of
be porous or nonporous, spherical or irregular, coated or
these possibilities. One simple application of IR is to de-
uncoated, pure silica or mixed mode (silica/polymer, sil-
termine whether two samples are identical. If the sam-
ica/alumina, etc.). The combination of longer capillaries
ples are the same, their IR spectra (obtained under identi-
packed with smaller-diameter particles makes it possible
cal conditions) must be the same. If the two samples are
to obtain and use columns with 100,000–500,000 and per-
both pure substances very similar in structure, the dif-
haps even 1,000,000 plates per meter.
ferences in the spectra may be so small that it would
In conventional HPLC, achieving more than about
not be easy to detect them; it may even be beyond the
25,000 plates per meter requires coupling columns and
power of the instrument to detect them. The absorption
long analysis times.
peaks of the spectrum of an impure sample is usually
Using small internal diameter columns and small par-
less intense than those of a pure sample and the spec-
ticle diameters it is possible to realize somewhat higher
trum will show additional peaks. The IR spectra of enan-
plate counts. It took 30 years, but today, capillary columns
tiomers are identical but those of diastereoisomers are
account for nearly 75% of the GC separations.
different.
Simple mixtures can be determined from their IR spec-
tra. For example mixtures of cycloalkanones show charac-
XII. SPECTROSCOPIC METHODS
teristic differences of the carbonyl group: cyclobutanone
(1788 cm−1 ), cyclopentanone (1746 cm−1 ), and cyclohex-
There has been a profound impact on the application
anone (1718 cm−1 ). IR spectroscopy has been applied to
of spectroscopic techniques in the detection of organic
deuteration of organic samples. It is found that the C H
compounds especially in the areas of natural products
stretch bands disappear from the 3000 cm−1 and the C D
and polymers. The commercially available Fourier trans-
stretch bands appear in the 2200 cm−1 region. It is possible
form signal handling techniques are rapidly revolutioniz-
by calibrating pure compounds to determine the isotopic
ing nuclear magnetic resonance (FTNMR), infrared spec-
purity in routine analysis.
troscopy (FTIR), and Raman spectroscopy (FTRS), as
Modern IR analysis utilizing high-performance detec-
well as prototypes of mass spectrometry (FTMS) which
tors can generate spectra (4000–200 cm−1 ) in 60 sec. Sam-
are not yet available commercially. Previously organic
ple amounts of 10 µg were found to be sufficient for such
chemists usually required all the spectroscopic data (UV,
spectra.
IR, NMR, and MS) to supplement microanalysis and gen-
eral physical characteristics such as refractive index, den-
sity, and melting and boiling points in order to elucidate
B. Fourier Transform Infrared Spectroscopy
the structure of the organic compound. Furthermore, it
was found that at least 50 mg of the material was needed Fourier Transform IR spectroscopy is widely used because
in order to determine the structure. It is now possible with of its rapidity of providing high-resolution spectra with
the modern available techniques to determine structures samples in the nanogram range. A complete spectrum for
with less than a milligram of material. a sample of 10−9 g can be obtained in less than 1 sec.
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
FTIR permits rapid quantitative characterization of solids, tion in a flow cell is very difficult. For normal-phase and
liquids, and gases. reverse-phase HPLC, continuous elimination of the sol-
Kinetic processes can be monitored by a technique vent is usually required before identifiable IR spectra can
known as time-resolved spectroscopy which involves be measured; no commercial instruments are yet available.
FTIR. This method has been applied to analysis of com- For at least one of these systems, a detection limit of 100
plex materials such as polymer film stretching which can ng has been reported for reverse-phase separations with a
be carried out in milliseconds and chemical transforma- water–methanol mobile phase.
tions involving, for example, coal pyrolysis; it also permits In conventional HPLC–FTIR, the interferograms are
on line analysis of products subject to chromatographic continuously recorded and stored during the analysis.
separation methods such as GC and LC. During the past Next, the absorption bands due to the solvent are sub-
five years GC–IR and GC–FTIR involved separating of tracted from the solution spectra. In order to facilitate these
mixtures and analysis of the individual compounds by IR measurements, the volume of the flow cell must be of the
spectroscopy. The sensitivity limitation of IR detectors order of 1 µL, implying that even with the packings of
with respect to GC and the time difference between the 3–5 µm diameter, the evolution volumes will be signif-
elution of a GC peak (measured in seconds) and the time icantly higher than the cell volume. Thus, only a small
scan were two of the problems encountered. GC–FTIR fraction (less than 1%) of each separated component will
allows an IR spectrum taken from a 5-µg GC peak of be in the cell when the measurements are made. Of the sev-
isobutylmethacrylate by repeatedly scanning with spectral eral interfaces proposed for solvent elimination, the most
accumulation and enhancement (Fig. 8). FTIR measure- promising one involves an initial concentration step in the
ments may be carried out by one of the following tech- concentrator tube using N2 gas and above ambient tem-
niques: (a) KBr pellets, (b) photoacoustic, and (c) diffuse perature, followed by the deposition of the concentrated
reflectance methods. solution on the KCl powder. Further elimination of the
The rapid-scanning property of FTIR spectrometers is solvent is achieved with a stream of air.
having its greatest impact in the field of GC–FTIR. FTIR
is now rivaling mass spectrometry for the identification 1. Diffuse Reflectance Fourier Transform
of unknowns eluting from gas chromatographs and in one Spectrometry (DRIFTS)
study was shown to identify more peaks than MS. It is
clear that GC–FTIR will be of importance for the analysis Diffuse reflectance Fourier transform spectrometry
of environmental samples. The interface between FTIR (DRIFTS) is a multifaceted technique for studying solid
and HPLC is still at a premature stage, and most of the samples. Diffuse reflectance has been known for some
HPLC–FTIR results reported to date have involved the use time as a solid sampling method. When infrared radia-
of size-exclusion chromatography with chlorinated sol- tion falls onto a sample surface, one of the following pro-
vents, since these have good transmission over much of cesses can occur. (a) The radiation can be absorbed; (b)
the infrared. The maximum concentration of most peaks it can be reflected from the sample (specular reflectance);
eluting from either normal-phase or reverse-phase HPLC or (c) it can penetrate the sample before being scattered
columns rarely exceeds 10 ppm, so that on-line detec- (Fig. 9). This latter effect is known as diffuse reflectance,
and it is this radiation that the technique of DRIFTS mea-
sures. The technique is applicable to most solids, involves
very little sample preparation, and avoids the unnecessary
complications of absorptions due to solvents or mulling
agents. Its greatest advantage by far, however, is its ability
to handle a wide range of samples, including conventional
FIGURE 8 IR spectrum of a small GC peak (5 µg) of isobutyl- FIGURE 9 The difference between specular reflectance and dif-
methacrylate using spectral accumulation (FTIR/GC cell). fuse reflectance from a solid surface.
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
FIGURE 10 Infrared spectrum of heroin by diffuse reflectance spectroscopy as a mixture in potassium bromide using
approximately 1 mg of heroin.
powders and crystals, solids with rough surfaces, gem- time-dependent IR signals to yield a spectrum defined by
stones, minerals, plastics, and fibers. The best results are two independent wave numbers. By spreading IR peaks
achieved with the sample in powdered form. The qual- over the second dimension, a complex spectrum consist-
ity of the spectra can be enormously enhanced by mixing ing of overlapped peaks can be substantially simplified,
the sample with potassium bromide in the ratio of 1:10 for and spectral resolution is enhanced. Peaks located on a
organic materials. 2D spectral plane provide information on connectivity and
Many pharmaceuticals can be studied successfully us- interactions among functional groups associated with the
ing this technique, for example 1 mg of heroin in potassium IR bands. The technique was recently developed by Isao
bromide (Fig. 10). It is possible to monitor the concentra- Noda and his co-workers, and he used it to demonstrate the
tion of heroin in a potassium bromide sample once a cali- known incompatibility of polyethylene and polystyrene by
bration graph has been prepared. The DRIFTS technique showing that the two resins are segregated at the molecular
can be applied to quantitative analysis. level (Fig. 11).
Another application of DRIFTS is the identification of
fibers. There are many types of fibers, both synthetic and 3. Chromatographic Optimization
natural, which can be identified by IR spectroscopy but of Combined GC–FTIR–MS
not so easily by other techniques.
By using a micro sampling cup it is possible to pro- Recent rapid developments of analytical instrumentation
duce diffuse reflectance spectra with a few milligrams have made it possible to obtain simultaneously GC–MS
of the sample, such as less than 1 mg of heroin. The and GC–FTIR information on a sample from one injec-
DRIFTS technique can thus be applied to the screening tion and one GC separation. Combining GC–MS and GC–
of illicit street drugs, such as a mixture of cocaine and FTIR into one system provides a higher confidence quali-
amphetamine. tative result and saves time, laboratory space, and money.
It should be realized that mass spectrometry and infrared
2. Two-Dimensional Infrared Spectroscopy spectrometry are complementary.
(2D IR) In terms of relative practical sensitivity, the Mass Spec-
trum Detector (MSD) in the scanning mode is approxi-
Two-dimensional infrared (2D IR) spectroscopy is a recent mately 10 times better than the Infrared Detector (IRD).
novel technique based on time-resolved IR spectroscopy. Thus, a mass spectrum on a level of 1 ng requires 10 ng
The spreading out of IR spectra over two dimensions sim- for an IR spectrum if the compound is a strong absorber.
plifies complex spectra with overlapping peaks. Corre-
lation of such peaks may identify or eliminate specific
4. Matrix Isolation Fourier Transform
intermolecular and intramolecular interactions. Such cor-
Spectrometry (MI-FTIR)
relations can assist in assigning peaks to particular vibra-
tional modes. In 2D IR, a system is excited by an exter- Matrix isolation techniques are used in the spectroscopic
nal perturbation, which induces a dynamic fluctuation of detection of (a) chemical species that are too reactive,
the IR spectrum. A correlation analysis is applied to the too unstable, or too transient to be investigated under
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
from a heated, linear capillary nozzle toward the surface tube that is electrically biased with respect to an entrance
of a slowly rotating germanium disc. The microdrops are orifice of a MS. The electrode shapes, geometries, and
constrained by a heated gas sheath that also rapidly va- potentials are important for penetrating a stable spray of
porizes the solvent, leaving a particle stream of solute droplets; however, many different experimental configu-
material that collects on the disc surface. Instrumentation rations can be used. Two widely held views of the mech-
based on this technology is used for a wide variety of ap- anism for ESI are that solvent-free molecular ions are
plications in polymer, pharmaceutical, and environmental formed either after a series of droplet fissions, followed
laboratories. by solvent evaporation from the ion, or by ion evaporation
The recent development of electrospray ionization from the droplet surface.
(ESI) and matrix-assisted laser desorption-ionization
(MALDI) sources for creation of macromolecular ions
C. Fourier Transform Raman
has brought about a renaissance in the field of mass spec-
Spectroscopy (FT-RS)
trometry (MS).
It is now common to determine high molecular weights Raman spectroscopy (RS) has been known for more than
such as proteins to within a few Daltons. 60 years. It is during the past decade that the organic
MS-based methods for sequencing are proliferating chemist has included RS as an additional tool for struc-
and replacing or complementing traditional biochemical ture determination. The Raman spectrum provides com-
methods for many applications. Mounting evidence sug- plementary information to the IR spectrum in that both
gests that some conformational properties of biomolecules are vibrational spectra. Often the weakest signals in the IR
in solution are preserved during the ionization process and spectrum are the strongest signals in the Raman spectra.
persist over the transience time that ions exist in mass spec- Modern Raman spectrometers operate with a monochro-
trometers. matic coherent laser light source. Sampling handling is
In the past 5 years a new field involving structural much easier in RS than in IR spectroscopy. The sample
studies of biomolecular ions has emerged. Studies of can be dissolved in a number of solvents since their Raman
biomolecular ions in the gas phase provide informa- absorptions are more limited than in the IR. Solid samples
tion about the nature of conformation in the absence of can also be examined as powders or crystals. Liquids can
solvent: be placed in small quartz or Pyrex glass capillaries, and the
The structural studies of protein ions have been grouped scattered radiation is observed through the open end of the
into two categories: chemical probes, in which structural capillary. Samples can also be examined without removing
characteristics are deduced by monitoring the products them from their bottles. If several spectroscopic analyses
of ion-molecule reactions, and physical probes such as are to be performed on a small amount of an organic sam-
measurements of dissociation patterns and measurements ple, then the RS is obtained first because it is totally nonde-
of cross sections. structive and does not require dilution in solvents, as in the
An important factor in folding of proteins involves case of UV or NMR. Certain group frequencies such as pri-
solvent–molecule interactions. Proteins appear to form mary amines, alkynes, and nitriles are strongly absorbed
physiologically relevant conformations in the presence of in the RS but are difficult to observe in the IR spectrum.
a minimum hydration shell. Raman spectroscopy has a number of experimental advan-
Recent studies have considered structural changes that tages over IR: (a) Window problems hardly exist, if visi-
occur when proteins are lyophilized by examining anhy- ble or near IR lasers are used as sources; glass is a robust
drous powder and thin films. Techniques such as FTIR window material with excellent transmission; (b) since
spectroscopy, Raman spectroscopy, and H-D exchange transmission through the sample is not involved, very lit-
NMR methods have shown structural changes upon re- tle preparation is required in RS; (c) water, which has an
moval of solvent. extremely strong IR spectrum, produces only a weak RS
When the proteins are heated, they denature, making and is the solvent of choice in biochemical and aqueous
them less volatile; the molecules decompose rather than studies.
vaporize. ESI has offered a simple, efficient means of Raman spectra are often plagued by the problem of flu-
producing gaseous, anhydrous biomolecules for study by orescence when conventional shorter (visible) wavelength
MS. Thus, anhydrous proteins also fold and unfold, and is used. It was estimated that from 1966 until 1989 less
new conformations are established in a grossly simpli- than 20% of the samples studied gave suitable spectra
fied environment whose only intramolecular constraints with visible laser Raman dispersive spectrometers. Even
are relevant. so, considerable time is wasted in obtaining a reason-
Charged droplets containing molecular ions are initially able Raman spectrum. Since 1987 Near Infrared Fourier
formed by pumping a solution through a narrow capillary Transform Raman Spectroscopy (NIR-FTRS) has been
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
available. This technique has the potential to be more ver- As technology advances, it provides new opportuni-
satile than conventional laser Raman spectroscopy. The ties. The gas-filled optical system technology opens up
technique in NIR-FTRS is to excite the spectrum with near the VUV for spectrochiral analysis. The terminology “far
IR laser, pass the scattered light through filters to remove ultraviolet” might be more appropriate since it includes
the reflected and Rayleigh scattered excitation radiation, access to this wavelength region by gas-filled optical
and then to process the Raman scatter on an FTIR instru- techniques.
ment. Being infrared excited, the spectra were essentially
clear of fluorescence, while color due to absorption was
E. Mass Spectrometry (MS)
far less of a problem in near infrared than with the conven-
tional techniques. For example, rubrene, which is deeply Mass spectrometry provides a means for studying sam-
colored and intensely fluorescent, produced a well-defined ples at the molecular level. Although MS is basically a
Raman spectrum at 1064 nm. Recently FTRS spectra of al- structure-identifying tool, it is not as specific for the de-
kaloids such as heroin, codeine, and morphine have been tection and determination of functional groups (IR) or rel-
obtained, but because of the complexities of their spec- ative positions protons or 13 C nuclei (NMR) spectroscopy.
tra they have not yet been fully analyzed. It is obvious Mass spectrometry, like UV, IR, and NMR spectroscopy,
that within the next few years group frequencies char- works best with pure samples. Structural elucidation of
acteristic of structural features in complex organic and mixtures is not possible without some means of separat-
bioorganic compounds will be determined. It appears that ing each constituent in the mixture as in GC–MS.
NIR-FTRS is the choice method in obtaining spectra of Besides being a useful structure-elucidating tool, MS
biochemical substances such as enzymes, proteins, and can be applied to detect very low levels of specific com-
polypeptides because such molecules can be studied in pounds and elements. Accurate determination of masses
aqueous media. Interpretation of the complex nature of can also be determined. An important advantage of the MS
these biomolecular spectra will also take time. Commer- technique is its high sensitivity and accuracy. MS is able
cial instruments offering mid-infrared (400–4000 cm−1 ), to provide more specific information per given amount of
near infrared (4000–10,000 cm−1 ), and Raman spectra all material than any other analytical technique. Furthermore,
on the same spectrometer will be available in the near even with minute amounts of material (10−6 –10−9 g), this
future. information can be provided in a reproducible and accu-
rate manner. In the 1940s MS was an important analyt-
ical technique for characterizing complex fuel mixtures.
D. Vacuumless Spectrochemistry
About 25 years ago interest was centered on a system-
in the Vacuum Ultraviolet
atic study of ionic fragmentation mechanisms in order
The emission wavelength region below 200 nm has tradi- to provide a set of rules that could be applied to eluci-
tionally been called the vacuum ultraviolet (VUV). This date structures of organic compounds. In the late 1960s
region contains the best spectral lines for many indus- mass spectrometers were coupled to gas chromatographs.
trially important elements, such as carbon, phosphorus, This provided such dramatic improvements in selectiv-
sulfur, and boron, in steel and cast iron by spark excita- ity and sensitivity that MS has since become one of the
tion, and the halogens for the inductively coupled plasma most generally useful analytical techniques for identifica-
(ICP) analysis of oils and fuels. tion and quantitation of organic substances at ultratrace
Wavelengths below 200 nm are absorbed by some com- levels. Chemical ionization mass spectrometry (CIMS)
ponents of air, primarily oxygen and water vapor. There- uses reagent ions rather than electrons to ionize a sample.
fore, the usual approach to spectrochemistry in this wave- Field desorption and rapid heating techniques are extend-
length region is to evacuate the optical system. ing mass spectrometry to high-molecular-weight, low-
The use of a nitrogen gas-filled system provides im- volatility compounds. During the past decade mass spec-
portant advantages for both the industrial and academic trometry has undergone a number of significant changes
spectroscopist. This technology opens up the low wave- that have far-reaching contributions especially in biol-
lengths of the emission spectrum which contains many ogy and medicine. The mass range of mass spectrometers
important spectral lines for elements of interest for practi- has been extended by approximately an order of magni-
cal spectrochemistry. It should be emphasized that none of tude in the past decade. Certain types of mass analyz-
the wavelengths below 140 run have been achieved with ers have been used to reach higher masses (150,000 in
a conventional vacuum spectrometer. some quadrupole experiments). Commercial instruments
One of the economic advantages of ICP analysis of are now available with mass ranges of 7500 compared
halogens in oils, fuels, and other matrices is that it is faster with 1000 about ten years ago. Another improvement is
than other methods such as X-ray fluorescence (XRF). desorption ionization [including fast atom bombardment
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
(FAB), secondary ion mass spectrometry (SIMS) and fis- analysis of mixtures. The two principal advantages of
sion fragment methods] which allows ionic, nonvolatile this system can be illustrated by considering a complex
compounds to be examined by MS. A further develop- coal liquid mixture. The signal due to a dioxin spike is
ment is the integration of separation and analysis tech- lost in the chemical noise from the other constituents
niques represented by tandem mass spectrometry (MS– and single-stage mass spectrometry is not capable of
MS) and the much improved liquid chromatography-mass analyzing for it. MS–MS filters against chemical noise
spectrometry (LC–MS) interfaces. The latter capability and allows a high-quality spectrum of the dioxin to be
is already proving its effect with respect to GC–MS but recorded. In addition to improving detection limits in this
without the limitation to volatile, low-mass compounds. way, tandem mass spectrometry provides alternative scan
MS–MS is a two-dimensional form of spectrometry which modes which can be employed to search the data bank for
often improves signal-to-noise ratios as well as provid- particular information. For example, chlorinated dioxins
ing entirely new capabilities, such as that of scanning a are characterized by the loss of COCI, so a scan for the
mixture for all constituents having particular structural reaction reveals all dioxins present in the mixture. The
subunits. GC–MS–MS system can decrease GC–MS detection lim-
The GC–MS combination has had considerable impact its by an order of magnitude. It has been reported that the
on biological and environmental research, providing a spe- drug, isosorbide-5-mononitrate, a coronary vasodilator, is
cific means of characterizing constituents of mixtures and metabolised to the glucuronide, which can be determined
having sufficient sensitivity and quantitative accuracy to in urine by a simple MS–MS procedure to 0.1 ng/mL.
trace constituents. The recent discovery in 1977 of the neu- The improved detection limit in the MS–MS experiment
ral excitotoxin, quinolinic acid, in the mammalian brain is the direct result of minimizing interferences. Both
at the level of nanomoles per gram was achieved by GC– the sensitivity and the speed of analysis with MS–MS
MS of the volatile hexafluoroisopropanol diester deriva- can be illustrated with the administration of the drug
tive with electron impact (EI) ionization. Quantitation was tetrahydrocannabinol in doses of 0.1 mg/kg for eight days
based on standard addition and single-ion monitoring. down to 10−11 g/ml by using a combination of GC–MS
The GC–MS technique was applied in the discovery of with simple MS–MS to avoid extensive sample cleanup.
19-hydroxylated E prostaglandins by first protecting the Sensitivities in parts per trillion have been reported in
unstable β-ketol system by oximation. This discovery, in MS–MS studies on animal tissue. High-resolution mass
1974, implied that the previously identified prostaglandins spectrometry and MS–MS have been used to achieve
could be artifacts. absolute detection limits of less than 10−12 g for tetra-
Challenging structural problems. including protein chlorodibenzodioxin and a GC–MS–MS combination has
structure determinations, can often be solved by a com- produced spectra with 250 fg samples (<20 parts per tril-
bination of techniques. Two mass spectrometric methods lion) at the rate of 30 samples per day. In terms of sample
were used to deduce the structure of the 112-amino-acid throughput, the determination of trichlorophenol in serum
antitumor protein macromycin derived from streptomyces at concentrations as low as 1 ppb and a rate of 90 samples
culture. Partial acid digestion gave a mixture of di- to per hour was reported. Characterization of the foodstuff
hexapeptides, which was derivatized and analyzed by GC– contamiant (and chemical warfare analog) vomitoxin at
MS with EI. 25 pg in wheat at a rate of 10 minutes per samples was also
Tandem mass spectrometry has been applied in the dis- reported.
covery and confirmation of the structure of the metabo- A different approach to the characterization of mixtures
lites of the drugs primidone, cinromide, and phenytoin in of nonvolatile compounds is LC–MS. The first practical
plasma and urine extracts; analyses were completed in less LC–MS interface was based on complete removal of the
than an hour, using concentrations of 1 to 50 µg/mL. The solvent and temporary storage of the solute during trans-
study is predicted on the speed and flexibility of MS–MS port by a moving belt or wire into the ion source. In the
scans made with a triple quadrupole instrument, and on source the sample is either thermally desorbed and ionized
the realization that metabolites often retain a large portion by electron or chemical ionization or the belt is bombarded
of the parent drug structure. by an energetic beam to create secondary ions. The large
Analysis of complex mixtures has often involved pumping capacities of chemical and atmospheric ioniza-
time-consuming procedures such as extraction, centrifu- tion sources make it possible to work at flow rates consis-
gation, and chromatography. MS–MS and LC–MS are tent with normal column operation, for example, 2 mL of
recent developments that reduce such time-consuming aqueous mobile phase per minute. The solvent itself acts
methods. By linking two mass spectrometers in tandem as the reagent gas in these experiments. An alternative,
it is possible to employ the first as a separator and the the thermospray procedure does not use any externalion-
second as an analyzer and, hence, to perform direct ization technique. An aerosol generated in the interface is
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
A routine sample on a 60-MHz instrument would re- in the works, including a 900-MHz and gigahertz instru-
quire about 5–50 mg of the sample in about 0.4 mL sol- ments being built in the United Kingdom, United States,
vent. A microtube consisting of a thick-wall capillary al- Germany, and Japan.
lows spectra to be obtained on less than 1 mg of sample. The new instruments will find cures for diseases, resolve
The use of FTNMR makes it possible to obtain spectra the molecular structures of protein, amorphous polymers,
on amounts as small as 5 µg. The ideal solvent should ceramics, and biomaterials, and elucidate the dynamics of
contain no protons in its structure, be low-boiling point, biological systems.
nonpolar, and inert. Carbon tetrachloride is often used in High-resolution NMR may serve as an ideal compli-
non-FTNMR spectroscopy, depending if the sample is suf- ment to X-ray crystallography because it provides infor-
ficiently soluble in it. The most widely used solvent is mation on molecular dynamics that crystals, being static,
deuterated chloroform (CDCl3 ). cannot. NMR enables observation of changes in the dy-
namics of a protein when it binds to DNA, or details the
relationship between catalytic function and motion. This
I. Nuclear Magnetic Resonance (NMR)
information would be useful in drug design.
New health and materials sciences for societal needs will Stoking the interest in faster access to protein structure
both fuel the drive to make high-field NMR spectroscopy are the rapid advances in genome research that are spewing
a more accessible research tool. Currently, only a hand- out genes with no known function.
ful of the most powerful 800-MHz NMR spectrometers A Swiss NMR technique makes it possible to obtain
and no more than a dozen 750-MHz instruments are in high-resolution solution NMR spectra of macromolecules
use in the United States. More powerful instruments are and supramolecular structures with masses up to several
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
hundred thousand Daltons—many times larger than those broadening with increasing molecular size, permitting
accessible by conventional NMR techniques. NMR analysis of molecules far beyond 100 KDa in size.
The technique could prove useful for
13
r Chemical shift mapping (NMR studies of interactions 1. C NMR Spectroscopy
of protein–protein or protein nucleic acid interfaces in The reason that 13 C NMR was developed much later than
supramolecular assemblies) the proton NMR was due to the low natural abundance
r Acquiring structure–activity relationship data on of the 13 C nucleus (1.108%) as compared to 1 H NMR
molecular interactions with large receptor molecules (100%) and the low gyromagnetic ratio. These two prob-
(SAR by NMR) lems were overcome during the past 15 years due to the
r Obtaining high-resolution NMR spectra of membrane arrival of Fourier transform where the accumulation of
proteins reconstituted in micelles. scans over a period of time overcomes the problem of the
r determining NMR solution structures of large low abundance of 13 C. The advantages of 13 C NMR com-
macromolecules. pared to 1 H NMR is that better resolution is obtained since
the 13 C absorptions for most of the organic molecules are
A technique called transverse relaxation-optimized spread over 200 ppm instead of 10 ppm. Secondly car-
spectroscopy (TROSY) now greatly reduces NMR line bons bearing no protons are revealed and finally a count
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
of the number of protons attached to each carbon results it is protons from 1 H NMR. Finally protons attached to
from comparison of the broad-band decoupled 13 C NMR heteroatoms are not visible. For the time being 13 C NMR
spectrum with the off resonance 13 C spectrum. Therefore, spectroscopy is a method that compliments other spec-
the number of methyl (quartet), methylene (triplet), me- troscopic techniques necessary to elucidate structures of
thinyl (doublet), and quarternary (singlet) carbons in a organic molecules. One commercial application of 13 C
fairly complex molecule such as in the natural sweetener, NMR is the determination of the purity of certain types
stevioside, C38 H60 O18 (Fig. 14) is easily determined by of organic compounds. For example, whether a reduction,
13
C rather than by 1 H NMR. oxidation, or esterification processes have gone to comple-
On the other hand there are a number of disadvantages in tion can easily and also quite rapidly (if sufficient material
13
C NMR. A larger sample size (up to 100 mg) and longer is available) be determined.
sampling time (up to several days for small samples) is re- Recent advances of even more sophisticated and more
quired. If 100 mg of a sample is available and that sample powerful NMR spectrometers have enabled high resolu-
is of high solubility, the time required may be only a few tion 13 C NMR spectra where signals as low as 3 ppb can be
minutes. A good spectrum can be obtained on as little as a distinguished at moderate field strength and without res-
1 mg sample when several days are available for scanning olution enhancement. This is evident in the enantiomeric
the sample provided the sample is completely stable in mixture of the alkane farnasane
the deuterated solvent. Another disadvantage is that due [CH3 CH (CH2 )3 CH (CH2 )3 CH CH2 CH3 ]
to variations in relaxation times and nuclear Overhauser | | |
effects (NOE), the areas of absorption differ for individ- CH3 CH3 CH3
ual carbons (up to a factor of about 10). Thus, it is not which gave two clear signals for all the carbon atoms ex-
as easy to tell relative numbers of carbons from 13 C as cept for one and with a minimum separation of 2.9 ppb.
FIGURE 14 13 C NMR spectrum of the natural sweetener stevioside (C38 H60 O18 ).
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
Enantiomeric separation and spectral simplification of must be found between the optional flow rate and the de-
overlapped resonances in high resolution 1 H NMR spectra tector volume. Moreover, the HPLC–FTNMR operation
can be obtained with optically active paramagnetic shift is limited to certain types of solvents. Since most com-
reagents such as Eu(tfc)3 . Detection of optical isomers monly used HPLC solvents contain hydrogen, the mem-
(±) 2,2-dimethyl-1-phenyl-1-propanol and (±) carvone, ory of the minicomputer used for data acquisition will be
as well as optically active terpenes, can rapidly be deter- saturated with the solvent signal, thus overshadowing the
mined by the use of optically active shift reagents. weak sample signal. This has necessitated the use of spe-
cial suppression techniques for solvent signals, such as
13 filtering, selective presaturation, selective excitation, and
2. C-CP/MAS NMR
spin-echo techniques, or increasing the dynamic range of
It has been observed that line broadening of NMR spec- the minicomputer. It is possible to use deuterated solvents
tra of solid powders arises jointly from 13 C–1 H static as the HPLC solvents but this involves considerable ex-
dipolar interactions and from chemical shift anisotropy. pense. During the past decade FTNMR has been applied
These effects are eliminated or greatly reduced by dipo- to solid states especially to elucidate microstructures of
lar decoupling (dephasing) of protons and by spinning polymers. The term microstructure pertains to the level of
the sample at an angle of 54.7◦ (magic angle) to the ap- isomerism of a polymer chain. For example oxidized sites
plied field. For solids a greater rate of spectral accumu- in polymers (carboxylic acids, ketones or alcohols) can be
lation is required, this method is called cross polarization detected and quantified at levels as low as two sites per
(CP). thousand residues from the 13 C NMR spectrum. Although
High resolution 13 C NMR which utilizes both cross po- carbon is ubiquitous in synthetic polymers, other nuclei
larization with dipolar dephasing and magic angle spin- have been used to advantage for polymer structure de-
ning (MAS) is called 13 C-CP/ MAS NMR. This tech- termination. For example, 15 N NMR for nylon polymers,
31
nique directly measures the organic carbon distribution P NMR for poly(phosphazenes), and 19 F NMR for fluo-
in terms of aliphatic and aromatic carbon structures in oil ropolymers. For some time 1 H FTNMR has been regarded
shales, coals, humic substances, melanoidins, and other as less informative for synthetic polymers than the carbon
polymeric substances. resonance; it is now the focus of renewed interest.
generated in which the F2 axis corresponds to 13 C chem- cynarin (isolated from artichokes), allowing the distinc-
ical shifts (δc ) or other nonproton nuclei and the F1 , axis tion between two possible structures to be made. The
which corresponds to the proton 1 H shifts (δH ). The di- complex structures of griseochelin and grisorixin metabo-
agonal relationship of the signals in such 2D-NMR spec- lites, which are antibiotics obtained from strains of Strep-
tra is a direct consequence of the parallel between car- tomyces, were solved by COSY techniques. HETCOR
bon chemical shifts and attached hydrogen shifts. The experiments appear for the most part to have been used
closely analogous Homonuclear shift correlation (HOM- as a simple tool for assignment in the area of natural
COR) or COSY (COrrelated Spectroscop Y) maps out all products. In the studies of bile acids, isomugineic acid
homonuclear couplings in a single spectrum. In a typical (a metabolite derived from barley), the heteronuclear cor-
COSY spectrum of stevioside (Fig. 15), both the F1 and F2 relation was used to transfer proton assignments that had
axes correspond to 1 H chemical shift scales. The normal been made by other means onto the 13 C spectrum. HET-
1 COR techniques have also been applied in long-range
H NMR spectrum appears along the diagonal, while all
off-diagonal signals corresponds to 1 H-1 H homonuclear experiments to obtain structural information; for exam-
couplings. ple, gracilin B (a bisnor-terpene isolated from the sponge
The HOMCOR technique was originally used by Spongionella gracilis), ekeberginine (an alkaloid extracted
Jenner in 1971, when he obtained a COSY spectrum of β- from Ekebergia senegalensis), and didehydroryanodine
chloroacrylic acid. It was only a decade ago that 2D-NMR (an insecticide obtained from Ryania speciosia) were all
became a routine technique. determined with the aid of long-range proton-carbon cor-
relations using the HETCOR method.
a. Applications of 2D-NMR spectroscopy. Ade- Perhaps one of the recent exciting and fruitful areas
quate COSY spectra were obtained from only 90 µg of of FTNMR research is in biochemistry. It is only been
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
a decade since Moon and Richards at California Insti- serve only those protons that are attached to the labeled
tute of Technology and Hoult at Oxford demonstrated that carbons.
FT–31 P NMR spectroscopy is an effective probe of the in
vivo generation and utilization of phosphate-bond energy.
FTNMR spectroscopy has since been applied to the study J. 1 H NMR “Polarimetry”
of many diverse systems, progressing from intact cells to
Optical purity of enantiomers is of great importance in
perfused organs to animals and man. In vivo FTNMR is
the chemistry of natural products, especially in the fields
noninvasive, and thus event-induced changes in the con-
of flavors, fragrances, and pheromones. Since odor recep-
centration of metabolites can be followed over a time pe-
tors in mammals and in insects can discriminate between
riod. 31 P is by far the most widely used nucleus for bio-
enantiomers, it is important to know the enantiomeric
chemical FTNMR determinations, as the key metabolite,
composition of chiral molecules that impinge on these
ATP, adenosine diphosphate (ADP), creatine phosphate,
receptors.
and sugar phosphates, for example, all contain phospho-
Determination of the enantiomeric composition of nat-
rus. The NMR sensitivity of phosphorus is high (100%)
ural semiochemicals by 1 H NMR spectroscopy using a
and offers selectivity not obtained with nuclei such as 13 C
chiral lanthanide shift reagent is a rapid, accurate, and ion-
or 1 H because it is not ubiquitous, but rather occurs pri-
destructive technique. It is also a quantitative method for
marily in cellular metabolites. Metabolic pathways are
detection of adulteration of essential oils and oleoresins
now being followed by 31 P FTNMR. Techniques using
by synthetic racemic chemicals.
whole animals (or humans) require that the NMR mea-
surements be localized to a particular organ or to a certain
portion of the body. More recently is the application of
K. Supersonic Jet Spectroscopy
FTNMR in the study of the metabolism of 13 C-labeled
drugs. The fate of the labeled carbons can be followed as Capillary column gas chromatography (GC) is the high-
a function of time using 13 C NMR spectroscopy. Also est resolution separation technique, but it is ultimately
new pulse sequences have now been developed to ob- limited by the volatility of the molecules in the sample.
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
HPLC extends the range of species which can be carried XIII. X-RAY CRYSTALLOGRAPHY
in the analyte stream, but it suffers from relatively low
column efficiency. Chromatography by itself provides ex- The purpose of X-ray analysis of crystal structures is to
tremely limited qualitative information. The most produc- provide information on the positions of the individual
tive approach is the combination of an instrument having atoms of a molecule, their interatomic distances, bond
high separating power and one having high discriminating angles, and other features of molecular geometry such
power. One measurement method with high discriminat- as the planarity of a particular group of atoms, the
ing power is supersonic jet (SJ) spectroscopy. Kantrowitz angles between the planes, and torsion angles around the
and Grey originally proposed supersonic expansions as a bonds. The resulting three-dimensional representation
means of increasing the intensity of molecular beams in of the atomic contents, rapidly determined by modern
the early fifties, but it was in the mid-seventies that the SJ computerized techniques, of the crystal establishes the
techniques were popularized for spectroscopic studies. It complete conformational structure and geometrical
was only in the eighties that the analytical application to details hitherto unknown.
the study of complex samples were recorded. The combi- This information is of primary interest to chemists and
nation of SJ spectroscopy with chromatography promises biochemists who are concerned with the relation of struc-
to be a powerful method for detailed analysis. Among tural features to chemical properties. It was only X-ray
all the line-narrowing spectroscopic techniques, SJ spec- analysis that finally determined the double helix of two
troscopy is the one which can be most directly coupled to DNA molecules held together by hydrogen bonds that
chromatography. Advantages of the SJ approach include verified Watson and Crick’s proposed model for DNA.
elimination of all matrix interferences, such as spectral Thus, X-ray analysis of organic compounds provides an
shifts, site inhomogeneous broadening, phono-coupling, unambiguous complete three-dimensional picture of the
all other extrinsic fluorescence quenching, and Raman molecule, whereas other chemical and physical methods
scatter and background from the matrx. The need to iden- of structure analysis involved different aspects which col-
tify a suitable matrix is avoided. No complex cryogenic lectively one can deduce the number and nature of the
equipment is required. A disadvantage of SJ is the very atoms bonded to each atom present (the topology of the
short time which sample molecules spend in the obser- molecule) or, for relatively simple molecules, provide
vation volume, limiting the amount of signal averaging some quantitative information from which geometrical
which can be done if only a small quantity of a sample is details can be derived. There are instances whereby only
available or if it is to be detected in the narrow time span X-ray analysis will decide absolutely the spatial arrange-
during which it elutes as a chromatographic band. Also, ment of groups of atoms in simple organic molecules. It
although the spectroscopic detection method may be non- was X-ray analysis that confirmed that the cyclic sulphite
destructive itself, the sample is usually lost. Among all (Fig. 17) had a boat–chair conformation with the S O
the line-narrowing spectroscopic techniques, supersonic group trans to the cyclopropane ring and in an equatorial
jet spectroscopy is the one which can be most easily used position.
for real-time detection in chromatography. The apparent disadvantages of X-ray analysis is that
The equipment required for SJ detection is less expen- it is time consuming even with the use of automated
sive than that for either MS or MI. Furthermore, the SJ ap- computerized techniques. Further-more, X-ray analysis
proach allows for adding on downstream detectors, such can only be carried out on pure crystalline materials.
as especially MS; the MS can be used in 100% and even The crystals must be a certain size and regularity in
weakly absorbing molecules should be detected. order that a good X-ray structure of the compound is
obtained.
1. Gas Chromatography–Supersonic Jet
Spectroscopy (GC–SJ Spectroscopy)
XIV. CHEMICAL ANALYSIS
It was in 1982 that the analytical potential of SJ spec-
troscopy was used to selectively detect the components The requirement for information on chemical composi-
of a mixture of naphthalene and α- and β-methyl naph- tion extends into most sectors of industry. The chemical
thalenes. Trace impurities of fluorene in biphenyl at less measurement analytical market is extremely diverse.
than 1 ppm could be detected. Once the analytical advan- Chemical analysis is a key process in drug discovery, in
tages of SJ were realized, it would seem that coupling it assessing the quality of feedstock materials and end prod-
to GC would be a narrow extension. A 100-ng detection ucts, in monitoring chemical processes, in demonstrating
has been recently achieved for perylene, and it has been regulatory compliance, in disease diagnostics, and in en-
estimated that 1-pg detection limit can be attained. vironmental conservation.
P1: LLL/LOS/GJM P2: GLM Final Pages
Encyclopedia of Physical Science and Technology EN0011A-541 July 25, 2001 17:27
Cooper, J. W. (1980). “Spectroscopic Techniques for Organic Chemists,” Heftman, E. (1983). “Chromatography,” Journal of Chromatography Li-
Wiley, New York. brary, Vol. 22B, Elsevier.
Craston and Cowen, S. (1998). “The Laboratory on a chip,” Science Ikan, R. (1982). “Chromatography in Organic Microanalysis,” Academic
Progress 81, 225–244. Press, New York.
Duddeck, H., and Dietrich, W. (1989). “Structure Elucidation by Modern Irwin, W. J. (1982). Analytical Pyrolysis, “Chromatographic Science
NMR,” Springer, New York. Series,” Vol. 22. Marcel Dekker. New York.
Firth, S. (1988). “Applications of diffuse reflectance infrared spectrom- Levy, G. C., Lichter, R. L., and Nelsen, G. L. (1980). “Carbon-13 Nuclear
etry,” Int. Lab. 18(8), 24–30. Magnetic Resonance Spectroscopy,” Wiley, New York.
Freemantle, M. (1999). “Downsizing chemistry,” Chem. Eng. News Feb. McLafferty, F. W. (1983). “Tandem Mass Spectrometry,” Wiley, New
22, 27–36. York.
Goates, R. G., and Sin, C. H. (1989). “Supersonic jet spectroscopy cou- Meuzelaar, H. L. C. (Ed.) (1990). “Computer Enhanced Analytical Spec-
pled to chromatography for very high resolution chemical analysis,” troscopy,” Plenum, New York.
Appl. Spectrosc. Rev. 25(2), 81–126. Pryde, A., and Gilbert, M. T. (1979). “Applications of High Performance
Granger, P., and Harris, R. K. (Eds.) (1991). “Multinuclear Magnetic Liquid Chromatography,” Chapman and Hall, London.
Resonance in Liquids and Solids—Chemical Applications,” Kluwer Schriemer, D. C., and Li, L. (1996). “Detection of high molecular weight
Academic, Dondrecht. narrow polydisperse polymers up to 1.5 million daltons by MALDI
Griffith, P. R., and Haseth, J. A. (1986). “Fourier Transform Infrared mass spectrometry,” Anal. Chem. 68, 2721–2725.
Spectroscopy,” Wiley, New York. Silverstein, R. M., Bassler, G. C., and Morrill, T. C. (1991).
Grob, R. L., and Kaiser, M. A. (1982). “Environmental Problem Solving “Spectrometric Identification of Organic Compounds,” Wiley, New
Using Gas and Liquid Chromatography,” Journal of Chromatography York.
Library, Vol. 21. Elsevier. Wilson, M. A. (1987). “NMR Techniques and Applications in
Gudzinowicz, B. J., and Gudzinowicz, M. J. (1979). “Analysis of Drugs Geochemistry and Soil Chemistry,” Pergamon, Elmsford, New
and Metabolites by Gas-Chromatography–Mass Spectrometry,” Chro- York.
matographic Science Series, Vol. 6, Marcel Dekker, New York. Wilson, N. K., and Childers, J. W. (1989). “Recent advances in the matrix
Hayes, J. M. (1987). “Analytical spectroscopy in supersonic expansions,” isolation infrared spectrometry of organic compounds,” Appl. Spec-
Chemical Rev. 87, 745–760. trosc. Rev. 25(1), 1–61.
P1: GTQ/GLE P2: GPJ Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN012C-562 July 26, 2001 15:30
Photoacoustic Spectroscopy
Konka Veeranjaneyulu
Roger M. Leblanc
University of Miami
I. Introduction
II. Photoacoustic Effect and Signal Generation
III. Advantages
IV. Instrumentation
V. Applications
1
P1: GTQ/GLE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN012C-562 July 26, 2001 15:30
2 Photoacoustic Spectroscopy
FIGURE 1 Various possible pathways of dissipation of incident electromagnetic radiation from a sample.
Ultraviolet(UV)/visible radiation is used to probe the elec- heat and light (fluorescence from singlet state and/or phos-
tronic transitions, while infrared and microwave radiations phorescence from triplet state). The excited molecules
are used to probe the molecular motions such as vibration transfer the energy to the neighboring molecules or release
and rotation. The UV and visible radiations are often used the energy by radiative and/or nonradiative de-excitation
to probe the energetics of a system. The probing light processes. The nonradiative de-excitation process results
incident upon the sample may be transmitted, absorbed, in heat emission, and the radiative process emits light at
or reflected by the sample. In order to probe the cellular longer wavelengths than the absorbed radiation (Fig. 1).
energetics of the system, the light must be absorbed by As mentioned, light emission is measured by fluorescence
the sample. The conventional methods such as absorption spectroscopy. The heat emission can be measured by pho-
spectroscopy or reflection spectroscopy measure the ab- toacoustic spectroscopy (PAS). Since every compound
sorbed and reflected light, respectively. In absorption spec- that absorbs light energy emits thermal energy, this tech-
troscopy, it is the transmitted light that is measured, and the nique is highly useful to probe directly the energetic and
absorbance is calculated by considering the incident light absorption characteristics of systems of interest. One par-
(Io) and transmitted light (I ). However, certain conditions ticular aspect of this technique is the probing of absorbed
such as nonhomogeneity and the light-scattering effect of light and not the transmitted light.
the samples eliminate the use of absorption spectroscopy Photoacoustic spectroscopy is essentially a technique
in a wide range of samples. Besides, the cellular biophys- that uses the ability of periodically modulated or pulsed
ical activity can be better probed by studying directly the light to produce sound upon absorption. The photoacoustic
absorbed light energy rather than the transmitted light en- (PA) effect is the result of the transformation of the ab-
ergy. Fluorescence spectroscopy has been used widely to sorbed modulated light energy into heat energy, which in
probe complex biological systems. As most molecules of turn generates sound. As a sample absorbs the modulated
biological interest have appreciable fluorescence quantum or pulsed light energy, it emits the heat periodically, which
yield, and the fluorescence can be measured from the sur- leads to the cycles of expansion and contraction causing
face of the sample, this technique has found wide usage the generation of sound waves. These sound waves are
and is considered the most sensitive tool. The fact that all detected by a sensitive microphone. There are also other
molecules are not fluorescent, and even some fluorescent means to detect the modulated heat emission, based on the
molecules exhibit low quantum yields, makes this tech- measurements of temperature changes. For example, in
nique limited to fluorescence high yielding systems. photochemical beam deflection spectroscopy, spatial and
Generally, when a sample absorbs light energy, a part of temporal refractive index changes caused by heat mod-
the absorbed energy is used for photochemistry (if the sam- ulation are followed. These are probed by a weak laser
ple is photochemically active), and the rest is emitted as beam, using the resulting deviations in the beam direction.
P1: GTQ/GLE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN012C-562 July 26, 2001 15:30
Photoacoustic Spectroscopy 3
4 Photoacoustic Spectroscopy
Photoacoustic Spectroscopy 5
are optically excited by groups of pulses, and data are limitation, a gas-permeable cell was designed to enrich the
acquired by a computer. carbon dioxide levels inside the closed cell. In addition,
from the time the PA method was initially used, in vivo
measurements were made on detached cut samples (either
C. Photoacoustic Cells
plant samples or animal or human skin samples). Though
During the past decade, PA cell architecture has trans- these measurements reflect the in vivo observations, real-
formed significantly from the basic concept of simple istically they may be called in situ observations. Hence,
closed cell with a microphone to that of a highly evolved an open PA cell was designed that can be attached to the
multichambered open cell capable of recording signals on leaf, and is reported to measure the photosynthesis of a
live objects such as intact plants and human beings. An- leaf still attached to the plant.
other factor that has improved significantly is the signal-
to-noise ratio, where the cells were designed for taking
2. Photoacoustic Open Cell
measurements in the field. In the literature, one can see
mention of a variety of photoacoustic cells. Considering certain limitations with closed cells, inves-
tigators at different labs started working on developing
open cells. Recently in our laboratory, we have designed
1. Basic Photoacoustic Cell
a PA cell that can be attached directly to a live human be-
The basic concept is that the PA cell must have a sensi- ing for measuring a drug or solar cream diffusion rates in
tive microphone, and it should be hermetically air-tight the skin. The cell is designed to have identical cylindrical
to generate an acoustic wave and to avoid external noise cavities fitted with microphones. A light pulse or beam is
(Fig. 4). Usually the cell is fitted with a frontal quartz passed through a fiberoptic light guide. One microphone
window for sending in the modulated excitation light onto measures the background noise from the cardiac pulse,
the sample, and the volume of the cell should be small. and the other detects the acoustic waves generated due to
Since it is air-tight, the gas composition of the cell can vary the excitation of compounds of interest inside the skin at
during the experiment. For example, a green leaf such as the contact site. The cell can be strapped to any part of the
photosynthetically active green leaf, which takes up car- body, and the signals can be recorded.
bon dioxide from and releases oxygen into surrounding
air, can alter the composition of the air in the cell. If one
3. Cell with Optical Microphone
measures the photochemical activity of leaf, there may be
limitation of the carbon dioxide to the photosynthetic phe- As mentioned above, the thermal deactivation-induced
nomenon during the measurements. In order to avoid this acoustic waves were detected by sensitive microphones
in a majority of PA studies. In order to improve the sen-
sitivity of the technique, a new detection system with op-
tical microphone was used to detect the acoustic wave
generated from the sample. In this system, the acoustic
wave-induced change in the position of a laser beam on
a pellicle is used to detect the signal. Instead of a micro-
phone to detect the signal, a laser beam positioned on a
Mylar pellicle (10 µM thickness) monitors the acoustic
waves. The vibrations of the pellicle caused by acoustic
waves deflect the laser beam, which is in turn is detected
by a silicon photodiode (Fig. 5).
4. Cuvette Cell
A standard cuvette with 1-cm path length is used in this
type of detection system. Samples are excited with a pla-
nar beam of light and the resulting pressure wave due to
solvent expansion through the solution is converted by a
transducer into a voltage pulse. As the measurements are
made using a cuvette with an attached transducer, the sys-
tem is easily adjustable to the needs of measurements in
FIGURE 4 Components of a typical photoacoustic cell. solutions.
P1: GTQ/GLE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN012C-562 July 26, 2001 15:30
6 Photoacoustic Spectroscopy
5. Layered Prism Cell ration, depth-profiling capability, and high signal satu-
ration limit. It can be used for analyzing any kind of
Layered prism cell was assembled using a pair of dove-
sample in any state. Thermophysical properties such as
tail prisms and a pair of piezoelectric transducers. The
thermal diffusivity and conductivity of certain samples
prisms are clamped together and separated by a shim. A
such as specialty papers have been investigated using
1-cm hole in the shim forms the sample compartment, and
Fourier transform-infrared–PAS (FTIR–PAS). This tech-
has the geometry of a thin disk. The inlet and outlet holes
nique has been used to improve our understanding of the
through one of the prisms permit the flow of the solution
chemical origin of depth inhomogeneities.
through the sample compartment. Light passes through the
FTIR–PAS can also be used as a rapid and nondestruc-
first prism, into the sample compartment, and out through
tive technique for quantitative characterization of the sur-
the second prism. This cell uses the transmission of light
face degradation of aged polymers and other compounds.
through the prisms, and it combines the enhanced time
Using FTIR–PAS spectroscopy, oxidation profiles of the
resolution capabilities of “layered” front face irradiation
various photooxidized polyisoprene–polystyrene copoly-
geometry with zero-background and broadband flexibility
mers have been investigated.
of classical cuvette geometry. As described, different labs
have developed different types of cells depending on the
B. Biophysics of Proteins
needs of their work.
and Organic Molecules
Photoacoustic Spectroscopy 7
determination of bond dissociation energies in complex, of tetracycline topically applied to skin. Tetracycline was
biologically important molecules. It has also proved to detected at 380 nm, where it strongly absorbs. The ease of
be a sensitive indicator of conformational distortions and identification of the substance in the spectral range sug-
mechanistic detail as biological molecules undergo reac- gests that penetration and distribution of other chemicals
tions. Using PAS, melanin content has been determined in through the skin can also be investigated by FTIR–PAS.
human hair samples. A recent study reports a novel percutaneous absorptio-
metric system using an open-ended PA cell combined with
C. Dermatology a light guide. This system is capable of measuring the de-
crease of the amount of a drug over the skin in vivo at any
Skin forms a protective boundary layer between the body given site with great ease, high sensitivity, and accuracy.
and its environment. With respect to transport proper- Using the UV laser PA method, it demonstrates a series
ties, the nonvascularized epidermis, the outermost layer of in vivo and in vitro percutaneous absorptiometric mea-
of which is formed by the stratum corneum (SC), serves surements in the presence of anti-inflammatory and anti-
as a barrier for the body. In the underlying vascularized itching agents. Melanin nonradiative de-excitation states
dermis, the transport properties are strongly influenced by and skin lipids have also been investigated in human skin
blood circulation. samples.
As mentioned earlier, PAS is a good tool for studying Using PAS, the influence of SC constituents such as
the optical and thermal properties of a sample. Hence, water, lipids, and hydrophilic substances on UV light ab-
this technique can be applied to fields of dermatological sorption have been investigated. The data reveal that upon
research such as drug detection and drug diffusion studies hydration of SC, there is a decrease of UV absorption coef-
in skin, and thermal properties and water content of the ficient by about 20%. Lipids were not shown to have any
skin. The optical and thermal properties of the sample are direct effect on UV absorption by SC. Further, the data
characterized by the optical absorption distance, µβ , and suggest that the hydrophilic substances of SC were pro-
the thermal diffusion distance, µs . The optical absorption tected in situ by polar lipids enmeshed within the structure
distance is the inverse of the optical absorption coefficient of SC.
β. The thermal diffusion length is defined by Using classical modulated PA and pulsed PA tech-
α niques, skin hydration, drug penetration rates, and dif-
µs =
πf fusion of solar creams into the human skin have been in-
vestigated. Using pulsed PAS (PPAS), diffusion kinetics
where α is the thermal diffusivity of the sample and f is
of five solutions at different concentrations in a mixture
frequency. The thermal diffusivity is given by
of chromophores, as used in commercial sunscreens, were
κ
α= studied, and a two-step data interpretation for better char-
ρc acterization of diffusion characteristics was reported. The
where κ is the thermal conductivity, ρ is the density, and diffusion process is monitored by considering the ampli-
c is the thermal capacity. Note that the thermal diffusion tude of the heat response pulse and the time delay of max-
length, µs , decreases with the increase in frequency and imum heat emission (Fig. 6). By analyzing the signal in
vice versa. Thus, by varying the modulation frequency, it is
possible to study the diffusion rates of the drug in the skin.
Based on the ability to perform depth profile analysis,
PAS has been shown to be suitable for the study of the
penetration of substances topically applied to skin, in the
visible spectral range. The absorption bands from specific
group components of substances can be easily identified
in the infrared spectral range, and a technique like PAS is
highly advantageous for studying the distribution of sub-
stances through the skin. Using PAS, the depth profile of
β-carotene and the differential penetration rate of
β-carotene and methylene blue dye in the skin were inves-
tigated. It is reported that methylene blue dye stayed at the
surface of the skin, while β-carotene penetrated up to the
epithelium cell layer. Using FTIR–PAS, dimethylsulfox-
ide diffusion into the skin has been studied. Another exam- FIGURE 6 Pressure pulse evolution during diffusion kinetics for
ple of drug detection in tissue using PAS is the detection a 20% chromophore solution deposited onto the skin.
P1: GTQ/GLE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN012C-562 July 26, 2001 15:30
8 Photoacoustic Spectroscopy
frequency and time domains, we found a clear difference tochemotherapeutic agents are irradiated in vivo, a rapid
in the diffusion behavior of the sunscreen compounds in photobleaching of the sensitizer is normally observed.
the skin. The combination of Fourier transform combined PAS has been used to monitor such photobleaching in a
with PPAS allowed us to determine the depth profile of highly scattering medium. It may also be possible to locate
the chromophores such as solar creams and other drugs. and delimit diseased areas in the case of external cancers
The frequency spectra helped to locate the chromophore from the hematoporphyrin signals and to follow the regres-
diffusion with respect to the different layers of skin. sion of the tumor. Similarly, photophysical parameters of
the naphthalocyanines, merocyanine 540 having various
applications in photobiology, have been investigated us-
D. Oncology and Photosensitizers
ing fluorescence and PA spectroscopies. A significant im-
Investigations on the photosensitized reactions in provement in instrumentation made during the past decade
medicine, particularly in photodynamic therapy, have been has resulted in much wider interest in the applications of
of considerable interest in recent years. Intensive research PAS in dermatology.
efforts are being made currently to the study of photosensi-
tized reactions in medicine, particularly in photodynamic E. Ophthalmology
therapy of tumors. Photodynamic therapy is a common
treatment for cancer and various pathologies such as pso- The visual process is a series of reactions beginning with
riasis. These treatments rely on the administration of pho- the absorption of a quantum of light energy. There are
tosensitizer and the irradiation of the zone treated. In the several temperature-dependent reactions during the con-
photosensitized reactions, the photosensitizer is excited version of the visual pigment rhodopsin to retinol. Sev-
by absorbed light and undergoes intersystem crossing to eral intermediates are formed during this process. Using
triplet state from which it either reacts with the nearby the depth profile method, a three-dimensional image of
molecules or produces cytotoxic oxygen derivatives, such the whole retina is recorded at low temperatures by vary-
as singlet oxygen or superoxide anion. The drugs used ing the frequency. At high frequencies the rhodopsin–
for photosensitization processes must show very low fluo- bathorhodopsin photostationary mixture appears to be the
rescent quantum yield and nonradiative decays with very main absorbing species. At low frequencies, deeper lay-
high efficiency. In this sense, PAS is an excellent tool for ers are probed, and the PA spectra reveal the increasing
studying the photophysical properties of the drugs and amplitude of the PA band at 412 nm, which is due to
their interaction with biological molecules. blood hemoglobin. The PA data suggest that in the 25-µm
The photophysical properties of the photosensitizers are layer which is probed between 50 and 60 Hz, the first half
usually monitored to determine the efficiency of the treat- (0–13 µm) essentially contains the rhodopsin (outer rod
ments. Parameters such as the lifetime of excited states, the segments).
intersystem crossing yield, and singlet oxygen O2 (1 g ) The bovine retinal pigment epithelium (RPE) is a single
production yield are known only in solution form. As the layer of cells underlying the photoreceptor outer segments
PA and photothermal spectroscopic techniques are least and is supported by the highly vascularized choroid. Upon
sensitive to light scattering and are known to be sensi- dissection and removal of the retina, the RPE appears as
tive to relaxation processes, several attempts have been an area of different colors, which are known as amelan-
made to use PAS to detect the photosensitizers in solu- otic and melanotic RPE. Using PAS, these amelanotic and
tion, in tissues, in excised organs, and in skin in situ and melanotic RPE areas (Fig. 7A, B) are detected in the RPE.
in vivo. Porphyrin derivatives are being actively investi- In healthy bovine eye lenses, PAS revealed the presence
gated as photosensitizers for use in photodynamic therapy of the 340- to 360-nm band, which increases in magnitude
of tumors. The aggregation tendency of many porphyrins with age and is particularly pronounced in brown nuclear
may be related to their selective retention in neoplastic tis- cataracts. UV irradiation effects on eye lenses of dogfish
sues. Triplet state energy levels and intersystem crossing sharks and rats have also been investigated using PAS.
quantum yields of porphyrin derivatives have been ob- Other biomedical applications of PAS include the study
tained by measuring the phase angle difference between of the different states of blood hemoglobin, cytochromes,
the PA signals of an unquenched sample and of the sample pharmaceutical powders, and diary products.
with triplet quenched. For molecules whose photophysi-
cal properties cannot easily be obtained by purely opti- F. Plant Biology
cal means, such an approach can prove to be a powerful
1. Biomass Determination
method. Similarly it is also possible to study the reaction
kinetics in environments that are not possible to monitor PAS was used to determine the number of cells in the
via absorbance. For example, when porphyrin based pho- fermentation of cyanobacteria, as this technique is not
P1: GTQ/GLE P2: GPJ Final Pages
Encyclopedia of Physical Science and Technology EN012C-562 July 26, 2001 15:30
Photoacoustic Spectroscopy 9
10 Photoacoustic Spectroscopy
Photoacoustic Spectroscopy 11
12 Photoacoustic Spectroscopy
Photoacoustic Spectroscopy 13
SEE ALSO THE FOLLOWING ARTICLES tation energy between the two photosystems in vivo: Photoacoustic
and fluorimetric study of an intact leaf,”Proc. Natl. Acad. Sci. USA
81, 1614–1618.
ACOUSTIC WAVE DEVICES • AUGER ELECTRON Charland, M., and Leblanc, R. M. (1993). “Photoacoustic spectroscopy
SPECTROSCOPY • INFRARED SPECTROSCOPY • MODULA- applied to biological systems,” Bulletin of the Institute for Chemical
TION • PHOTOCHEMISTRY, MOLECULAR • POLLUTION, Research, Kyoto University, 71, 226–244.
AIR Crippa, P. R., Vecli, A., and Viappiani, C. (1994). “Time-resolved pho-
toacoustic spectroscopy: New developments of an old idea,” J. Pho-
tochem. Photobiol. B: Biol. 24, 3–15.
Fork, D. C., and Herbert, S. K. (1993). “The application of photoacoustic
BIBLIOGRAPHY techniques to studies of photosynthesis,” Photochem. Photobiol. 57,
207–220.
Autrey, T., et al. (1998). “A new angle into time-resolved photoacoustic De Paula, M. H., Vinha, C. A., and Badini, R. G. (1992). “High-
spectroscopy: A layered prism cell increases experimental flexibility,” sensitivity optical microphone for photoacoustics,” Rev. Sci. Instrum.
Rev. Sci. Instrum. 69, 2246–2258. 63, 3487–3491.
Bock, G., and Harnett, S. (1989). “Photosensitizing Compounds: Malkin, S., and Canaani, O. (1994). “The use of the characteristics of
Their Chemistry and Clinical Use,” Wiley, Chichester, UK. the photoacoustic method in the study of photosynthesis,” Annu. Rev.
Boucher, F., and Leblanc, R. M. (1981). “Photoacoustic spectroscopy Plant Physiol. Plant Mol. Biol. 45, 493–526.
of cattle visual pigment at low temperature,” Biochem. Biophys. Res. Repond, P., and Sigrist, M. W. (1996). “Photoacoustic spectroscopy on
Commun. 100, 385–390. trace gases with continuously tunable CO2 laser,” Appl. Optics 35,
Braslavsky, S. E., and Heibel, G. E. (1992). “Time resolved photother- 4065–4085.
mal and photoacoustic methods applied to photo-induced processes Rosencwaig, A. (1975). “Photoacoustic spectroscopy. A new tool for
in solutions,” Chem. Rev. 92, 1381–1410. investigation of solids,” Anal. Chem. 47, 592–604.
Buschmann, C. (1990). “Photoacoustic Spectroscopy—Photoacoustic Rosencwaig, A., and Gersho, A. (1976). “Theory of photoacoustic effect
and Photothermal Effects,” In “Modern Methods of Plant Analysis, with solids,” J. Appl. Phys. 4, 64–69.
New Series,” Vol. 11, Springer-Verlag, Berlin. Sigrist, M. (1998). “Encyclopedia of Environmental Analysis and Re-
Cahen, D., et al. (1980). “Photoacoustics in life sciences,”J. Biochem. mediation,” Wiley, New York.
Biophys. Method. 3, 293–310. Veeranjaneyulu, K., et al. (1991). “Photoacoustic study of changes in
Canaani, O., Barber, J., and Malkin, S. (1984). “Evidence that phos- energy storage of photosystems I and II during state 1–state 2 transi-
phorylation and dephosphorylation regulate the distribution of exci- tions,” Plant Physiol. 34, 271–334.
P1: GLQ Final pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
Photoelectron Spectroscopy
G. Hohlneicher
A. Gildenpfennig
University of Cologne
57
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
58 Photoelectron Spectroscopy
Before photoionization takes place, the system is in a well- scale for the binding energy at the bottom. The binding
defined electronic state, usually the electronic ground state energy scale is obtained from the kinetic energy scale by
M0 , the initial state of the photoemission process. Irra- means of Eq. (2). It must be kept in mind, however, that
diation of the system with radiation of sufficiently high the quantity originally measured is E kin (e− ), even when
energy hν leads to the ejection of a photoelectron. The only a scale for E B is shown. The scales in Fig. 1 run in
ion M + that is created by this process is again in a well- different directions for different examples. This is not a
defined electronic state Mi+ which is the final state of the mistake but is due to different presentations of PE spec-
photoemission process. In general, the lifetime of Mi+ is tra in the literature. Some researchers show values of the
long enough to prevent successive changes in the ion state measured quantity E kin (e− ) increasing from left to right,
from influencing the kinetic energy of the photoelectron. which results in the scale for E B running from right to left.
The final state is either the electronic ground state M0+ or, This kind of display is especially common for PE spectra
provided hν is sufficiently high, an electronically excited of solid surfaces. Others are interested mainly in E B , so
state of the ion. From conservation of energy it follows that they draw the E B scale with increasing energies from left
to right. Sometimes only the E kin (e− ) scale or only the
E(Mi+ ) − E(M0 ) + E kin (e− ) + E kin (M + ) = hν (1)
E B scale is provided. As far as possible we will show both
where E(M0 ) and E(Mi+ ) are the energies of the initial scales throughout this article, but when using spectra from
and final states, respectively, and E kin (e− ) and E kin (M + ) the literature one should always be aware of the scale used.
are the kinetic energies of the electron and of the ion. Figure 1 gives a first impression of different types of PE
Since E kin (e− )/E kin (M + ) is determined by the mass spectra. From Eq. (2) we expect photoelectrons to appear
ratio m(M + )/m(e− ), E kin (M + ) can be neglected in most only at kinetic energies that correspond to a certain final
applications. This leads to state Mi+ . We therefore expect a PE spectrum to consist
of lines with widths defined by some experimental param-
E B (i) = E(Mi+ ) − E(M0 ) = hν − E kin (e− ) (2)
eters. This is the result observed for neon (Fig. 1a). For
where E B (i) is called the “binding energy.” To avoid molecules, vibrational and rotational states are coupled to
confusion, this expression is used throughout this article. the electronic states and, as in optical spectroscopy, band
However, specifically in connection with the investigation spectra are obtained (Fig. 1b). In this case we speak of pho-
of free molecules, the energy difference E(Mi+ ) − E(M0 ) toemission bands rather than photoemission lines. With
is also referred to as the “ionization energy” or “ionization the exception of the technique discussed in Section I.I, the
potential.” resolution achievable in PES is much lower than in opti-
If the excitation energy hν is known, Eq. (2) allows cal spectroscopy (Section III.B). If the sample is a solid
determination of E B (i) from the kinetic energy of the cre- (Fig. 1c), each band or line is preceded by a tail extending
ated photoelectrons. Photoelectron spectroscopy (PES) is toward lower kinetic energies. For low kinetic energies this
basically the measurement of the kinetic energy of pho- leads to a considerable background. The tails are due to
toelectrons with the goal of deriving information about inelastic scattering of the photoelectrons within the solid.
binding energies. This definition distinguishes PES from The progress achieved with PES results from the fact
other methods in which photoionization is used mainly for that it is an energy-resolved method. Older methods for the
detection (e.g., laser-induced multiphoton ionization) and determination of binding energies were mainly based on a
not for determining binding energies. measurement of the photoionization current and depended
on a variable excitation energy. For a given excitation en-
ergy E a the photoionization current is proportional to the
B. Photoelectron Spectra
integral over the PE spectrum from E B = 0 to E B = E a .
A photoelectron (PE) spectrum is the number of photo- Even if a sufficiently variable source is available for ex-
electrons with kinetic energy E kin (e− ) observed per unit citation (which was not the case prior to the invention
time, displayed as a function of kinetic energy. Three of synchrotron radiation), an energy-resolved method is
examples are shown in Fig. 1: The gas-phase spectrum always preferable to an integral one. In an energy-resolved
of neon excited with h ν = 1253.6 eV (Fig. 1a), the gas- measurement, only electrons with a kinetic energy that
phase spectrum of H2 C O excited with hν = 21.2 eV falls in the window defined by the resolution of the ana-
(Fig. 1b), and the solid-state spectrum of copper excited lyzer contribute to the statistical noise, whereas in an inte-
with h ν = 1486.7 eV (Fig. 1c). The meaning of the as- gral method, all electrons with E kin ≤ E a contribute. The
signments given in these spectra is explained below. In older methods therefore permitted determination of only
all three cases two energy scales are shown correspond- the first or, under fortunate conditions, the first few binding
ing to the IUPAC recommendations: The scale for the energies. The photoemission processes leading to higher
kinetic energy is given at the top of the spectrum and the excited final states only became accessible with PES. In
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
Photoelectron Spectroscopy 59
FIGURE 1 Examples of photoelectron spectra: (a) neon, hν = 1253.6 eV; (b) formaldehyde, hν = 21.2 eV; (c) metallic
copper, hν = 1486.7 eV. [Spectrum of neon, from Siegbahn, K., et al. (1971). “ESCA Applied to Free Molecules,”
Elsevier, Amsterdam.]
Section II we will discuss in detail why and how these details). Only valence electrons can be photoionized with
processes in particular contribute to a better understand- this energy. The development of UPS in its application to
ing of the electronic structure of the systems investigated. free molecules is mainly due to work by D. W. Turner and
First, however, we will clarify two terms frequently used associates in Oxford, who reached a breakthrough in the
in connection with PES: low- and high-energy PES. We early 1960s.
will also discuss some historical aspects that will aid the In high-energy PES, characteristic x-irradiation with
reader in understanding the development of PES. energies between 100 and 2000 eV and half-widths of
In low-energy PES, usually termed ultraviolet PES or about 1 eV is used for excitation. Because of the higher ex-
UPS, far-UV radiation is used for excitation. The most citation energy, valence electrons as well as electrons from
common source for UPS is the helium resonance lamp, inner shells (core electrons) can be photoionized. The core
which provides radiation with an energy of 21.2 eV and electrons yield direct information on elemental composi-
a half-width down to 10 meV (see Section III for further tion and on the chemical state of a given element. Changes
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
60 Photoelectron Spectroscopy
in the chemical state, like different oxidation states or ex- gion is very likely for larger molecules), a photon can be
change of substituents, lead to small changes in the bind- absorbed by the neutral system. In case the final state Mk
ing energy of core electrons. These changes or “chemi- of this process has a higher energy than the ground state
cal shifts” are the key to most analytical applications of of M + , Mk can decompose into an electron and a low-
high-energy PES. Because of the lower resolution, the in- lying state of M + . Since both the initial and the final state
formation obtained for valence electrons is more limited of the autoionization (AI) process are well-defined elec-
than in UPS. The development of high-energy PES, which tronic states, the electron created in this process has the
is usually termed electron spectroscopy for chemical anal- well-defined kinetic energy
ysis (ESCA) or X-ray photoelectron spectroscopy (XPS), AI −
E kin (e ) = E(Mk ) − E(Mi+ ) (3)
began in 1914 with Robinson and Rawlinson’s investi-
gation of the photoemission of gold. Steinhardt and Ser- In Eq. (3) we again neglected the rebound energy of the
fass succeeded in 1951 in the first analytical application, heavy particle. Unlike the kinetic energy of an electron
AI −
which was then perfected by K. Siegbahn and coworkers produced in process 1, E kin (e ) does not depend on hν.
in Uppsala in the late 1950s. In 1981 Siegbahn was hon- Therefore, the binding energy scale has no meaning for
ored with the Nobel Prize for this significant development. autoionization peaks. These peaks appear primarily at low
Much of his pioneering work on the analysis of solid or- kinetic energies since autoionization is usually efficient
ganic and inorganic materials was published in a single only when the initial and final states of the process are
volume in 1967. A second volume following 2 years later close in energy.
contained applications to free molecules. Process 3 in Fig. 2, is a direct transition from M0 to
states of M 2+ under simultaneous emission of two elec-
trons. As discussed in Section II, such a process has a much
C. Competing Photoemission Processes lower probability than process 1 and leads to a continuous
The photoemission processes M0 → Mi+ are not the only energy distribution of the produced photoelectrons.
processes that lead to the appearance of electrons with Process 4 represents an Auger transition. As discussed
well-defined kinetic energies. Depending on the energy above, photoionization can lead to the creation of a core
of the exciting radiation, some other processes also con- hole provided the energy of the exciting radiation is high
tribute to the PE spectrum. These processes are indicated enough. The resulting state of M + is highly excited. In
by the circled numbers in Fig. 2, which shows a schematic about 10−16 sec, it relaxes to a lower excited state of M +
diagram for the electronic states of the initial system (M) by emission of an X-ray photon or to a lower lying state of
and the systems that have lost one (M + ) or two (M 2+ ) M 2+ by emission of another electron. The latter transition,
electrons. known as an Auger process, has a higher probability for
Process 1 is the process discussed in the previous sec- light atoms, up to about Z = 40. Since the initial and final
tion. It corresponds to photoionization by emission of ei- states of an Auger transition are well-defined electronic
ther a valence (1a) or a core (1b) electron. states of M + and M 2+ , respectively, the emitted Auger
Process 2 represents an autoionization. If the energy electron has the well-defined kinetic energy
hν of the exciting radiation coincides with an electronic 2+
Au −
E kin (e ) = E M + j − E Mk (4)
transition of the neutral system (which in the valence re-
Auger transitions contribute strongly to high-energy PE
spectra. For example, in Fig. 1c all peaks indicated with
an asterisk result from Auger transitions. As in autoion-
ization, the kinetic energy of the Auger electron does not
depend on hν. Autoionization and Auger processes, there-
fore, can be separated from photoemission processes by
variation of the excitation energy.
Photoelectron Spectroscopy 61
62 Photoelectron Spectroscopy
Photoelectron Spectroscopy 63
64 Photoelectron Spectroscopy
field in which ARPES is important is the investigation the beginning, it was assumed that it is the inelastic mean
of two- and three-dimensional periodic structures. Two- free path (IMFP) that has to be used in connection with
dimensional periodic structures are found in layer com- these equations. The IMFP is an intrinsic property of a
pounds or in well-ordered adsorbate systems formed at material which can be measured using other techniques,
the surface of a single crystal. Three-dimensional periodic such as electron backscattering. In the mid-1980s it was
structures are found in all crystalline materials. In most found that the appropriate length is not the IMFP but the
cases the application of ARPES asks for single crystals of attenuation length (AL). Powell defined the AL as “a value
at least a few millimeters in the two surface dimensions. resulting from overlayer-film experiments on the basis of
With an illuminated area of less than about 1 mm2 one a model in which elastic electron scattering is assumed to
either loses too much intensity or risks too much radiation be insignificant.” The AL is typically 10 to 25% shorter
damage. than the IMFP because of the contribution of elastic scat-
The important difference between free or space-fixed tering to the IMFP. If the thickness of the overlayer is
single molecules and periodic structures is as follows. In larger than about three times the AL, the substrate is no
the first case, the intensity of the peaks in the PE spec- longer detectable. This limits the applicability of ARPES
trum is angle dependent but the position of the peaks is to overlayers that are not more than a few nanometers
not. In the second case, the energy of the most prominent thick. Some terms concerning electron transport in matter
features in the spectrum of the valence region depends on are defined in Table II.
the angles
, θ , and φ as well as the polarization of the To convert the “composition versus take-off angle”
ionizing radiation. One therefore speaks of energy distri- curves obtained from an angle-resolved XPS (ARXPS)
bution curves instead of PE spectra in such a case. We will
come back to this type of investigation in Section IV.B.
ARPES has become more and more common in the in- TABLE II Some Definitions with Regard to Electron
vestigation of substrate/overlayer structures. An overlayer Transport in Matter
is a zone of thickness d that differs in chemical composi-
IMFP I nelastic Mean Free Path. The average distance
tion from the substrate as, for example, a metal oxide on that an electron with a given energy travels
the surface of a metal. In materials science, the knowledge between inelastic collisions.
of d is often very important (see, e.g., Fig. 30 in Section ED Escape Depth. The distance normal to the surface at
IV.E). To determine d, one makes use of the limited es- which the probability of an electron escaping
cape depth of photoelectrons and the difference in binding without significant energy loss due to inelastic
energy that leads to distinguishable peaks for pure metal processes drops to e−1 (38%) of its original value.
and metal oxide. The larger the d, the smaller becomes AL Attenuation Length. The average distance that an
electron with a given energy travels between
the intensity ratio Imetal /Imetal oxide . Increase of the polar inelastic collisions as derived from a particular
angle θ increases the effective thickness of the overlayer model in which elastic scattering is assumed to
and leads to characteristic changes in the intensity ratio. be negligible.
Equations that connect the observed θ dependence of the ID I nformation Depth. The average distance normal to
intensity ratio with d were derived under the following the surface from which a specified percentage of
two assumptions: (1) the surface is atomically flat and the detected electrons originates.
(2) electrons which reach the detector without energy loss Sampling Three times the IMFP corresponding to an ID for
Depth which the percentage of detected electrons is 95%.
have traveled in straight lines (no elastic scattering). In
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
Photoelectron Spectroscopy 65
experiment to a plot of “depth versus composition,” it is states. The resulting electrons are ejected from the ex-
necessary to deconvolute the observed data. A number of citation region and measured. This combination of laser
algorithms for such a deconvolution has been published excitation and pulsed-field ionization makes it possible to
but as the experimental data always contain some amount resolve the rotational structure of small molecular ions
of statistical error, it is often more appropriate to compare and the vibrational structure of larger ones.
curves obtained for certain layer models with the experi-
mental data.
II. PHOTOEMISSION FROM ATOMS,
H. Spin-Resolved Photoelectron Spectra
MOLECULES, AND SOLIDS
gion. An electrical pulse then field ionizes the Rydberg obtained with the same basis set.
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
66 Photoelectron Spectroscopy
FIGURE 8 Orbital representation of (a) the initial and (b–f) different final states of the photoemission process.
of systems with such a ground state. It is an exception one of the occupied orbitals without disturbing the remain-
for atoms but is very common for molecules. Apart from ing electrons. This is called the “frozen orbital approxima-
the orbitals that are occupied in the ground state, there tion” (FOA). This process creates a hole in the manifold
are unoccupied or “virtual” orbitals. For example, for the of the occupied orbitals, and we call the resulting state a
neon atom, with a ground state configuration 1s 2 2s 2 2 p 6 , “hole state.” If the electron is removed from the highest
the lowest unoccupied orbital is 3s. In the following dis- occupied orbital, the final state of the photoemission pro-
cussion, it is important to keep in mind that “occupied” cess is the ground state M0+ of the ion (Fig. 8b). If the
always refers to the set of orbitals occupied in the ground electron is removed from a lower lying orbital, an excited
state M0 . Correspondingly, “unoccupied” always refers to state of the ion is reached which we call a “hole excited
the set of orbitals not occupied in the ground state. state” because moving a hole downward is the same as
When we turn from atoms to molecules, the situation moving an electron upward (Fig. 8c). When the electron
changes only slightly. The major difference is that the is removed from a core orbital, we speak of a “core hole
valence orbitals no longer belong to a single atom. To a state” (Fig. 8d). Such a state is usually labeled by the
greater or lesser degree, they are extended over the whole chemical symbol of the atom and the orbital from which
molecule. The same is true for the unoccupied orbitals. the electron has been removed (e.g., C 1s, O 1s, P 2 p).
The core orbitals are still localized at their respective If we start from M0+ (Fig. 8b) and excite one of the
atoms. Degenerate valence orbitals are less common in remaining electrons to an unoccupied orbital (Fig. 8e),
molecules than in atoms; they are usually found in mole- a new type of excited ion state is reached. Compared to
cules of higher symmetry with at least one three-fold axis. the ground state M0 of the initial system, this state has
For solids, the number of atoms and therefore the two holes in the occupied orbitals and one electron in an
number of orbitals goes to infinity. The valence orbitals unoccupied orbital. Therefore, it is called a “two-hole one-
develop into continuous energy bands. The details of the particle (2h1p) state.” Analogously, we can define three-
electronic structure of a crystalline solid are described in hole two-particle states, and so on.
terms of the “band structure.” The energy up to which the
bands are occupied is the Fermi energy (E F ) already men-
tioned in Section I.E. When the Fermi energy lies within C. Secondary Structures in PE Spectra
a band, the solid is a metal; when it lies in a gap between
The nh(n-1) p states are frequently called “shake-up
two bands, the solid is a semiconductor (small gap) or
states,” based on the idea that photoionization is such
an insulator (large gap). As in molecules, the core orbitals
a strong perturbation that the whole electron system is
in solids still behave like those in atoms.
“shaken” and one or more of the remaining electrons are
“shaken up” to unoccupied orbitals. If the “shake-up elec-
B. The Frozen Orbital Approximation
tron” receives enough energy to leave the system, the final
We now return to photoemission. For a first approxima- state of the photoionization is a state of the doubly ion-
tion, we assume that we can remove a single electron from ized system (Fig. 8f). This type of transition was shown as
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
Photoelectron Spectroscopy 67
process 3 in Fig. 2 and, in a further extension of the above valence orbitals and the vibrational structure of the cor-
ideas, is called a “shake-off” process. Since the second responding photoemission bands (Section II.G), shake-up
electron is no longer bound, shake-off excitations lead to structures in the valence region are more difficult to ob-
a continuum. This corresponds to the continuum following serve. Figure 3 shows an example where it was possible to
discrete excitations in the optical spectrum of the ion. detect a shake-up structure (indicated 2h1p) bycomparison
Examples of the processes discussed above are seen in with the optical spectrum of the ion. A transition from the
the photoemission spectrum of neon (Fig. 1a). The first ion ground state M0+ to this particular 2h1p state is allowed
line appears at 21.6 eV, corresponding to photoionization and leads to a strong band in the absorption spectrum of
of one of the six 2 p electrons. The second line, at 49 eV, the ion. In the PE spectrum, which starts from M0 instead
corresponds to photoionization of a 2s electron and the of M0+ , the 2h1p final state gives rise to only a weak struc-
third line, at 870 eV, to photoionization of a 1s electron. ture. From the PE spectrum alone it would be difficult to
On an enlarged intensity scale (Fig. 9) shake-up satellites detect this weak structure as a result of an independent
are seen at the high binding energy side of the Ne 1s line, transition. Comparison with the UV spectrum of the ion,
which result from transitions into 2h1p states. The lines however, allows an unambiguous assignment.
designated 3, 5, 6, and 7, for example, have been identified In the examples shown, the shake-up transitions only
as transitions into states where, in addition to the creation lead to weak structures in the PE spectrum. This is a very
of the 1s hole, a 2 p electron is excited to 3 p, 4 p, 5 p, general observation. Only in special cases can shake-up
and so on. In addition, a shake-off continuum underlies satellites reach an intensity comparable to that of the main
the high-energy part of the 1s shake-up spectrum. Shake- line (see Section IV.D and Fig. 28). Why are shake-up tran-
up satellites are especially common for the 2 p spectra sitions usually weak? In the FOA, transitions are allowed
of metals with d-electrons, as for example in the Cu 2 p only when the initial and final states differ by one electron.
spectrum of copper(II) oxide shown in Fig. 10a, and in Transitions that involve more than one electron are forbid-
the spectra of aromatic organic compounds where they den. Since all 2h1p states differ from the initial state by at
result from transitions between bonding and antibonding least two electrons (Fig. 8), transitions from M0 to 2h1p
π-orbitals. Much weaker shake-up satellites have been states (and also to shake-off states) are strictly forbidden
reported for the σ → σ ∗ transitions in aliphatic organics. in the FOA. Only transitions to final single-hole states are
The appearance of shake-up structures is not restricted allowed. When degenerate orbitals are counted only once,
to the core region. Because of the close spacing of the the number of lines or bands observable in a PE spectrum
FIGURE 9 Shake-up satellites accompanying Ne 1s photoionization; hν = 1486.7 eV. [From Siegbahn, K. (1974). J.
Electron Spectrosc. 5, 34.]
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
68 Photoelectron Spectroscopy
FIGURE 10 Energy loss features in XPS spectra: (a) Cu 2 p shake-up satellites, (b) Ni 2 p multiplet splitting,
(c) plasmon loss features from clean aluminum. [From Watts, J. F. (1994). Vacuum 45(6/7), 653.]
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
Photoelectron Spectroscopy 69
should be equal to the number of occupied orbitals. There- ally less than 10% (see Table III). Therefore, KT is a close
fore, all the prominent bands in the valence region of free approximation.
molecules and all main lines in the inner shell region of However, one point always must be kept in mind if KT is
free molecules and solids are usually assignable in the applied: orbital energies are quantities solely related to the
single-particle picture. Basically, this type of interpreta- initial state of the photoemission process. The application
tion also holds for photoemission from solids. However, in of KT is equivalent to neglecting any influence of the final
this case some special aspects must be considered, which state on the binding energies. Obviously, this cannot be
will be discussed in Section IV.B. true. To take final state effects into account, we must go
Additional structures in PE spectra can result from mul- beyond the FOA. We return to this point in the following
tiplet splitting (Fig. 10b shows an example), which is dis- section.
cussed in more detail in Section II.H, plasmon losses, and In the application of PES to the study of molecular
even the X-ray source itself. Plasmon losses, which are electronic structure, KT is seldom used in the form of
of little analytical use, are observed in the PE and Auger Eq. (7). In most cases, a direct comparison of calculated
spectra of clean metal surfaces, as shown in Fig. 10c for εi and measured E B (i) is not the main objective. Often it is
a clean aluminum surface. They result from collective os- more interesting to investigate how a certain ε and corre-
cillations in the conduction band excited by the outgoing spondingly a certain E B vary with alterations in chemical
electron which thereby suffers a discrete energy loss. The constitution. In this case we use a “weaker” form of KT:
plasmon frequency and multiples thereof are observed as
a characteristic series of peaks on the high binding energy E B (i) = εi (8)
side of the main core line. where we connect binding energy shifts (E B ) with or-
The radiation source also gives rise to weak additional bital energy shifts. In considering these chemical shifts,
photoelectron signals, as, for example, both the AlKα and we do not completely neglect final state effects. We only
the MgKα lines have satellites some 10 eV below the assume that they are approximately constant within a cer-
main line with around 10% of its intensity. In addition to tain class of compounds. The chemically appealing fea-
these satellites, an oxidized or damaged anode (CuKα ) as ture of this approach lies in the great variety of models
well as cross-talk from the complementary side of a badly available to estimate orbital energy shifts. For example,
aligned twin anode can lead to the appearance of unwanted all the models that have been developed to describe the
radiation. influence of different substituents on physical properties
and chemical reactivities of molecules can be applied, and
D. Koopmans’ Theorem frequently a more direct proof of a given model is possi-
ble with PES. Examples of this type of application are
The correspondence between orbital picture and PE spec- discussed in Sections IV.A and IV.C.
trum goes even further than the one-to-one correspon- Of special importance is a model which connects core
dence between main lines and occupied orbitals. In the electron binding energy shifts to atomic charges. Chemists
FOA, the binding energies relative to the vacuum level are usually attribute partial charges q A to the different atoms of
directly connected with the orbital energy of the occupied a molecule, even though this concept is problematic from
orbitals a strictly theoretical point of view. In the “point charge
E B (i) = −εi (7) approximation,” the energy εi (A) of a core orbital at atom
A can be expressed in terms of partial charges by
This relation was derived by T. Koopmans in 1932 and is
now known as Koopmans’ theorem (KT). It is the ba- εi (A) = k(i, A)q A + V (q B ) + k0 (i, A) (9)
sis of most applications of PES in electronic structure
elucidation. If KT were strictly valid, we could experi- where q A is the charge at the considered atom, V (q B )
mentally observe orbital or single-particle energies, which the “off-atom potential” created by the charges at all other
in reality exist only in the theoretical framework of the atoms, and k0 and k are parameters specific for atom A and
independent-particle model. It must be clearly understood orbital i. Within the limits of applicability of Eq. (8), the
that orbital energies are not observable in the sense of model allows one to estimate changes in atomic charges
quantum mechanics. However, they can be calculated by from measured shifts of core electron binding energies.
a variety of different methods, and these calculations can
be performed with a high degree of accuracy for small
E. Final State Effects
and medium size molecules. A comparison of orbital ener-
gies from such calculations with experimentally observed In the preceding section we saw that the most prominent
binding energies shows that the deviation from KT is usu- structures in PE spectra can be explained adequately in the
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
70 Photoelectron Spectroscopy
single-particle approximation. Two facts, however, clearly theoretical model. It can be described, however, by a su-
show the limitations of this type of interpretation: (1) the perposition of the eigenstates Mk+ of the ion, where each
deviations from KT [Eq. (7)] and (2) the appearance of of these states contributes with a certain weight factor gk :
shake-up and shake-off satellites. + ∗
From these two observations, it is obvious that we can- Mi = gk Mk+ (11)
k
not completely neglect final state effects. Using KT as a
first approximation, we can write In the high-energy limit where the electron leaves the ion-
ized system very quickly (sudden limit), the probability
E B (i) = −εi − Ri (10) of a transition to a final state Mk+ is equal to gk2 . The main
where Ri is called the “relaxation energy.” The meaning contribution usually comes from a state Mi+ that has the
of Eq. (10) can be understood by the following Gedanken same orbital occupation as the hypothetical frozen orbital
experiment. We separate the photoemission process into state (Mi+ )∗ . However, because of the presence of the hole,
two steps. In the first step, we use the FOA and remove one the orbitals of Mi+ differ somewhat from the orbitals of
electron from a given orbital i, leaving all other electrons the initial state M0 . Since the weight factor gi is usually
unperturbed. To remove the electron we need the energy greater than 0.5, we normally can give an assignment of
−εi . In the second step, we consider the reaction of the the main bands or main lines of a PE spectrum in terms
remaining electrons to the presence of the hole created in of KT, as discussed in the previous section. The remainder
the first step. The system will “relax” to a new, energeti- of the weights is frequently distributed over a variety of
cally favorable situation. Therefore, the relaxation energy final states Mk+ including the continuum states, thus ex-
is usually positive. Only in rare cases can specific quan- plaining shake-up satellites and the shake-off continuum.
tum mechanical effects (so-called correlation effects) lead When we are able to describe Mk+ to a good approximation
to small negative values of R. In Table III we have com- by a single 2h1p state, we reach a situation as discussed in
pared experimental binding energies to calculated orbital the previous section in connection with Fig. 9, where sev-
energies. The relaxation energies derived from these two eral of the observed shake-up satellites could be assigned
sets of data vary from orbital to orbital. R is much larger to specific electronic excitations in the core ionized sys-
for core electrons than for valence electrons; however, rel- tem. For larger systems, where the number of possible
ative to the magnitude of the binding energy, R is similar electronic excitations becomes very large, we will be able
for all shells. to identify the shake-up satellites only in cases where the
From our Gedanken experiment, we suggest that Ri will remaining weight is not distributed more or less equally
be connected strongly to the mobility of the electrons in over a large number of final states. When gk has a some-
the system. If the orbital from which the electron is re- what larger value for a specific final state Mk+ , we will
moved is highly localized, as in a core orbital or a lone observe a characteristic shake-up satellite even for an ex-
pair orbital, the most efficient stabilization of the hole tended system. An example of such a situation will be
state will be achieved by transfer of negative charge to discussed in Section IV.D.
the vicinity of the hole. If a direct transfer is not possi- We now turn to binding energy shifts. From Eq. (10)
ble, the stabilization can be achieved only by polarization we obtain
of the surroundings. From this consideration we expect E B (i) = −εi − Ri (12)
larger relaxation energies for the core ionization of met-
als, where the electrons in the valence band move almost Binding energy shifts depend as much on initial state ef-
freely, than for the core ionization of insulators, where the fects (via εi ) as on final state effects (via Ri ). Often we
polarization of the nearest neighbor atoms yields the most are specifically interested in initial state effects, because
important contribution to the relaxation. For delocalized we want to derive information on the electronic structure
holes that result from the photoionization of π electrons of of the initial system M0 and its dependence on variations
unsaturated molecules or valence electrons of solids, the in chemical constitution. This information, however, can
relaxation contribution is expected to be smaller and less be derived only if R is negligibly small or if we are
dependent on the individual orbital. This is in accordance able to obtain independent information on R. The re-
with the data shown in Table III; for formaldehyde, for laxation contribution itself also contains valuable infor-
example, the smallest relaxation energy is found for the π mation, since it is connected with electronic relaxation
orbital 1b1 . processes that can take place during a chemical transfor-
The considerations discussed above can also be viewed mation. In a wide variety of chemical reactions the transi-
in a somewhat different manner. The ion state (Mi+ )∗ tion state is charged. The better this charge can be screened
formed in the FOA is not a real state (eigenstate) of the by a relaxation of the whole electronic system, the lower
investigated system. It exists only in the framework of the the energy of the transition state.
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
Photoelectron Spectroscopy 71
72 Photoelectron Spectroscopy
long as the temperature of the sample is not much higher and initial state. For example, the first ionization band of
than 300 K, vibrational excitations in the initial states do formaldehyde (Fig. 1b) shows a very intense 0-0 transi-
not perturb the observed spectra; low-frequency vibrations tion and little vibrational fine structure. From this we can
excited at these temperatures are covered by the limited conclude that the electronic ground state of the formalde-
resolution. hyde cation is very similar in geometry to the electronic
To first order, the intensity distribution of the different ground state of formaldehyde. For the second photoion-
lines of a vibrationally structured PE band can be inter- ization band the vertical transition corresponds to the fifth
preted in terms of the Franck–Condon principle (Fig. 12). vibrational level, telling us that the equilibrium geometry
The most probable transition is the “vertical transition.” In of the first electronically excited ion state differs consid-
a potential diagram, where the electronic energy is drawn erably from the ground state of the neutral system. In the
as a function of interatomic distances, the vertical transi- fourth ionization band of formaldehyde we observe a rapid
tion is best approximated by a vertical line drawn from loss of vibrational fine structure after the fifth vibrational
the minimum in the potential of the initial state to the excitation. This is a typical pattern for an excitation into
potential curve of the final state. It is the binding energy a dissociative final state (see Fig. 12). Excitation with an
corresponding to the vertical transition [the “vertical ion- energy greater than the dissociation limit leads to frag-
ization potential” (VIP)] that must be used in connection mentation of the ion. The vibrational structure becomes
with KT [Eq. (7)]. The transition from the lowest vibra- broadened in the region of the dissociation limit and then
tional level of the initial state to the lowest vibrational continuous. From the examples discussed it is clear that
level of the final state is called “adiabatic transition.” If the vibrational fine structure of a photoionization band
the minimum of the final state is not displaced with re- contains information on the geometry and the potential
spect to the initial state, vertical and adiabatic transitions surface of the final ion state.
fall together (Fig. 12). The difference between adiabatic For larger molecules we usually do not observe vibra-
and vertical ionization potentials is therefore a measure tionally structured bands (Figs. 19, 21, and 22). The num-
of the change in equilibrium geometry between ion state ber of possible vibrations that can be excited in the final
ion state increases rapidly with the size of the system,
and the superposition of the different vibrational excita-
tions leads to more or less continuous bands. For these
unstructured bands the VIP is attributed to the band maxi-
mum. Some bands show vibrational fine structure even for
larger molecules (Figs. 3 and 22). In planar unsaturated
compounds, for example, an ionization from the π system
usually couples to skeletal modes that lie near 1400 cm−1 ,
resulting in a well-developed vibrational structure. How-
ever, for the same molecules the bands that result from
ionizations from CH σ bonds are mostly broad and
unstructured.
The analysis of the vibrational fine structure can be car-
ried even further when we combine it with the assignment
of the main bands in the single-particle approximation.
From theoretical calculations, for example, the highest
occupied orbital of formaldehyde is a nonbonding orbital
(n orbital) that is mainly localized at the oxygen atom.
When an electron is removed from such an orbital the
geometry is expected to change only little, in agreement
with what we concluded from the fine structure of the first
band. A detailed analysis of this structure reveals that the
CO stretching vibration, which is 1744 cm−1 in the ground
state of the neutral molecule, is only slightly reduced to
1590 cm−1 in the lowest ion state. The second highest or-
bital is the π orbital of the CO double bond, an orbital
FIGURE 12 Schematic representation of potential curves for the that is strongly CO bonding. When an electron is removed
ground state of the initial system (M0 ) and for different ion states from this orbital, the bond is considerably weakened. This
Mk+ . should lead to an increased CO bond length, in accordance
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
Photoelectron Spectroscopy 73
with the difference between vertical and adiabatic ioniza- TABLE V Labeling Schemes Used in XPS and X-Ray
tion potential observed for the second band. The weaken- Spectroscopy
ing of the CO bond is also seen in the lowering of the CO XPS 1s 2s 2 p1/2 2 p3/2 3s 3 p1/2 3 p3/2 3d3/2 3d5/2
stretching frequency, which is reduced to 1210 cm−1 in the X-Ray K LI LII LIII MI MII MIII MIV MV
first excited state of the ion. These examples show how a
careful analysis of the vibrational structure of photoion-
ization bands and a comparison of observed vibrational
resolution, we observe a doublet in the XPE spectrum. In
frequencies to the vibrational frequencies known for the
the case of resolved doublets the individual lines are la-
molecular ground state provide information on the bond-
beled n I J . Table V compares this type of labeling with the
ing characteristics of molecular orbitals.
notation commonly used in X-ray spectroscopy.
The intensity ratio of the two doublet lines is given by
H. Multiplet Splitting
I I +1/2 /I I −1/2 = (I + 1)/I (17)
As mentioned earlier, core electrons behave in an atom-
like way, even when the respective atom is part of a The line with the higher intensity is always found at lower
molecule or a solid. As a consequence, the angular quan- binding energy (see Fig. 14). Deviations from this ideal
tum number is always a good quantum number for in- intensity ratio exist because of the effects discussed in
ner shell electrons. We can denote such an electron by Section II.E: one of the lines can lose more intensity to
its principal quantum number n and its angular quantum satellites than the other.
number I . We therefore speak of 2s, 2 p, or 3d electrons Spin–orbit coupling is less important for the valence
and correspondingly of a 2s, 2 p, or 3d hole. However, electrons because I is no longer a good quantum number.
an additional effect must be taken into account for core Only for molecules that contain at least one three-fold or
holes: the spin–orbit coupling. Since we can treat a hole higher symmetry axis, the angular momentum around this
in an otherwise completely filled shell in the same way axis can be described by a quantum number I ≥ 0. If the
as a single electron in an otherwise empty shell, we can angular momentum is nonzero, the angular motion can
attribute a spin +1/2 or −1/2 to the hole. Under the action still couple to the spin. The intensity ratio in this case is
of spin–orbit coupling, the angular momentum resulting 1:1 for the two doublet lines. In the valence region spin–
from the spin couples to the angular momentum corre- orbit interaction is observed only when at least one atom
sponding to the spatial motion of the hole to form a total with Z > 10 is present in the molecule. For light atoms, the
angular momentum J . The two possible J values that can doublet splitting due to spin–orbit coupling is too small
be formed are to be resolved in normal UPE spectra. Spin–orbit interac-
tion is also of great importance for the band structure of
J = I ± 1/2 (16)
solids.
Only J = + /2 is possible for I = 0 (s orbitals). Photoion-
1 For states that contain more than a single unpaired
ization of a core electron with angular quantum number electron (free radicals, atoms with partially filled d- or
I > 0 therefore leads to two possible final states that corre- f -shells, nh(n-1) p states, double hole states created by an
spond to the two possible J values. Because of spin–orbit Auger process), multiplet splitting can become extremely
coupling, the two final states have different energies. For complicated. Splittings of this type are observed in the
different inner shells, spin–orbit coupling decreases with XPE spectra of Co and Ni (2 p signals), Mn and Cr (3s sig-
increasing n and within the same shell it decreases with nals), and the rare earths (4s signals). Multiplet splittings
increasing I . To give an impression of the magnitude of can be used as a diagnostic tool as it distinguishes, e.g.,
spin–orbit splitting, data for xenon are provided as an ex- the spectrum of nickel(II)oxide (Figure 10b) quite conclu-
ample in Table IV. If the splitting exceeds the experimental sively from nickel(II)hydroxide, where multiplet splitting
does not occur. Auger signals usually consist of multi-
TABLE IV Spin-Orbit Splitting in Xenon ple lines, which are often only partially resolved. This can
cause considerable difficulties in the proper determination
Splitting (eV)
of Auger parameters.
n I =1 I =2
5 1.3
III. INSTRUMENTATION
4 2.0
3 61.5 12.6
A PE spectrum is measured with an instrument known as
2 321.5
a photoelectron spectrometer. Figure 13 shows a sketch of
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
74 Photoelectron Spectroscopy
A. Vacuum Requirements
Since PES measures electrons, the whole path between
the region where photoionization takes place and the de-
tector must be kept at a pressure of about 10−5 torr or
less to prevent collisions with the rest gas. For an excit-
ing radiation with hν > 11.3 eV, no material exists that is
sufficiently transparent to serve as a window. Therefore,
the path from the radiation source to the photoionization
region (PIR) must also be kept under vacuum.
Photoelectron spectrometers are very sensitive to con-
tamination of their inner surfaces. Deposition of the sam-
FIGURE 13 Scheme of a photoelectron spectrometer.
ple or its decomposition products on these surfaces can
lead to local charges that strongly influence the path of
the basic structure of this instrument, which is the same the electrons. Therefore, the vacuum must not only pro-
whether the instrument is designed for gases or for solids. vide a sufficient mean free path for the electrons but also
The technical details of the instrument, however, differ prevent contamination as much as possible. Since even
considerably depending on its application. heavy pumping does not always prevent contamination,
all vital parts of a PE spectrometer should be bakeable to
remove contamination.
Photoelectron spectrometers designed for the study of
solids require even better vacuum conditions. Because of
the surface sensitivity of PES (see Section I.E and Fig. 4),
any contamination of the sample surface contributes
strongly to the measured PE spectrum. For a gas with
a sticking coefficient of 1 (the sticking coefficient is the
probability that a particle hitting the surface will remain
on the surface), a clean surface is covered by a monolayer
if 1 Langmuir of gas is offered to the surface (1 Langmuir
is an exposure to 10−6 torr · sec). To maintain a reasonably
clean surface for 1/2 hr (about 103 sec), a vacuum better
than 10−9 torr is necessary in the collision chamber. If
well-defined adsorbates in the mono- or submonolayer
range are the object of investigation, the vacuum in the col-
lision chamber must be even better (down to 10−11 torr).
Spectrometers designed for this type of investigation are
therefore built in ultrahigh-vacuum (UHV) technology.
B. Radiation Source
The type of radiation source strongly depends on the kind
of PE spectroscopy that is performed. With UPS the com-
monly used radiation source is a resonance lamp, usually
operated with helium. The radiation emitted from this
source results from the transitions 1 P(1snp) → 1 S(1s 2 )
(compare Table VI). The radiation is produced by a high-
voltage, direct-current discharge in a capillary, a high-
current arc discharge with a heated cathode, or a mi-
crowave discharge. The helium pressure is usually a few
hundred millitorrs. Since this is higher than the pressure
maintained in the collision chamber, differential pump-
FIGURE 14 Au 4f doublet measured with different excitation ing is necessary. The lamps have an intensity of about
sources. 1012 photons/cm · sec and a half-width of 10–15 meV,
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
Photoelectron Spectroscopy 75
TABLE VI Sources of Exciting Radiation are used only for special investigations. However, some
Source Energy (eV) Width (eV) of these lines can show up in the helium discharge if the
helium is not of very high purity. Because of the lower
He I 21.22 0.001 energy of the corresponding excited states, the relative
Iα 21.2182 intensity of the impurity lines is much higher than the
Iβ 23.0848 concentration of the impurity itself.
Iλ 23.7423 For higher excitation energies, the only easily avail-
II 40.81 0.01 able monochromatic radiation comes from characteristic
IIα 40.8140 X-ray emissions. Because of the short lifetime of the in-
IIβ 48.3718 volved core hole states (10−16 sec), the half-width of the
IIλ 51.0170 emitted radiation is much larger than for the resonance
IIδ 52.2415 lines discussed above and increases rapidly with the or-
Ne Iα 16.6709 0.001 dering number. The number of applicable photon sources
16.8482 is rather large and some of the more viable examples are
IIα 26.8141 0.01 listed in Table VI. The standard materials used in most in-
26.9110 struments are aluminum and magnesium, often in a twin
Ar Iα 11.6237 anode configuration, which provide Kα radiation at 1486.6
11.8282 and 1253.6 eV, respectively. The filament is kept at ground
Y Mζ 132.3 0.5 while the anode is held at a high potential (12–20 kV).
Zr Mζ 151.4 0.8 Twin anodes have two filaments which allow changes be-
Ti Mζ 452.2 tween Mg Kα and Al Kα by simply switching the power
Mg Kα 1253.6 0.7 from one filament to the other.
Al Kα 1486.7 0.8 Figure 14 shows the 4f doublet of polycrystalline gold
Si Kα 1739.9 0.9 measured with Si, Al, and Mg Kα radiation. The intensity
Zr Lα 2042.4 1.7 obtained with characteristic X-ray lines is about two or-
Ag Lα 2984.4 2.6 ders of magnitude lower than that of the resonance lines.
Ti Kα 4510.9 2.0 High-power X-ray sources operating with a rotating anode
Cr Kα 5417.0 2.1 have been designed for this reason. To gain spectra from
Cu Kα 8047.8 2.5 deeper core levels, sources with higher energy are useful.
However, the use of such sources is not without penalties
in the form of decreasing spectral resolution and reduction
corresponding to about 100 cm−1 . The half-width, which in sensitivity for the very light elements.
determines the resolution obtainable in the spectrum, can Like the resonance lines, the X-ray lines are always ac-
be reduced by altering the operating conditions, but at the companied by satellites with intensities of up to 10% of
cost of rapid intensity loss. The numbers given above are that of the main line. This must be considered in the evalu-
a compromise between intensity and half-width. Under ation of spectra. For example, the small structures seen at
normal operating conditions about 98% of the emission the low binding energy side of the photoelectron lines in
consists of the He Iα line at 504 Å. Thus, the helium reso- Figs. 1a, 10c, and 14 are due to the satellites of the excit-
nance lamp is a source of fairly monochromatic radiation. ing radiation. In addition to the satellites, there is always
If the lamp is operated at lower helium pressure and higher a bremsstrahlung continuum underlying the characteristic
voltage, it also emits He II radiation, which is the radiation X-ray lines. This continuum can be reduced by inserting
emitted from He+ . The strongest emitted line is the He IIα a thin metal foil (the nature of which depends on the an-
line at 40.8 eV (compare Table VI), and under optimal con- ode material) between the X-ray source and the collision
ditions up to 40% of the emitted radiation is He IIα . How- chamber. The foil must be sufficiently thin and cannot be
ever, He I radiation is also present and has to be removed used to maintain a reasonable pressure difference between
by optical techniques if pure He IIα radiation is needed. the source and the chamber.
Under the conditions for optimal He II emission, the total To reduce the half-width of the characteristic X-ray
intensity of the emitted radiation is lower than the intensity lines, X-ray monochromators have been designed in which
under the standard conditions for He I operation. bent crystals are used to disperse the radiation. The method
Resonance lines from other atoms can also be used for depends on the dispersion of X-rays by diffraction, as pre-
excitation. Data for some of these lines are included in dicted by the Bragg equation
Table VI. Since there is usually more than one intense line
and since all the energies are lower than He Iα , these lines nλ = 2d sin ϕ (18)
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
76 Photoelectron Spectroscopy
Photoelectron Spectroscopy 77
Vapors from liquids usually can he measured in the under UHV conditions. Among the techniques most
same way as gases. If the vapor pressure is lower than 10−3 widely used for the preparation of suitable surfaces are
torr, the liquid must be heated. To prevent condensation
in the inlet system this should be heated, too. 1. Bombardment of the surface with rare gas ions with a
To determine a gas-phase spectrum from a sample that is kinetic energy of a few hundred to a few thousand
solid under normal conditions, the sample must be volatile. electron volts
If the volatility is fairly high, the same inlet can be used as 2. Heating of the sample, sometimes in the presence of
for liquids. For samples with low vapor pressure, a direct hydrogen gas (to reduce oxides)
inlet system can be used. A small capillary is filled with 3. Mechanical preparation of fresh surfaces within the
the sample and placed in a heatable sample holder, and UHV.
the opening of the capillary is then brought close to the
PIR. Molecules evaporating from the capillary reach the Which preparation technique is most suitable always
PIR directly, and difficulties with deposition at narrow or depends on the nature of the sample. For example, ion
cold parts of the inlet system do not arise. In addition, bombardment can change the surface constitution and
the molecules do not come into contact with heated metal heating can lead to phase transitions. To prevent contam-
parts, which often leads to catalytic decomposition. The ination of the collision chamber, the cleaning and prepa-
amount of substance needed for the measurement of a ration procedure is usually carried out in a separately
gas-phase PE spectrum is about 20 mg, and it cannot be pumped chamber. The sample is then transferred from the
recovered. preparation chamber to the collision chamber.
To study photoemission from a solid, the sample must
be brought directly to the ionization region. When the
UHV requirements are not too high (up to 10−10 torr), D. Electron Energy Analyzer
the sample can be introduced through a lock. As extreme The electron energy analyzer is the heart of a PE spectrom-
UHV conditions are required for most surface investiga- eter. It is here that the electrons are discriminated with re-
tions, a fully metal sealed vacuum system must be used. spect to their kinetic energy. The most important features
The sample is usually mounted on a manipulator that can of an electron kinetic energy analyzer (EKEA) are sensi-
be operated from the outside and is inserted through a tivity and resolution. High resolution and high sensitivity
system of two or more separate vacuum chambers. This contradict each other, and a given analyzer will always
prevents the pressure in the analysis chamber from rising be a compromise between them. As chemical shifts are
upon introduction of a sample and thus guarantees a high often small (<0.5 eV) the spectral resolution, E, of an
operational capacity. analyzer used in XPS should not exceed a few tenths of
If the sample is a conductor, it is brought in direct elec- an eV. E is mainly determined by three contributions:
trical contact with the sample holder. Metallic indium is the analyzer resolution E A , the linewidth of the exciting
often used to provide the necessary electrical and thermal radiation E X , and the intrinsic width of the photoelec-
contact between sample and sample holder. This is espe- tron line E P . If we assume Gaussian distributions for all
cially convenient because samples that are small particles three, E can be expressed as
can be pressed directly into the indium. If the sample is
an insulator, even pressing it into indium does not always 0.5
E = E 2A + E 2X + E 2P (19)
prevent charging. In this case, charge compensation can be
achieved by using a floodgun to bombard the sample with As discussed in Section III.B, E X is at best 0.3 eV in
low energy electrons. Powder samples can be mounted on XPS, when using a monochromated X-ray source, and 10
double-stick tape or pressed into a tablet, similar to the meV in UPS. To minimize the influence of the analyzer,
preparation used in infrared spectroscopy. Soluble sam- E A should by no means exceed these values. In both
ples can be dissolved and a drop of the solution brought cases, this corresponds to a relative resolution of at least
to the surface of a metallic sample holder and evaporated. 1:1000.
The latter method can yield a thin film, which is less sen- Modern EKEAs usually consist of a complex combina-
sitive to charging. tion of electron transfer lenses and a hemispherical elec-
When a sample is brought into the vacuum, its surface is trostatic deflection analyzer. Figure 16 shows a sketch of
always contaminated, usually with O2 , H2 O, CO2 , and hy- the analyzer and input lense system presently used on a
drocarbons. In many cases, the sample surface itself has a VG ESCALAB® 220iXL produced by VG Scientific® .
different chemical constitution than the bulk. The surface The input lens captures the maximum possible number of
of most metals, for example, consists of oxides. In nearly emitted phototelectrons and transfers them to the deflec-
all cases, the sample surface must therefore be cleaned tion analyzer.
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
78 Photoelectron Spectroscopy
Photoelectron Spectroscopy 79
80 Photoelectron Spectroscopy
analyzing weak features in the spectrum. Most modern G. Instrumentation for Angle-Resolved
instruments use a sweep technique in which the investi- Photoelectron Spectroscopy
gated energy range is measured within a few seconds, and
Instruments designed for angle-resolved measurements
the measurement is repeated at intervals. The data from
must meet two specific requirements: (1) The analyzer
each sweep are accumulated and stored in a data handling
must be movable around the ionization region and (2) the
system. Well-designed sweep programs also take care of
acceptance angle at the first input lens of the analyzer
sensitivity differences between individual channeltrons in
system must be small enough (usually a few degrees) to
the array. The development of the spectrum can be watched
achieve angular resolution. The requirements for an angle-
on a screen, and the measurement can be terminated
resolving spectrometer differ depending on whether the
when the signal-to-noise ratio has reached an acceptable
sample is a gas or a solid.
limit.
The angular distribution for atoms or molecules in the
For spin-resolved PES (Section I.H), the detector must
gas phase is fully described by the asymmetry parameter
be sensitive to the direction of the electron spin. Such a
β (Section I.F). To measure β, the angle θ between the
detector was constructed in 1943 by Shull et al. based on
incoming radiation and the outgoing electrons must be
the theoretical prediction by Mott, that the scattering of
varied [Eq. (6)]. It is sufficient for this type of measurement
an electron from a nucleus with a high atomic number Z
to move the analyzer in one plane.
depends on the direction of the incident electron’s spin, if
For solid samples there are three degrees of freedom: the
the electron has an energy greater than 50 keV and is scat-
angles
, θ, and φ defined in Section I.G. If the incident
tered through an angle greater than 90◦ . This Mott detector
radiation is polarized, an additional degree of freedom is
has been widely used to measure the spin polarization of
obtained because of the orientation of the polarization vec-
electrons. In the 1980s, Mini Mott detectors were real-
tor. Angle-resolved measurements with unpolarized radi-
ized by the use of cylindrical or sperical retarding optics.
ation can be carried out with an instrument in which the
Among them, the SPLEED and the low-energy diffusive
analyzer can be moved in one plane, provided that the sam-
scattering detector are of practical importance. The lat-
ple is mounted on a manipulator that allows adjustment of
ter is preferable because the SPLEED demands frequent
the direction of the surface normal. With this setting
,
cleaning of the target crystal under UHV. The low-energy
θ , and φ can be varied over a fairly wide range. However,
diffuse scattering detector is based on the spin–orbit inter-
a better range for all three angles is achieved if the ana-
action between an incident electron and a polycrystalline
lyzer can be moved in two planes that are perpendicular
gold film that is used as scattering target, because gold
to each other. Because of the additional degree of freedom
is not chemically reactive and exhibits strong spin-orbit
obtained with polarized radiation, it is necessary to use an
interaction due to its high atomic number Z . According to
instrument that allows movement of the analyzer in two
the Ls term of the spin–orbit interaction, the spin analyzer
dimensions in order to obtain all the available information.
is sensitive to the transverse component of the spin polar-
Some limited information on the angular distribution
ization which is normal to the scattering plane, while the
of the electrons can be derived even with an instrument in
longitudinal component along the incident electron direc-
which the angle between excitation source and analyzer is
tion cannot be observed.
fixed: if the sample is mounted on a suitable manipulator,
it is possible to vary the orientation of the analyzer to
F. Magnetic Shielding the surface normal while keeping the angle of incident
radiation constant.
Low-energy electrons are very sensitive to the stray mag-
netic fields that exist in any laboratory, such as the mag-
netic field of the earth. These stray fields strongly influence IV. SOME EXAMPLES
the trajectories of the traveling electrons. As a result, all
parts of a PE spectrometer between ionization region and In this section we will discuss a few examples of the appli-
detector must be carefully shielded. This can be done with cation of PES in different fields of research. The number
Helmholtz coils, which compensate the external fields, or of examples is too limited to allow more than a glimpse
with magnetically shielding materials like mu-metal. The onto the broad and highly active field of PES. Further ex-
latter type of shielding is used in most modern instruments, amples and other types of application can be found in the
because shielding with Helmholtz coils is difficult for the references cited in the bibliography. Current work in PES
dynamic stray fields produced by many types of labora- is published in a variety of scientific journals. A journal
tory equipment. Improper shielding leads to reduced sen- specifically devoted to this field is the Journal of Elec-
sitivity and resolution, as well as to asymmetric skewed tron Spectroscopy and Related Phenomena, published by
lines. Elsevier, Amsterdam.
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
Photoelectron Spectroscopy 81
FIGURE 20 Orbital diagram for stilbene: (a) ordering as in trans-stilbene; (b) ordering as in trans- p, p -dimethoxystilbene.
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
82 Photoelectron Spectroscopy
Photoelectron Spectroscopy 83
FIGURE 22 UPE spectra of cyclohexane, cyclohexene, and 1,4-cyclohexadiene. [From Bischof, P., et al. (1969).
Helv. Chim. Acta 52, 1745.]
the electron must escape into the vacuum, and in going surface (see inset in Fig. 25). A variation of the polar angle
from the interior of the crystal through the surface into the θ selects different parallel components
vacuum only the parallel component k of the momentum −1/2
is conserved and not the component perpendicular to the k = 2m E kin π 2 sin θ (23)
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
84 Photoelectron Spectroscopy
Photoelectron Spectroscopy 85
86 Photoelectron Spectroscopy
D. Investigation of Adsorbates
When a molecule is adsorbed at the surface of a solid it FIGURE 28 Comparison of C 1s and O 1s spectra for free
can be either physisorbed or chemisorbed. We speak of CO, W(CO)6 , and CO adsorbed on W(110) and Cu(100). [From
“physisorption” when the bonding is mainly caused by Freund, H. J., and Plummer, E. W. (1981). Phys. Rev. B 23, 4859.]
van der Waals interactions and of “chemisorption” when
we have some type of a chemical bond between adsorbate
and substrate. However, there is a more or less continuous contribute strongly to the understanding of the chemisorp-
transition between both types of bonding. The nature of tion bond. Figure 28 shows an example. The C 1s and O
the bonding between adsorbate and substrate is very im- 1s spectra of gaseous CO are shown at the bottom. Weak
portant for the understanding of reactions that take place satellite structures at the high binding energy side of the
in heterogeneous catalysis. Catalytic processes of this type main lines are blown up in the figure in order to reveal
have great technical importance but often they are not fully their structure. The indicated satellites are connected with
understood on a molecular level. Because of its extreme final ion states in which, in addition to the removal of a
surface sensitivity, PES is a powerful tool in achieving core electron, a second electron is excited from the π to
such an understanding. The applications of PES to the the π ∗ orbital. When CO is adsorbed on the 110 surface
study of adsorbates are too widespread to be reviewed of tungsten, a new satellite is observed closer to the main
here. We only show two examples to give an impression line. This new satellite increases strongly in intensity with
of how these studies can be carried out. decreasing strength of the chemisorption bond. For the
For a variety of adsorbate systems such as the adsorption system CO on copper (100) (at the top in Fig. 28), where
of small molecules (CO, N2 , H2 O, etc.) on transition metal CO is only weakly bound, this new satellite becomes very
surfaces, a “molecular view” of the bonding often provides strong. What is the nature of this satellite and what does
a useful description. In a molecular view we do not look it tell us about bonding? The fact that the same satellite as
at the substrate as a metal with band structure and all the observed for CO on W(110) is also observed in the metal
typical solid-state properties. Instead, we look primarily complex W(CO)6 reveals that this satellite cannot result
at a few metal atoms to which the adsorbed molecule is from the special properties of metallic tungsten. It must
bound. Such a view immediately connects the bonding in be related to the local bonding between CO and a tungsten
an adsorbate to the bonding in a metal complex. Thus, atom. The satellite is connected with a final state where, in
studying adsorbates together with related complexes can addition to the core ionization, an electron is transferred
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
Photoelectron Spectroscopy 87
from an occupied d orbital of the tungsten atom to the π ∗ two-dimensional band structure with the methods outlined
orbital of CO (charge transfer excitation). A detailed anal- in Section IV.B and its comparison to theoretical results.
ysis reveals that the intensity of the satellite is a measure
of the “backbonding” caused by the overlap of occupied
E. Studies in Corrosion
d orbitals of the metal atom with empty π ∗ orbitals of
the ligand. The similarities and differences observed in Corrosion is of tremendous economic importance. In the
the satellite structure of the C 1s and O 1s photoemission United States, for example, a total damage of more than
line can be used to derive additional information on the $300 billion is caused every year by corrosion. Among the
chemisorption bond. important corrosive reactions are all kinds of oxidation
The second example concerns well-ordered adsorbate reactions, anodic dissolution, passivation, breakdown of
layers on single crystal surfaces. A wide variety of such passivation, and absorption of hydrogen, usually the first
layers has been investigated with metals and metal oxides step of hydrogen embrittlement. All these reactions take
as substrates. The three most important questions one tries place in the outermost layers of the material, but many of
to answer in these studies are the following: them are not well understood on an atomic or molecular
level. The surface science approach to corrosion tries to
1. What is the crystallographic structure of the adsorbate clarify the mechanisms of corrosion reactions on that level
overlayer? and to link this understanding to the macroscopic manifes-
2. What are the adsorption sites (on top of a surface tations of corrosion. PES as a highly surface sensitive tech-
atom, between two or more, etc.)? nique became an important tool in this area of research.
3. How are the individual molecules adsorbed (standing, In most cases, the surface of a metal or an alloy, and
lying, tilted)? often also of a polymer, differs from the bulk. This is due
to spontaneous reactions with the environment like oxida-
The first question is usually answered by LEED (low- tion, which can lead either to passivation or to ongoing de-
energy electron diffraction) but for the other two ARPS is struction. In alloys, which account for the majority of tech-
the method of choice. Figure 29 shows the result of such a nically important materials, the composition of the surface
study: a dense monolayer of CO on a Ni(110) surface. The often differs from the bulk composition to dealloying or to
layer has p2mg symmetry. Each CO molecule is bound in segregation processes that accompany surface reactions.
a bridging position between two neighboring Ni atoms The surface zone, which is important for passivation or
along the [11̄0] direction. The C-terminus points down to corrosion, often consists of only a few atomic layers. In
the Ni surface. The CO molecules are tilted by (17 ± 2)◦ these cases, PES can contribute significantly to the investi-
along the [001] direction in order to avoid the close in- gation of the overlayer. The key is that for most elements,
termolecular contact along [11̄0], which would occur if oxides are easily distinguished from the element itself.
the molecules were oriented perpendicular to the surface. Angular dependent measurements do not only allow a de-
Most of these details resulted from the determination of a termination of the thickness of the oxide layer: because of
the element specificity of PES, such measurements also
can be used to study the change of composition perpen-
dicular to the surface.
Figure 30 shows an example. The sample is an
iron/chromium alloy that had been exposed to an aque-
ous solution of sulfuric acid. The alloy is covered by a
surface layer of approximately one nanometer that con-
sists of oxides and hydroxides of iron and chromium. The
composition of the material immediately underneath the
passive film equals the bulk composition. This is not al-
ways the case, as elements are often depleted underneath
the passive film if an enrichment of these elements occurs
in the film. Technical problems with this kind of investi-
gation mainly concern the transfer of the corroded sample
into the vacuum of the spectrometer. Passive layers that
contain hydroxides in solution especially tend to degrade
FIGURE 29 Schematic representation of CO adsorbed in a
(2 × 1) p2mg overlayer on a Ni (110) surface. [From Freund, H.-J., in the vacuum.
and Kuhlenbeck, H. (1995). In “Applications of Synchrotron Radi- As mentioned at the beginning of this section, these few
ation” (Eberhardt, W., ed.), Springer-Verlag, Berlin, Heidelberg.] examples can only provide a glimpse of a very broad field.
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
88 Photoelectron Spectroscopy
Photoelectron Spectroscopy 89
core line, this procedure allows the imaging of the lateral cuprates. Recent experiments on the Bi-based materials
distribution of a single element within a multicomponent indicated that a description of the normal state as a Fermi-
surface. Important areas of research that may benefit from liquid with well-defined quasiparticles is not adequate. In
these kinds of experiments include studies of impurity seg- contrast, below Tc such well-defined quasiparticles appear
regation at grain boundaries, Schottky barrier formation to exist.
on semiconductors, and diffusion. PEEM is also expected Further PES investigations have been devoted to the
to gain importance in studies of the complex surface struc- technically important question of corrosion stability. Wet
tures of highly integrated circuits. air and CO2 lead to an increase of hydroxides or carbonates
There are two new fields of PES research that have and a depletion of Cu in the surface layers. The depth
gained increasing interest during the last decade: Clus- of the corroded, nonsuperconducting surface zone ranges
ters and high Tc superconductors. A cluster consists of n from a few nanometers up to several tens of nanometers,
atoms or molecules that stick together to form a kind of depending on the nature of the material and on the manner
supermolecule. Progress in preparation techniques now in which the sample is prepared (polycrystalline material,
allows the study of ionic and, more recently, neutral clus- single crystals, or epitactically grown films).
ters with a well-defined number of atoms or molecules.
Investigation of a series of such clusters with increasing n
yields information on the development of a given physi- SEE ALSO THE FOLLOWING ARTICLES
cal property from the single atom or molecule to the solid
state. PES is especially suited for examining the devel- ADSORPTION • ATOMIC SPECTROMETRY • AUGER ELEC-
opment of the outermost orbitals of the free atom or free TRON SPECTROSCOPY • MICROWAVE MOLECULAR SPEC-
molecule into the valence band of the corresponding solid. TROSCOPY • MÖSSBAUER SPECTROSCOPY • MULTIPHO-
Pronounced stabilities for certain values of n are usually TON SPECTROSCOPY • RADIATION SOURCES • SURFACE
found in small clusters. PES has become extremely help- CHEMISTRY • VACUUM TECHNOLOGY • X-RAY ANALY-
ful in explaining these stabilities. For elements that are SIS • X-RAY, SYNCHROTRON RADIATION, AND NEUTRON
metals in the solid state, it is of special interest to deter- DIFFRACTION
mine the number of atoms that are necessary to develop
metallic properties. In the case of mercury, for example,
this problem was successfully studied by PES.
BIBLIOGRAPHY
The discovery of the so-called high Tc super-
conductors with critical temperatures up to 100 K
Asami, K., and Hashimoto, K. (1987). Langmuir 3, 897.
(such as YBaCu3 O6+x and La2−x Srx CuO4 and Bi2 Sr2 Baalmann, A., et al. (1985). Solid State Commun. 54, 583.
Cax−1 Yx Cu2 O8 ) immediately prompted a variety of PES BESSY II (1986). “Eine optimierte Undulator/Wiggler-Speicherring
investigations. In the meantime, electron spectroscopy Lichtquelle für den VUV- und XUV-Spektralbereich,” p. 20, BESSY,
turned out to be the key technique to understand the Berlin.
electronic structure of these cuprates. These experiments Bischof, P., et al. (1969). Helv. Chim. Acta 52, 1745.
Briggs, D., and Seah, M. P., eds. (1983). “Practical Surface Analysis by
are, however, extremely difficult because of the surface Auger and X-Ray Photoelectron Spectroscopy,” Wiley, New York.
sensitivity of PES and the fact that the surface of these Cardona, M., and Ley, L., eds. (1978). “Topics in Applied Physics,”
materials usually consists of a few layers of noncon- Vol. 26, “Photoemission in Solids I,” p. 193, Springer-Verlag, Berlin.
ducting material, which, due to segregation phenomena, Eberhardt, W., ed. (1995). “Application of Synchrotron Radiation,”
exhibit a chemical constitution different from the bulk. Springer-Verlag, Berlin, Heidelberg.
Eckardt, H., Fritsche, L., and Noffke, J. (1983). J. Phys. F 14, 97.
The most reliable experiments have been performed on Eland, J. H. D. (1984). “Photoelectron Spectroscopy,” 2nd ed. Butter-
the Bi-based materials that have a natural cleavage plane worths, London.
with an unreconstructed surface. The electronic structure Freund, H.-J., and Kuhlenbeck, H. (1995). In “Applications of Syn-
derived from the PES studies revealed the importance chrotron Radiation” (W. Eberhardt, ed.), Springer-Verlag, Berlin, Hei-
of the CuO2 planes which appear in all these materials. delberg.
Freund, H.-J., and Plummer, E. W. (1981). Phys. Rev. B 23, 4859.
The compounds can be described as doped Mott insula- Heimann, P., Miusga, H., and Neddermeyer, H. (1979). Solid State Com-
tors. The highest occupied valence band in the insulating mun. 29, 463.
parent compounds have in-plane O 2 p character, so that ICES (1997). “Proceedings of the 7th International Conference on Elec-
the systems are charge transfer insulators in the so-called tron Spectroscopy,” ICES 7, Chiba, Japan, September 8–12, Elsevier.
Zaanen–Sawatzky–Allen classification scheme. Further Jung, C., and Bressler, P. R. (1996). Electron Spectrosc. 78, 503.
Kevan, S. D., ed. (1992). “Angle Resolved Photoemission, Theory and
important information came from ARPES. At present the Current Applications,” Elsevier.
most pertinent questions concern in-plane anisotropies re- Plummer, E. W., et al. (1977). Phys. Rev. A 15, 2339.
lated to the unusual normal state properties of the doped Siegbahn, K. (1974). J. Electron Spectrosc. 5, 34.
P1: GLQ Final pages
Encyclopedia of Physical Science and Technology EN012C-568 July 26, 2001 15:32
90 Photoelectron Spectroscopy
Siegbahn, K., et al. (1971). “ESCA Applied to Free Molecules,” Elsevier, niques,” Wiley, Chichester.
Amsterdam. Watts, J. F. (1994). “X-ray photoelectron spectroscopy,” Vacuum 45
Sodhi, R. N., and Cavell, R. G. (1983). J. Electron Spectrosc. 32, 283. (6/7), 653.
Sokell, E., et al. (1998). J. Electron Spectrosc. 94, 107. Wagner, C. D., et al. (1979). “Handbook of XPS,” Perkin-Elmer Corp.,
Vickerman, J. C., ed. (1997). “Surface Analysis—The Principal Tech- Eden Prairie, MN.
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
Raman Spectroscopy
R. P. Van Duyne
C. L. Haynes
Northwestern University
845
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
µind = α E between the induced dipole µind vector and dent light is scattered. Although most of the scattered light
the electric field E in linear Raman scattering. αρσ is the has the same wavelength as the incident radiation, a small
polarization element where ρ and σ are the Cartesian part of it occurs at different wavelengths. The scattering
coordinates x, y, or z. of light at different wavelengths is called Raman scatter-
Raman microscope Combination of microscope optics ing, after the Indian scientist Sir C. V. Raman, who, with
and a Raman spectrometer equipped with a charge- K. S. Krishnan, first reported the phenomenon in liquids
coupled device camera that enables the observer to view in 1928. In the same year Landsberg and Mandelstam in
a microscopic object according to its Raman-active Russia reported a change in frequency of the light scat-
vibrations. tered from quartz. The physical origin of Raman scattering
Resonance Raman scattering The intensity of Raman lies in inelastic collisions between the molecules compos-
scattering is amplified by 103 –105 times when the ing the system (e.g., the liquid) and photons, the particles
wavelength of the incident light is coincident with an composing the light beam. “Inelastic collision” means that
intense molecular electronic absorption transition. there is an exchange of energy between the photon and the
SPP-Raman The intensity of Raman scattering is en- molecule and a consequent change in energy, and hence
hanced by 102 –103 times when a smooth Ag or Au wavelength, of the photon. Moreover, since total energy is
surface is excited by incoming radiation. conserved during the scattering process, the energy gained
Stokes Raman scattering Light scattering in which the or lost by the photon must equal the energy change within
photons lose energy as a result of photon–molecule the molecule. By measuring the energy gained or lost by
collisions. the photon, changes in molecular energy can be probed.
Surface-enhanced hyper-Raman scattering (SEHRS) The changes are usually related to rotational, electronic,
The intensity of the two-photon Raman scattering is and vibrational energy levels. Since much of present-day
enhanced by 1011 –1012 times from molecules (i) ad- Raman spectroscopy is concerned with vibrational energy
sorbed at a Ag, Cu, or Au surface roughened on the levels, thisarticle will concentrate on vibrational proper-
nanometer scale; and (ii) having an electronic absorp- ties. However, the theoretical principles and experimental
tion band at or near the excitation wavelength. techniques enunciated for the vibrational states are readily
Surface-enhanced Raman scattering (SERS) The in- extended to other types of molecular properties.
tensity of Raman scattering is amplified by 106 –108
times from molecules adsorbed at a Ag, Cu, or Au sur-
face roughened on the nanometer scale. I. CLASSICAL MODEL OF
Surface-enhanced resonance Raman scattering RAMAN SCATTERING
(SERRS) The intensity of Raman scattering is
amplified by 109 –1011 times from molecules adsorbed Light is a traveling wave of electric and magnetic fields,
at a Ag, Cu, or Au surface roughened on the nanometer of which only the electric component gives rise to Raman
scale. scattering. When a light wave meets a molecule consisting
Surface Plasmon Polariton (SPP) Evanescent, p- of electrons and nuclei, the electric field of the wave at any
polarized, nonradiative electromagnetic wave associ- instant will be the same throughout the molecule because
ated with the propagation of a charge wave at a smooth the molecule, perhaps 1 nm in size, is small compared to
metal–dielectric interface. the wavelength of the light, which is typically 500 nm.
Time-resolved resonance Raman spectroscopy (TR3 S) Thus, the field will exert the same force on all electrons
Study of transient chemical species or excited elec- in the molecule and will tend to displace them from their
tronic states via their Raman spectra, often with pulsed average positions around the positively charged nuclei.
lasers and multichannel photon detectors. For the Raman process, the displacements result in an
Ultraviolet resonance Raman scattering (UVRRS) Se- induced dipole moment, µind in the molecule that is, to
lective excitation of a UV resonance enhanced Raman- a good approximation, proportional to the electric field
active mode of vibration. strength E.
Thus
µind = α E, (1)
RAMAN SPECTROSCOPY is a technique that uses
scattered light resulting from photon–molecule collisions where the proportionality factor α is called the electric po-
to investigate molecular properties. When a monochro- larizability of the molecule. In general, the vector µind will
matic light beam is incident on systems such as trans- have a different direction from the vector E, and therefore,
parent gases, liquids, or solids, most of it is transmitted α is not a simple scalar quantity. In fact, the magnitudes of
without change. However, a very small portion of the inci- the three components defining the induced dipole moment,
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
transitions) that gives the Raman effect its importance in ings between the vibrational levels contained within each
the study of molecules. state. Moreover, for the vibrational energy E vib of a di-
The information about molecular vibrational frequen- atomic molecule it is a good approximation to write
cies provided by infrared absorption spectroscopy is of
E vib = v + 12 hνvib v = 0, 1, 2, . . . , (10)
the same kind as that provided by the νvib values of the
Raman lines. Moreover, for molecules with little or no where the vibrational quantum number v has only integral
symmetry, the infrared and Raman spectra can, in some values, so the vibrational energy levels in the ground state
cases, have quite similar appearances. However, infor- in Fig. 1 are equally spaced by the amount h νvib .
mation from both techniques is important for molecules In the quantum mechanical model, light scattering is
having a center of symmetry, since there can be comple- depicted as a two-photon process. The first step in this
mentarity between the Raman and infrared spectra with process is the combination of a photon and a molecule
no correspondence between the main features exhibited to raise the molecule to a higher-energy, short-lived state.
in each spectrum. The complementarity derives from the This state is reached by the upward arrows in Fig. 1 and,
rule of mutual exclusion, which states that for molecules as shown in the figure, may or may not correspond to
with a center of symmetry, vibrational transitions that are a quantized energy state of the molecule. The second
allowed in the infrared are forbidden in the Raman ef- step, indicated by the downward arrows in Fig. 1, in-
fect and vice versa. The reason for the complementarity volves the release of a photon after a very short time in-
of Raman and infrared spectra lies in the different natures terval (<10−14 sec). The energy of this second photon is
of the physical processes involved in the two effects. The given by the length of the downward arrows in Fig. 1.
Raman process is a scattering effect involving an induced For Rayleigh scattering the upward and downward tran-
dipole µind , which in turn depends on a change in molec- sitions have the same length and have therefore, apart
ular polarizability during a vibration. In contrast, infrared from a change in sign, the same energies. Thus, in the
spectroscopy is an absorption process caused by a change Rayleigh process, no change in frequency of the photon
in the permanent molecular dipole µ with change in bond occurs.
length during a vibration. The various kinds of Raman processes can now be out-
lined. If the downward arrow stops on a vibrational energy
level that is higher than the starting level, a Stokes pro-
II. QUANTUM MECHANICAL MODEL cess has occurred. In this case, the second photon has a
OF RAMAN SCATTERING frequency ν0 − νvib corresponding to the third term in Eq.
(8) of the discussion of the classic model for the Raman
The quantum mechanical approach to the scattering pro- effect. Conversely, an anti-Stokes process results from the
cess is quite different from the classic model: the wave– transition terminating at a vibrational energy level lower
particle duality of a light beam is incorporated by consid- than the starting level. In the anti-Stokes process, the sec-
ering that the beam is made up of packets or quanta of light ond photon has a frequency ν0 + νvib , giving the same
particles known as photons. Moreover, the quantization of result as the second term in Eq. (8). Of course, in both
molecular energy levels is taken into account and a means processes total energy is conserved, so for Stokes scat-
is provided for calculating the polarizability α, and thus tering the molecule gains a quantum of energy, hνvib ,
Raman intensities, in terms of the electronic properties of while for anti-Stokes scattering the reverse is true. For
a molecule. both Stokes and anti-Stokes processes a selection rule can
It is a reasonable approximation, for a gaseous diatomic be derived from Eq. (10) which says that ν , in Fig. 1, can
molecule, to write the molecular energy E max as a sum of only change by ±1. Thus Eq. (8) derived from the clas-
terms: sic model agrees with the results obtained by considering
E max = E elec + E vib , (9) quantized energy levels in that both models predict that the
difference in frequency between the incident and scattered
where the respective subscripts refer to electronic and light, ν0 − (ν0 − νvib ), corresponds directly to the molec-
vibrational components of the total energy. For present ular vibrational frequency νvib . The quantum mechanical
purposes the contributions due to molecular rotation and model also illustrates an important generalization, namely,
translation can be ignored. Electronic energy transitions that the position of Raman peaks is a property solely of
involve much larger quantities of energy than vibrational the electronic ground state. This follows from the fact that
transitions,with values of 10,000 to 50,000 cm−1 for the hνvib is the energy change in a vibrational transition within
former and 10 to 4000 cm−1 for the latter. This situation the lower or ground electronic state in Fig. 1.
is depicted in Fig. 1 with a large energy spacing between In the classical model Eq. (8) indicates no differ-
the ground and excited electronic states and smaller spac- ence in the expected intensities of Stokes and anti-Stokes
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
FIGURE 1 Some of the possible consequences of a photon–molecule interaction. The lengths of the upward-pointing
arrows are proportional to the frequencies of the incoming light, and the lengths of the downward-pointing arrows
are proportional to the frequency to the scattered (or in the case of flurescence, emitted) light. The energy spacing
between the ν levels is proportional to νvib (E = hνvib ).
transitions, since the coefficients of the two terms in the higher energy electronic state shown in Fig. 1. Consider
equation are the same. the situation when the photon energy is approaching that
However, the model of quantized energy levels depicted of the energy gap between the lower and higher electronic
in Fig. 1 shows that for anti-Stokes transitions to take place states. A preresonance Raman transition is due to a light
the molecule must be in a higher (ν > 0) vibrational state frequency that has almost enough energy to produce direct
within the electronic ground state. Since the population electronic absorption by the molecule. Under this condi-
of vibrational levels is governed by a Boltzmann distribu- tion the intensity of Raman scattering shows a marked
tion, only a small percentage of molecules are in higher increase. For normal or nonresonant conditions Raman
vibrational states. The ratio of the numbers of molecules intensities are proportional to the fourth power of the scat-
in the ν = 1 and ν = 0 vibrational states in the ground tered light frequency νs . However, as preresonance Raman
electronic state (Fig. 1), which we will denote N1 and N0 , conditions are approached the intensity of scattering goes
respectively, is given by up much more rapidly than νs4 .
A slight increase in the energy of the exciting radia-
N1 hνvib
= exp − , (11) tion over that for the preresonance case will place the
N0 kT upward transition in Fig. 1 within the higher electronic
where h is Planck’s constant, T the absolute temperature, state. Absorption of a photon can now occur and by the
and k the Boltzmann constant. When T = 300 K and the prompt reemission of a second photon can give rise to the
vibrational frequency is 480 cm−1 , N1 is 0.1N0 . As a result resonance Raman process. Band intensities in resonance
of the exponential nature of Eq. (11) an anti-Stokes line Raman spectra can be orders of magnitude greater than
at 3 × 480 cm−1 , or 1440 cm−1 , would be 0.001 times those in normal Raman spectra. The reason for this can be
as strong as the corresponding Stokes line. In practice seen by considering the quantum mechanical expression
this means that the feeble anti-Stokes scattering is usually for Raman intensities.
ignored in conventional Raman spectroscopy and only the In the quantum mechanical treatment for randomly ori-
Stokes spectrum is recorded. ented molecules, the total intensity of the scattered light
In normal Raman scattering, the energy of the incident resulting from a molecular transition between states m and
light is considerably less than that needed to reach the n is
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
FIGURE 4 Schematic diagram of a Raman spectrometer showing the options of a single monochromator and mul-
tichannel detection.
intensified photodiode array (IPDA) detectors. A CCD small number of incoming wavelengths centered at the
is a multichannel detector made up of large arrays of laser resonance. Holographic notch filters are inserted af-
individual metal–oxide–silicon capacitors. All the Raman ter the Raman light is scattered from the sample in order
lines are registered on different elements of the detector to filter the Rayleigh line from the detected light. These
at the same time. Thus, it is possible to observe the entire filters are especially important in single monochromator
Raman spectrum on a television screen or an oscilloscope detection schemes where the Rayleigh line cannot be re-
in real time. When using multichannel detection, the moved as it is in double or triple monochromators.
grating is turned only to change the spectral region across The recent development of the liquid crystal tunable
the detector. interference filter (LCTF) has provided significant perfor-
Recent advances in photodiode technology have pro- mance advantages over nontunable systems. The LCTF
duced a new detector for high-sensitivity, low-light de- is useful for high-definition Raman imaging because the
tector applications in which PMTs, and even CCDs, are color of the transmitted light can be electronically con-
unsuitable. This ultrasensitive device is known as an trolled from the visible to the near infrared wavelengths
avalanche photodiode (APD) and is based on standard of the spectrum. When a LCTF is paired with a CCD de-
photodiode technology. The extra sensitivity of the APD tector, the result is a multispectral, high-resolution image.
is gained because a high bias potential accelerates elec-
trons to ionize semiconductor atoms. These semiconduc-
VI. BASIC EXPERIMENTAL TECHNIQUES
tor atoms trigger the release of free electrons which are
then accelerated to continue the cycle in an “avalanche
Though normal Raman spectroscopy is a very selec-
effect.” This effect gives rise to thousands of electrons for
tive technique for chemical analysis, there are some se-
each initial impinging photon.
rious experimental disadvantages related to the sensi-
tivity, large fluorescence interference, and lack of time
resolution of the technique. These weaknesses have been
D. Filters
addressed in the creation of new Raman-based techniques.
As mentioned in previous sections, optical filters play an The weak Raman signals due to inherently small Raman
important role in Raman spectrometers. An interference scattering efficiencies has been addressed by resonance
filter is usually inserted before the laser beam reaches the Raman, surface-enhanced Raman and SPP-Raman tech-
sample in order to improve the monochromatic charac- niques. Fourier transform-Raman spectroscopy and con-
ter of the wavelength. Interference filters block all but a focal Raman microscopy address the disadvantage of
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
FIGURE 5 Schematic depiction of nonresonant and resonant impinging radiation on a metallic roughness feature.
The resonant condition demonstrates the concept of the LSPR.
fluorescence interference. Time-resolved resonance Ra- length strikes a metallic roughness feature, the plasma of
man spectroscopy allows kinetic studies of transient reac- conduction electrons will oscillate collectively. Because
tion intermediates. Nonlinear Raman techniques add yet this collective oscillation is localized at the surface of this
another dimension to Raman spectroscopy; these tech- plasma of electrons, it is known as a localized surface plas-
niques are addressed in Section VII. mon resonance (LSPR). The LSPR allows the resonant
wavelength to be absorbed and scattered, creating large
electromagnetic fields around the roughness feature. The
A. Surface-Enhanced Raman Spectroscopy LSPR is depicted in Fig. 5. If a molecule is placed within
Though Raman spectroscopy gives detailed chemical and the electromagnetic fields, an enhanced Raman signal is
conformational information, the small scattering inten- measured. Current estimates suggest that these electro-
sity has prohibited widespread use. Experiments done magnetic fields reach further than 65 nm from the metal
in the mid 1970s demonstrated that molecules adsorbed surface.
to a roughened metal surface generated anomalously The discovery and understanding of SERS was impor-
large Raman intensities. This phenomenon is known as tant not only because it made Raman a more viable an-
surface-enhanced Raman spectroscopy (SERS). SERS en- alytical method but also because it introduced the con-
hancements of greater than 106 are seen regularly when cept of surface-enhanced spectroscopies in general. With
molecules are adsorbed to roughened silver, gold, and cop- the SERS precedent, surface-enhanced resonance Raman
per surfaces. Enhanced signals have also measured from spectroscopy (SERRS) and surface-enhanced hyper-
molecules on lithium, sodium, potassium, cesium, rubid- Raman spectroscopy (SEHRS) have both been discovered
ium, aluminum, gallium, indium, platinum, and rhodium. and put to use as analytical tools. In fact, enhancement
Today, the SERS community generally agrees that the factors as large as 1012 have been measured in SEHRS
106 enhancement factor can be attributed to the combina- experiments (see Section VII.B). This immense enhance-
tion of two separate enhancement mechanisms, the chem- ment was only recently surpassed by the 1014 enhancement
ical enhancement mechanism (CHEM) and the electro- measured in single molecule SERS (see Section X).
magnetic enhancement mechanism (EM).
The CHEM enhancement mechanism, thought to con-
B. Surface-Enhanced Resonance
tribute an enhancement factor of 102 , asserts that a charge-
Raman Spectroscopy
transfer state is created between the metal and adsorbate
molecules. This mechanism is site-specific and analyte Surface-enhanced resonance Raman scattering (SERRS)
dependent. The molecule must be directly adsorbed to is a combination of the aforementioned CHEM and EM
the roughened surface in order to experience the CHEM enhancement mechanisms with fulfillment of a resonance
enhancement. condition. This resonance condition is satisfied when an
The EM enhancement mechanism contributes greater incoming photon instigates a molecule’s electric dipole
than 104 times enhancement over normal Raman scat- allowed transition into an excited state, rather than an in-
tering. In order to understand the EM enhancement, one termediate state (see Fig. 1). In a standard SERRS exper-
must consider the size, shape, and material of the surface’s iment, the molecules are first adsorbed to a Ag, Cu, or Au
nanoscale roughness features. If the correct laser wave- substrate, and then, the absorption spectrum is measured.
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
A laser wavelength close to the molecule’s electronic ab- giving rise to fluorescence by irradiating the sample
sorption band is chosen as the Raman excitation wave- with near infrared (NIR) wavelengths rather than visi-
length. When the Raman spectra are captured, the signals ble wavelengths. Until the development of Fourier trans-
are 109 –1011 times greater than normal Raman scattering form (FT)-Raman in 1986, this solution was not imple-
signals. This large enhancement factor is attributed to the mented because longer excitation wavelengths translate
cumulative effects of the >106 SERS enhancement and into smaller Raman peaks intensities by the 1/λ4ex de-
the 103 –105 resonance Raman enhancement. pendence of the Raman scattering intensity. For example,
using a 1064 nm Nd:YAG laser instead of a 514.5-nm
Ar+ laser produces 18 times less Raman scattering in-
C. SPP-Raman
tensity. FT-Raman makes use of the high throughput and
Just as a standing wave is generated when a SERS active wavelength multiplexing features of a Michelson interfer-
surface with nanoscale roughness is irradiated, a prop- ometer in order to compensate for the excitation wave-
agating wave is generated when a smooth SERS-active length dependence loss. Current FT-Raman instruments
surface is irradiated. This traveling wave is known as a typically use a diode-pumped, air-cooled cw Nd:YAG
Surface Plasmon Polariton (SPP). The SPP can be located laser source, a broad-range quartz beamsplitter, gold-
by measuring the Attenuated Total Reflectance (ATR) of coated optics, and an InGaAs or Ge diode detector. The
a thin film surface; the SPP is seen as a deep minimum in spectral range for a modern commercial FT-Raman in-
the reflectivity versus incident angle dependence when the strument is 3600 − 70 cm−1 Stokes shift and −100–
surface is excited by p-polarized light. SPP measurements −2000 cm−1 Anti-Stokes shift. FT-Raman advantages in-
give information about the thickness and refractive index clude fluorescence elimination and accuracy in frequency
of ultrathin films. measurements.
Even if two systems are found to be optically identical In the FT-Raman instrument, all frequencies are mea-
(i.e., refractive index and film thickness are the same), a sured simultaneously so that the entire spectrum can be
sensitive tool like Raman spectroscopy can show how they acquired in a short period of time. The replacement of
are different at the molecular level. Although SPP-Raman monochromator slits with large apertures also improves
enhancement (103 –104 ) is smaller than SERS enhance- the overall spectral throughput. Most FT-Raman systems
ment, SPP-Raman is necessary when nanoscale rough- utilize an internal He–Ne optical standard to ensure wave-
ness would complicate or disturb the adsorbate/substrate length accuracy. The wavelength accuracy of the data is an
system. By coupling the SPP technique with Raman spec- order of magnitude higher than that from scanning instru-
troscopy, it is possible to obtain information about the mentation, allowing the kind of sophisticated data manip-
thickness, refractive index, and chemical identity of thin ulation that has been developed by FTIR spectroscopists.
films. Thus, Raman data can be obtained from samples such as
A standard experimental apparatus uses a thin metal industrial polymers and dyes as well as pharmaceutical
(Ag or Au) film on one side of an equilateral triangu- and petroleum products, which were often inaccessible
lar, hemicylindrical, or hemispherical prisms. The metal to earlier generations of dispersive Raman spectroscopy
film is placed in contact with the adsorbate molecules to using blue-green visible excitation.
be investigated. Laser light is focused through one side
of the prism, to the back of the metal film. Raman scat-
E. Confocal Raman Microscopy
tered light is collected from the front surface of the Ag
or Au film and focused unto the monochromator/detector. In another application of a multichannel detector, the
The total enhancement is dependent on the thickness and Raman microscope provides an image of a microscopic
dielectric properties of the adsorbate, media, and metal. sample according to the Raman-active frequencies of the
When the interaction between the SPP and the Raman ex- components making up the sample. There are two main
citation wavelength is maximized by adjusting either the classes of multichannel detectors: a diode array, which is
incident or scattering angle, the Raman signal intensity is a linear array of minuscule diodes each of which func-
maximized. tions as a miniature photomultiplier tube, and an image
intensifier tube, which is akin to a low-light-level televi-
sion camera. Diode arrays can collect information in only
D. Fourier Transform Raman Spectroscopy
one dimension, whereas an image intensifier permits col-
One major limitation to the routine use of Raman spec- lection over the photoactive surface of the intensifier—that
troscopy lies in the interference caused by the concomi- is, in two dimensions. Hence, an image intensifier affords
tantly excited fluorescence signals. The simplest solu- an entire two-dimensional picture and gives rise to the po-
tion for this problem is to avoid the absorption process tential for performing Raman microscopy. In the Raman
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
microscope, the laser beam is focused into a sample modes to garner structural information, (ii) monitoring
and the focal point is transferred by a microscope and peak intensity changes as a function of time to get kinetic
a monochromator to the active surface of an image inten- information, and (iii) studying the dynamics of the vibra-
sifier. By adjusting the monochromator, it is possible to tional modes while changing environmental parameters to
select different positions of the sample according to their get information about molecular interactions.
Raman peaks. A widely used TR3 S configuration is the pump–probe
The advantages of Raman microscopy can be further experiment. A pump pulse initiates a photochemical or
extended by adding a confocal pinhole to create a confocal photophysical event while the second pulse, a probe pulse,
Raman microscope. By placing a screen with a pinhole on generates resonance Raman scattering. In order to produce
one side of the lens system, only the light from the focal a high Raman signal-to-noise ratio, the lasers typically
point of the lenses will pass through the pinhole to the have 10- to 1000-Hz repetition rates with picosecond—
detector. Any light resulting from outside the lens’ focal 100 femtosecond pulse widths. Because these two pulses
point will be out of focus when it reaches the pinhole (see are independent of one another, both the time delay be-
Fig. 6 inset). The focal point of the objective lens and the tween the pulses and their excitation wavelengths can be
pinhole are known as conjugate points because the lenses selected. This allows for maximum time resolution as well
form an image exactly at the pinhole. Because the pinhole as optimization of the resonance wavelength. By varying
is conjugate to the focal point of the lenses, this is known the time delay between the two laser pulses, a reaction’s
as “confocal” microscopy. kinetic profile can be acquired by measuring the resonant
Figure 6 illustrates the confocal Raman microscope. In Raman spectrum for a sequence of delays. Recent TR3 S
this apparatus, laser light is reflected by scanning mirrors, experiments have explored such issues as the dynamics of
through the microscope, to the sample. The scanning mir- ring-opening reactions, the mechanisms of photobiologi-
rors are used to investigate the sample pixel by pixel so cal systems, and electron-transfer in radical reactions.
that a Raman image can be obtained. The scattered light
reflects back through the microscope, and the longer wave-
VII. NONLINEAR RAMAN EFFECTS
length (scattered) light passes through a dichroic mirror,
focuses at the pinhole, and registers on the detector. Femto-
In Eq. (1) it was assumed that the induced dipole var-
liter sample volumes can be investigated with this method.
ied in a linear fashion with the electric field. However,
for electric field intensities above 109 V/m, as are often
F. Time-Resolved Resonance produced by pulsed lasers, the linear dependence breaks
Raman Spectroscopy down. New spectroscopic phenomena arise from the non-
Time-Resolved Resonance Raman Spectroscopy (TR3 S) linear interaction of a system with intense monochromatic
is a technique used to get structural, kinetic, and molecu- radiation. Each of the four examples considered here in-
lar interaction data from chemical and biological systems volves changes in wavelength of the radiation as a result
by recording resonance Raman spectra in a short time of interaction with the system and can be considered to
span. Using TR3 S, a transient molecular species can be be a variant of the Raman effect. The four examples are
analyzed by (i) monitoring the frequency of vibrational the hyper-Raman scattering, the surface-enhanced hyper-
Raman scattering, stimulated Raman scattering, and co-
herent anti-Stokes Raman scattering (CARS).
A. Hyper-Raman Spectroscopy
Hyper-Raman scattering arises from illuminating a sys-
tem with a focused, pulsed laser that has an irradiance
just above the threshold for nonlinear interaction. If the
incident radiation is of frequency 1ν0 , the scattered radi-
ation is found to include frequencies of the type 2ν0 and
2ν0 ± ν1 , where ν1 , is a frequency associated with a tran-
sition between two levels of the scattering molecules. The
hyper-Raman radiation arises from the second-order non-
linear induced dipole. Thus, Eq. (1) can be expanded to
1
µ = α · E + β · E · E + ···, (24)
FIGURE 6 Schematic diagram of a confocal Raman microscope. 2
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
and the hyper-Raman scattering is controlled by the 50% of the incident radiation at ν0 may be converted to
hyperpolarizability tensor β. The symmetry factors gov- Stokes radiation at ν0 −ν1 . The high-conversion efficiency
erning the factors of the β tensor are not the same as those of stimulated Raman scattering, taken with the coherent
for the components of the α tensor. As a consequence, the nature of the scattering, gives rise to the possibility that
vibrational selection rules for hyper-Raman scattering are the effect can be used to generate laser-like sources at new
significantly different from those for linear Raman scat- frequencies. For example, H2 or D2 , gas under pressure
tering. For example, the torsional vibration in CH2 CH2 is used routinely to shift the incident radiation by 4160
is neither IR nor Raman active but is hyper-Raman-active. or 2942 cm−1 (the H2 or D2 , stretching frequencies, re-
spectively) to gain access to the deep ultraviolet, vacuum
ultraviolet, and mid-infrared spectral regions that are not
B. Surface-Enhanced Hyper-Raman
easily accessible in other ways. Stimulated Raman scatter-
Spectroscopy
ing in H2 and D2 has been used extensively in ultraviolet
Surface-enhanced hyper-Raman scattering (SEHRS) is resonance Raman spectroscopic studies of proteins.
the analog of hyper-Raman scattering just as SERS is the
analog of normal Raman scattering. By adsorbing hyper-
Raman-active molecules to a nanoscale roughened Ag, D. Coherent Anti-Stokes Raman Spectroscopy
Cu, or Au surface or the corresponding colloids, enhance- The CARS technique involves the use of two tunable dye
ment factors as large as 1012 have been measured. The lasers, a pump laser and a Stokes laser, set at frequencies
main utility of the SEHRS technique lies in the new chem- νP and νS , respectively. If these two light beams cross in
ical information available as a consequence of the differ- the sample at the phase-matching angle θ, coherent anti-
ent selection rules. While low-symmetry molecules will Stokes emission at νAS = 2νP − νS is generated through
have very similar SER and SEHR spectra, high-symmetry the third-order linear polarization. In fact, ICARS ∼ IP2 IS .
molecules will have spectra that show new vibrational The laserlike beam due to νAS is greatly enhanced when
features. The combination of three factors: (1) the 1012 the frequency interval, νP − νS = , is equal to a Raman-
enhancement factor of SEHRS; (2) solid-state, femtosec- active molecular vibrational frequency. Thus, vibrational
ond lasers with high repetition rate, high peak power, and Raman spectra are obtained by fixing the frequency of νP
low average power; and (3) high-efficiency multichannel and varying the frequency of νS . Fluorescence rejection
detectors portends a new era in vibrational spectroscopy occurs in a CARS experiment because the signal beam,
using SEHRS to extract new structural information from νAS , is spatially and temporally removed from the fluores-
important chemical and biological systems. cence signal.
In previous years, the major application of CARS was
C. Stimulated Raman Effect in the analysis of gases and gas mixtures, but recent ad-
vances have demonstrated that CARS can be applied to
When monochromatic radiation from a pulsed laser of suf- nongaseous systems as well. A problem in the application
ficiently large irradiance is incident on a scattering sys- of CARS to molecules in solution is interference from the
tem, hyper-Raman scattering is replaced by a different solvent, which can contribute a background emission re-
phenomenon: stimulated Raman scattering. In a typical sulting from the third-order susceptibility of the solvent.
experimental configuration, the laser beam is focused into To a certain extent, this problem may be overcome when
the sample and the scattering is observed in the forward the CARS experiment is carried out under resonance con-
direction, along the axis of the laser beam direction, and ditions. As the νP beam frequency approaches an elec-
at a small angle to this direction. The forward-scattered tronic transition of the solute, the solute’s CARS signal
radiation is found to consist of the incident frequency ν0 is resonance enhanced whereas the background emission
and Stokes and anti-Stokes frequencies of the general for- remains unchanged.
mula ν0 ± nν1 , where ν1 is usually associated with just
one Raman-active vibration of the scattering molecules
and n = 1, 2, 3, . . . . For example, if liquid benzene is
VIII. CHEMICAL AND BIOCHEMICAL
illuminated, the Stokes and anti-Stokes shifts are all ex-
APPLICATIONS
act multiples of 992 cm−1 , which represents the strongest
feature in the normal Raman spectrum of benzene.
A. Chemical Applications
Stimulated Raman scattering differs from normal
Raman scattering in its angular dependence, intensity, and Five representative applications in which Raman spec-
frequency distribution. A major fraction of the incident troscopy has proved to be very powerful are cross-
light is converted to radiation at ν0 ± nν1 . In benzene, sectional imaging of live cells with CARS, investigation
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
of industrial polymers with Fourier Transform (FT) technique. Polymer samples are historically plagued by
Raman Spectroscopy, the study of self-assembled mono- large fluorescence backgrounds and sample degradation
layers (SAMs) with SERS, the assignment of vibrational under the heating effects of visible wavelength laser irra-
bands by the combination of SERS and SEHRS, and SPP- diation. In fact, estimates suggest that up to 95% of poly-
enhanced Raman studies of carbon clusters. mer samples cannot be examined by normal Raman spec-
troscopy. Until the advent of FT-Raman, most polymer
1. CARS Cross-Sectional Imaging samples were investigated by infrared (IR) spectroscopy.
Often, the strenuous sample preparation required for IR
Recent CARS experiments have targeted cross-sectional
investigations alters the polymer structure, thus sacrific-
imaging by making use of solid state femtosecond lasers
ing an accurate spectroscopic view. The high symmetry
with λp and λs in the NIR region. The long wavelength
of most polymer samples also acts as a disadvantage in
pulses give two major advantages over visible excitation:
IR experiments because many symmetric stretching mode
(i) the excitation wavelengths are far from electronic tran-
vibrations are IR inactive. Often, IR spectra are miss-
sition bands, so the background signals are small, and
ing all information regarding the homonuclear polymer
(ii) Rayleigh scattering is minimized in heterogeneous
backbone.
samples, allowing deeper penetration into the sample vol-
FT-Raman addresses many of the disadvantages of IR
ume. By collinearly coupling the two pulses into an optical
polymer analysis. First, because sample preparation is
microscope and through a high numerical aperture objec-
minimal for FT-Raman analysis, no valuable information
tive, the excitation pulses can be focused tightly onto the
is lost. Second, because the Raman selection rule is based
sample. Because CARS is a nonlinear optical process,
on changing bond polarizability rather than on the chang-
only a small volume of the sample is excited. The advan-
ing dipole moment (as in IR spectroscopy), the homonu-
tages of small volume excitation include background sig-
clear backbone stretches can be seen in Raman spectra.
nal rejection, decreased photodecomposition of the sam-
Third, detailed information about the chemical compo-
ple, and the ability to section a three-dimensional object
sition, structure, and stereoregularity can be obtained by
by changing the focal plane of the excitation pulses. The
FT-Raman measurements. Some of the most exciting data
consequence of a long wavelength, small volume exci-
gained in FT-Raman investigations of polymers lie in the
tation is that three-dimensional imaging of live cells has
details of chain conformation due to intermolecular inter-
been achieved. Figure 7 shows CARS images of live, un-
actions. The chain-packing details help to explain physical
stained bacteria tuned to the Raman shift of 2878 cm−1 ,
characteristics of polymers such as absorbance spectra and
an aliphatic C–H stretching band. The lipid bilayer of the
transition temperatures.
bacterial cell membrane is rich in aliphatic C–H bonds.
CARS will find more application in coming years because
it gives vibrational contrast and high sensitivity, but the 3. SERS of Self-Assembled Monolayers
power levels are tolerable to living cells. at the Solid/Liquid Interface
2. FT-Raman Analysis of Polymers SERS experiments give scientists a window into the
complex processes occurring at solid/liquid, solid/gas,
The spectroscopic advances made in the field of poly- solid/UHV, and solid/solid interfaces. When molecules
mer chemistry demonstrate the impact of the FT-Raman adsorb to a SERS-active substrate, the measured spec-
tra give valuable information concerning the structure
and reactivity of the adsorbate molecules. Comparison
of SERS spectra to bulk Raman spectra demonstrates
any chemical or conformational variance between the
free and adsorbed states. This information is important
when using SERS to understand complex phenomena such
as heterogeneous catalysis and electrochemistry at the
molecular level.
As an example, consider the pervasive use of self-
assembled monolayers (SAMs) for the surface modifica-
tion of gold and silver substrates. A recent SERS/SERRS
study of the Au film/SAM adsorbate interface demon-
FIGURE 7 CARS image of a live bacterial cell. Imaging was
tuned to the Raman shift of 2878-cm−1 band (aliphatic C–H strates that thin films of Au(111) on mica substrates,
stretch). [Reproduced from Zumbusch et al. (1999). Phys. Rev. previously thought to be SERS-inactive, is SERS-active
Lett. 82(20), 4142, by permission.] without any additional treatment. This discovery is quite
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
significant because it allows SERS experiments to be exe- attributed to the ∼100 nm wide, atomically flat islands that
cuted on undisturbed, unroughened Au(111)/SAM adsor- comprised the surface of Au(111)/mica samples shown in
bate surfaces. The impact of better understanding SAM Fig. 8B or the roughness features between the terraces
adsorption to a Au(111) substrate is significant because of the Au(111) surface. Future experiments will address
this system acts as a model for other substrate/adsorbate these hypotheses.
systems.
In the aforementioned experiment, the SERS activity
4. SEHRS Assisted Assignment
of rough Au films is compared with that of single crystal
of Vibrational Bands
Au(111) films. Figures 8A and B show nanometer-scale
AFM images of a rough Au surface and a Au(111) surface A particularly salient example of the complementarity of
before SAM modification. The corresponding SERRS of normal Raman spectroscopy with other techniques is ex-
SAMs on these surfaces are in Figs. 8C and D, respectively. pressed in a recent publication of the combined data from
The resonance Raman condition is a result of overlap be- infrared, normal Raman, SERS, SEHRS, and theoretical
tween the adsorbate molecule’s electronic absorption band predictions for one molecule. First, ab initio theoretical
and the excitation wavelength. Although not shown within predictions were made for the vibrational characteristics
this text, SERS spectra were also measured for each sam- of trans-1,2-bis(4-pyridyl)ethylene (BPE) at the Hartree-
ple. With these two spectra, the surface resonance Raman Fock 6-31G∗ level. When the spectra were collected, com-
(SRR) enhancement (∼102 ) could be separated from the parisons were made between the theoretical and exper-
EM enhancement (∼103 ). This weak EM enhancement, imental results as well as among the different spectra.
compared to the standard EM contribution of 104 –105 , is Based on the known selection rules for each spectroscopy
and the matching of wavenumber shifts to theoretical pre-
dictions, all vibrational bands were assigned.
Though this study only gives exact information for
BPE, it demonstrates the power of combined theoreti-
cal/spectroscopic characterization. The development, in
recent years, of ab initio electronic structure calculations
of high accuracy has revolutionized theoretical chemistry.
The ability to efficiently calculate vibrational properties
from first principles has or will completely replace normal
coordinate analysis.
Figure 9 shows the SER and SEHR BPE spectra as
well as the theoretical prediction for each. Although
the hyper-Raman scattering efficiency is eight orders
of magnitude smaller than the linear Raman scattering
efficiency, the SEHR signals are only three orders of
magnitude smaller than the SERS signals. This trans-
lates to 1012 -fold enhancement over normal hyper-Raman
scattering. There has been renewed interest in hyper-
Raman spectroscopy due to this immense SEHR enhance-
ment. Vibrational modes that are active in only hyper
Raman spectroscopy can be measured reliably for the first
time.
FIGURE 9 Theoretical SER (A) and SEHR (C) and experimental SER (B) and SEHR (D) spectra of BPE
on a roughened silver surface. [Reproduced from Yang et al. (1996). J. Chem. Phys. 104(11), 4313, by
permission.]
of interest. In cluster science, a common solution to the acetylene ring conformation, but not a fullerene or bowl
problem of high reactivity is to embed the clusters inside structure.
solid matrix of condensed inert gas. SPP-enhanced Raman
spectroscopy allows Raman spectra to be measured from
B. Biochemical Applications
smooth surfaces.
A recent SPP-resonance Raman experiment explored A major advantage of Raman spectroscopy for the anal-
the structural conformations of several carbon clusters. ysis of biomolecules stems from the fact that water has
The C16 , C18 , and C20 clusters were created by laser ab- a weak Raman spectrum. Spectra can be recorded for
lation of a graphite rod and then deposited into a N2 ma- aqueous solutes at 10−1 –10−2 M with little interference
trix on the silvered SPP prism surface. After finding the from the solvent. For a chromophore under the RR con-
SPP resonance condition (as described in Section VI.B), dition the accessible concentration range becomes 10−4 –
Raman spectra were collected with six different excita- 10−6 M. Moreover, the intensity enhancement associated
tion wavelengths. SPP-Raman enhancement is operative with the RR effect confers the important advantage of
in all six spectra. The strong dependence of the Raman selectivity, allowing one to observe selectively the vibra-
spectra for the C20 cluster on λex shown in Fig. 10 in- tional spectrum of a chromophore that is just one compo-
dicates that enhancement due to RRS is simultaneously nent of an extremely complex biological system. Because
operative. It is important to emphasize that these signals many biomolecules have chromophores with an ultravio-
shown in this figure are obtained from ca. 1010 clusters let (UV) resonance condition, one may also selectively ex-
in the laser focal spot. Upon comparing the Raman peak cite a chromophore by irradiating these molecule with UV
frequencies for all three carbon clusters to theoretical pre- light. This technique is known as Ultraviolet Resonance
dictions, the researchers were able to hypothesize that all Raman Spectroscopy (UVRRS). In recent years, Raman
three carbon clusters adopt either a linear chain or poly- difference spectroscopy (RDS) has been developed in
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
However, many biological sites of importance do not Resonance Raman labeling is used in order to highlight
contain a suitable chromophore, and for these, the RR the spectrum of small portions of the large molecule. This
labeling technique was developed. An RR label, usu- technique has limited applications to protein molecules
ally mimicking a natural component, is introduced into because the chromophore dominates the observed spec-
the system as a reporter group, and there has been suc- trum and not all proteins will accept chromophore addi-
cess in using chromophoric ligands to study antibody- tion. A more generally applicable method known as RDS
haptcn, enzyme-inhibitor (and drug), DNA–drug, and has become technically feasible in recent years to study
cell–dye interactions. Moreover, the labeling technique events such as protein–ligand binding, enzymatic cataly-
can aid the understanding of the molecular details of sis, and protein assembly.
enzymolysis. The hydrolysis of thionoesters (of the In RDS, spectra of both the normal and the modi-
type RC( O)NHCH2 C( S)OCH3 ) by the enzyme pa- fied version of the macromolecule are collected. By sub-
pain occurs via the formation of a transient dithioester tracting the normal spectrum from the tagged molecules’
RC( O)NHCH2 C( S)–S–papain (the thiol sulfur be- spectrum, information about the tagged portions of the
longs to a cysteine side chain in the active site). The molecule is obtained. Two methods used in the tagging
dithioester absorbs at 315 nm, and thus, the label is gen- process are ligand binding and isotopic editing. By sub-
erated at the time and location of catalysis. The 324-nm tracting a bare protein’s spectrum from the spectrum of
excited RR spectrum in Fig. 11 shows how different spec- the protein bound to a ligand, the RDS spectrum of the
tral features monitor conformations in different parts of bound ligand is obtained. In isotopic editing, the substitu-
the enzyme–substrate complex. By this means it is pos- tion of an atom in the bond of interest with another stable
sible to monitor the critical events in the bonds under- isotope will cause shifts in the Raman bands for motions
going catalytic transformation in a complex of molecular involving the edited nucleus. By subtracting the normal
weight 24,000. spectrum from the isotopically edited spectrum, Raman
Raman analysis of large macromolecules, such as pro- peaks will be seen only for the modes of vibration involv-
teins, often yields complicated, spectrally crowded results. ing the isotopically edited atoms.
A. Symmetry of Vibrations
from Single-Crystal Spectra
Raman spectroscopy offers a valuable means of unam-
biguously determining the symmetry of vibrations for
single oriented crystals. Planned selection of the direc-
tion and polarization state of both the incident light and
the observed Raman scattering enables the magnitude of
each tensor component [Eq. (2)] to be determined. By this
means, the symmetry classes of the Raman-active vibra-
tions can be identified.
FIGURE 12 Polarized Raman spectra of calcite; 632.8-nm exci-
A convenient notation for describing polarization data
tation. [Reproduced from Porto et al. (1966). Phys. Rev. 147, 608.]
from single crystals is due to S. Porto. This involves four
symbols (usually chosen from x, y, z axis designations),
single elements) and extrinsic semiconductors are doped
which define the propagation direction of the incident ra-
materials. Raman spectroscopy is an efficient tool used
diation, the direction of the E of the incident radiation,
to investigate semiconductor surfaces or interfaces made
the direction of the E of the scattered radiation being
up of a semiconductor and a metal or an insulator, or to
examined, and the direction of propagation of the scat-
characterize the strains brought about by heavy doping in
tered radiation. The second and third symbols are placed
semiconductors. In one instance, Raman scattering from
in parentheses, for example, z(x y)x, and define the com-
an extrinsic semiconductor can give rise to a tunable laser
ponents of the scattering tensor.
source. The example system involves a wide-band-gap
Crystalline calcite furnishes an instructive example of
II–VI semiconductor, such as InSb, subjected to a strong
this notation. Calcite (a form of calcium carbonate) has
magnetic field at low temperature. Under these conditions,
the point group symmetry D3d and symmetry arguments
interaction with a photon causes two electrons in the semi-
determine that the vibrational Raman modes will be made
conductor to change their energy levels. The overall result
up of one vibration of a1g symmetry and four doubly de-
is the change of electron spin quantum numbers S of ±1.
generate pairs of vibrations of eg symmetry. The polarized
That is, one spin flip. The spin–flip energy is proportional
Raman spectra of calcite are shown in Fig. 12 for four dif-
to the strength of the magnetic field and is approximately
ferent orientations. For the illumination observation con-
7 cm−1 when the magnetic field is 8T . Thus, intense peaks
figuration y(zz)x, the scattering is proportional to αzz 2
.
are observed at 7 cm−1 in the Stokes and anti-Stokes
Since this is nonzero only for the a1g mode, only this
Raman spectra. The dependence of the peak position or
vibration will be seen in the y(zz)x configuration. Thus,
magnetic field offers the possibility of constructing a tun-
from Fig. 12, the 1088-cm−1 feature is immediately iden-
able, spin–flip laser. For example, the 943-cm−1 output of
tified as the a1g mode.
a pulsed CO2 , laser may be tuned from 700 to 1100 cm−1
by using the magnetic field dependent spin–flip of InSb.
B. Semiconductors
Raman spectroscopy has been used to obtain valuable
C. Phase Transitions
information on the properties of semiconductors. There
are two major classes of semiconducting materials: Structural detail associated with phase transitions can
intrinsic semiconductors are very high purity (i.e., be obtained by recording the Raman spectra of oriented
P1: GNH/LBD P2: GNY/GUR QC: GNB Final Pages
Encyclopedia of Physical Science and Technology EN013A-651 July 26, 2001 20:36
X. RECENT DEVELOPMENTS—SINGLE
MOLECULE SERS (SMSERS)
FIGURE 14 Correlated SERS and tapping-mode AFM images obtained from spatially isolated, single Ag nanopar-
ticles. These particles were selected by wide-field screening for maximum enhancement at (A) 488 nm, (B) 567 nm,
and (C) 647 nm, respectively. [Reproduced from Nie et al. (1998). JACS 120, 8009, by permission.]
Ott, A. K., Rechtsteiner, G. A., Felix, C., Hampe, O., Jarrold, M. F., and Yang, W. H., Hulteen, J. C., Schatz, G. C., and Van Duyne, R. P. (1996).
Van Duyne, R. P. (1998). “Raman spectra and calculated vibrational “A surface-enhanced hyper-Raman and surface-enhanced Raman scat-
frequencies of size-selected C16 , C18 , and C20 clusters,” J. Chem. tering study of trans-1,2-bis(4-pyridyl)ethylene adsorbed onto silver
Physics 109, 9652–9655. film over nanosphere electrodes. Vibrational Assignments: Experi-
Schrader, B. (ed.) (1995). “Infrared and Raman Spectroscopy,” VCH ment and theory,” J. Chem. Physics 104, 4313–4323.
Publishers, Inc., New York. Zumbusch, A., Holtom, G. R., and Xie, X. S. (1999). “Three-dimensional
Turrell, G., and Corset, J. (eds.) (1996). “Raman Microscopy, Develop- vibrational imaging by coherent anti-Stokes Raman scattering,” Phys-
ments and Applications,” Academic Press, New York. ical Rev. Lett. 82, 4142–4145.
P1: GPB/GLE P2: GNH/GLT P3: GTY Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN014G-675 July 30, 2001 16:29
469
P1: GPB/GLE P2: GNH/GLT P3: GTY Final Pages
Encyclopedia of Physical Science and Technology EN014G-675 July 30, 2001 16:29
Since their invention in the 1980s, STM and AFM have late the probe–sample distance. Different SPM methods
provided scientists with exquisitely detailed snapshots of a are distinguished by the specific feedback mechanism
wide variety of structural and chemical phenomena occur- employed and the type of data, in addition to topogra-
ring on surfaces. The impact of STM on our understanding phy, that are recorded. The different feedback mecha-
of surface chemistry and physics was apparent immedi- nisms are best understood by considering specific SPM
ately. In 1986, Gerd Binnig and Heinrich Rohrer were techniques.
awarded the Nobel Prize for their invention of the STM
only 5 years earlier. The success of STM and AFM has
A. Scanning Tunneling Microscopy (STM)
led to the proliferation of scanning probe technology in all
aspects of science and industry. New scanning probe mi- Figure 1A shows a schematic diagram of an STM. The
croscopies have been developed that allow high-resolution STM probe is a thin metal wire that has been cut or chem-
imaging of a large spectrum of sample characteristics in ically etched to a sharp point. The probe is attached to a
addition to topography, including frictional properties, lo- piezoelectric tube scanner described above allowing pre-
cal electric and magnetic domains, thermal behavior, opti- cise sweeping of the probe tip back and forth across the
cal absorption, fluorescence, and birefringence. SPMs are sample. When the probe is within a few angstroms of the
now valued not just as imaging instruments, but as impor- substrate, application of a voltage produces a quantum me-
tant analysis and metrology tools for examining a range of chanical tunneling current that is extremely sensitive to the
material properties on nanometer length scales. Scanning probe–sample separation. The tunneling current increases
probe techniques continue to be developed and to have an exponentially with decreasing probe–sample distance, and
impact on virtually every area of technology ranging from thus it can be used to regulate the probe–sample separation
molecular biology to semiconductors, surface chemistry, in the Z direction. While the probe is being scanned, the
catalysis, and nanofabrication. tunneling current at a given applied bias is detected by
the feedback control electronics and compared to a “set-
point” current that the user chooses (e.g., 1 nA). If the
I. BASIC OPERATING PRINCIPLES
measured current is higher than the setpoint current, the the voltage dependence of the tunneling current is the ba-
feedback electronics applies voltages to the Z axis of the sis of local tunneling spectroscopy, which is not covered in
tube scanner to make the tip retract until the current equals this article, but is discussed extensively by Wiesendanger
the setpoint. Conversely, if the measured current is too (1994).
small, voltages are applied to the tube scanner to extend STM imaging can be accomplished in air, vacuum, or
the tip closer to the substrate until the current increases under fluids, though the best quality images are usually
to the setpoint value. Essentially, the goal of the feedback taken under vacuum since contamination of the tip or sam-
electronics is to zero out the “error” signal, or the differ- ple by adsorbates (e.g., water) is minimized. The time re-
ence between the measured current and the setpoint. This quired to produce an atomic-resolution STM image is tens
electronic feedback process is very quick, requiring only of seconds to minutes depending on the scan size and scan
fractions of a second, so that the tip may track the surface speed. Samples for STM analysis must be conductive.
topography while scanning at speeds from 100 to several
thousand nanometers per second.
The tube scanner is calibrated and therefore it is known B. Atomic Force Microscopy (AFM)
how many angstroms the tip moves in the X , Y , and Z Atomic force microscopy uses the repulsive or attractive
directions for given applied voltages. Topographic height forces between the probe and sample rather than tunneling
images are produced by recording the Z -axis position of current to track sample topography. Figure 2A shows an
the tip as a function of its X and Y coordinates (pixel po-
sition). It is a remarkable fact that the constant-tunneling-
current methodology allows the tip to track the topo-
graphic contours of individual atoms, as shown schemat-
ically in Fig. 1B. The resolution of STM height images
is typically better than 1 Å in the X and Y directions
and 0.1 Å in the Z direction, sufficient to visualize indi-
vidual atoms. This superb resolution is achieved because
the tunneling current is emitted essentially from the last
atom on the end of the tip. In reality, tips are not smooth
as indicated by Fig. 1B, but have atomic corrogation and
many different asperities or “minitips” that can emit tun-
neling current. However, because the tunneling current (in
vacuum) decreases by about a factor of 10 for every 1 Å of
separation, and because generally there is one minitip that
is a few angstroms closer to the sample than the others, it
is usually the case that the vast majority of the tunneling
current flows through only one protrusion. It is really the
exponential distance dependence of tunneling current that
gives STM its tremendous resolution.
It is important to note that the magnitude of the tunnel-
ing current depends not only on the tip–sample separation,
but also on the electron density of states (number of states
per volume per unit energy) in the sample and the tip bias.
If the tip is biased negatively, electrons tunnel from the tip
into empty electronic states in the sample surface. If the tip
is biased positively, electrons tunnel from filled states in
the sample to the tip. Therefore, the contrast or corrogation
seen in STM images is really a map of the spatial distri-
bution of electron states, either empty or filled, depending FIGURE 2 (A) Schematic illustration of atomic force microscopy.
on the tip bias. In many cases, the electron density directly Cantilever deflection recorded at the photodetector is used to gen-
reflects atomic positions on the sample. As we shall see in erate a Z-position feedback signal, analogous to the way tunneling
current is used to control Z position in STM. (B) Scanning electron
Section II, however, this is not always the case. Addition-
micrograph of a commercial silicon nitride AFM tip and cantilever.
ally, there can be a nonlinear dependence of the tunneling [Reprinted with permission from Takano, H., Kenseth, J. R., Wong,
current on the tip voltage, reflecting the spectrum of the S.-S., O’Brien, J. C., and Porter, M. D. (1999). Chem. Rev. 99,
electronic energy states at the sample surface. Examining 2845–2890. Copyright 1999 American Chemical Society.]
P1: GPB/GLE P2: GNH/GLT P3: GTY Final Pages
Encyclopedia of Physical Science and Technology EN014G-675 July 30, 2001 16:29
illustration of a typical AFM apparatus. The probe, shown signal indicating how much the cantilever has deflected
in Fig. 2B, consists of a sharp tip on the end of a small sideways. The optical beam method allows deflections as
cantilever a few hundred micrometers in length. Forces small as an angstrom to be detected.
exerted on the tip cause the cantilever to bend. The force AFM instruments are designed to scan either the sample
is directly proportional to the cantilever deflection and is or the cantilever-tip assembly. Figure 2A shows a sample
given by scanning configuration. In contact mode operation, the
sample is scanned underneath the cantilever tip, which
F = kz, remains stationary. The user defines a setpoint cantilever
where k is the cantilever force constant (in nN/nm, anal- deflection (actually, a voltage proportional to the deflec-
ogous to a spring constant) and z is the vertical deflec- tion) that is to be maintained by the feedback electronics
tion. Cantilevers can be fabricated with force constants during scanning. If the cantilever deflection increases, the
as low as 0.01 nN/nm, so that forces as small as 0.1 nN feedback electronics applies a voltage to the Z axis of the
give nanometer deflections that can be detected easily. piezo to cause the sample to retract from the lever, thus
The sensitivity of AFM to forces smaller than the bond- lowering the deflection back to the setpoint value. Con-
ing forces between individual atoms was the motivation versely, if the cantilever deflection decreases, the feedback
for the name “atomic force” microscopy. Some investi- electronics applies a Z -axis voltage to cause the sample to
gators instead use the name scanning force microscopy move toward the lever, increasing the deflection back to
(SFM) to emphasize the relationship to other scanning the setpoint value. As in STM, the feedback is done very
probe techniques. quickly, and recording the Z position as a function of X –Y
In the simplest version of AFM, called contact mode coordinates produces a topographic map or height image.
or normal force mode, the tip is brought into hard con- Resolution in AFM is governed principally by the
tact with the sample surface such that the cantilever bends sharpness of the probe tip, although the sample char-
due to the tip–sample repulsion (Fig. 3A). The cantilever acteristics and specific operating mode are also impor-
deflection can be measured in several ways, but the most tant. Commercial cantilever-tip assemblies are made from
common is the optical beam method. In this approach, a silicon or silicon nitride using conventional microfabri-
laser beam from a semiconductor laser is focused on the cation methods common to the semiconductor industry.
back side of the cantilever, as shown in Fig. 2A. The re- Figure 2B shows an electron micrograph of a microfab-
flected light hits a position-sensitive detector that produces ricated silicon nitride probe. Commercial probes can be
a voltage indicating how much the cantilever has deflected made with tip radii as small as 10 nm, but 20–40 nm is
up or down. In some cases, the detector also produces a more common. With a 20-nm-radius tip it is possible to
obtain images of surfaces that show topographic corroga-
tion due to atomic or molecular packing.
As AFM has matured, a number of more sophisticated
imaging modes have been developed. In lateral force mode
(Fig. 3B) the sample is scanned sideways relative to the
long axis of the cantilever. Torsional forces exerted on
the tip cause the cantilever to twist and consequently to
deflect the optical beam horizontally on the photodetec-
tor. Recording the lateral deflection or twisting of the can-
tilever as a function of X –Y position gives a lateral force
map. Lateral force images are particularly sensitive to fric-
tion force between the tip and sample, and therefore these
images are also called friction force maps.
Two other popular scanning modes involve oscillating
the cantilever-tip assembly while the sample is scanned
(Figs. 3C and 3D). In tapping mode (Fig. 3C) the tip
is oscillated vertically with an amplitude of 50 to sev-
eral hundred nanometers and makes contact with the sam-
ple at the bottom of every downstroke. The optical beam
and photodetector assembly is used to detect the oscilla-
FIGURE 3 Different AFM imaging modes. (A) Contact or normal
force mode. (B) Lateral or friction force mode. (C) Intermittent tion amplitude, which is dependent on how close the tip
contact mode, also known as noncontact or tapping mode. (D) is to the sample. The feedback control electronics main-
Force modulation mode. tains the oscillation amplitude at a user-defined setpoint
P1: GPB/GLE P2: GNH/GLT P3: GTY Final Pages
Encyclopedia of Physical Science and Technology EN014G-675 July 30, 2001 16:29
FIGURE 10 Simultaneously acquired topography (A), lateral force (B), and elasticity (C) maps of a phase-separated
hydrocarbon and fluorocarbon film taken by atomic force microscopy. Each image is 3 µm × 3 µm. The raised (bright)
islands in the topographic image contain the hydrocarbon molecules that lie on top of a sea of fluorocarbon. The
lateral force map shows that the hydrocarbon islands, which appear dark, exhibit lower friction. The elasticity map
has the opposite contrast and shows that the islands are less compliant (or stiffer) than the fluorocarbon regions.
These images allow correlation of friction with mechanical compliancy. [Reprinted with permission from Overney,
R. M., Meyer, E., Frommer, J., Guntherodt, H.-J., Fujihira, M., Takano, H., and Gotoh, Y. (1994). Langmuir 10, 1281–
1286. Copright 1994 American Chemical Society.]
consequently has an experimental voltage resolution of a Appropriate calibration of SCM has been achieved in
few millivolts. The spatial resolution is determined by the important semiconductor materials, such as silicon and
sharpness of the probe and is approximately 50 nm for gallium arsenide, and SCM has been used successfully
microfabricated commercial probes. to measure dopant profiles across p–n junctions in these
materials with 20-nm resolution. Figure 17 shows SCM
images of three different dopant regions in a silicon wafer
D. Scanning Capacitance Microscopy (SCM)
covered with a thin layer of silicon oxide (SiO2 ). The con-
Scanning capacitance microscopy (SCM) is a contact trast in the image corresponds to different concentrations
mode variant of atomic force microscopy (AFM) in which of n-type (regions I and III) and p-type (region II) dopants.
changes in tip–sample capacitance are imaged simultane-
ously with surface topography. Samples for SCM anal-
E. Scanning Electrochemical
ysis are generally covered with a thin insulating layer
Microscopy (SECM)
that serves as the capacitance dielectric between the tip
and sample. The spatially mapped capacitance variations In SECM, a tiny electrode is scanned over a surface im-
can be attributed to changes in the dielectric thickness, mersed in electrolyte solution. The electrode probe senses
changes in the dielectric constant, or variation in the lo- either electrochemical current or potential, and can be used
cal charge carrier density underneath the dielectric. In the to map chemical reactivity or the distribution of ions across
most typical application of SCM, which is the character- surfaces. A typical amperometric (current sensing) probe,
ization of semiconductor devices, capacitance variations shown in Fig. 18A, is fabricated by sealing a Pt or Au wire
are converted via a theoretical model into a dopant con- or a carbon fiber in a glass capillary. Potentiometric probes
centration profile in the semiconductor underneath the di- are usually ion-selective electrodes that are sensitive to a
electric. given ion, for example, Na+ or Ca2+ . The potential mea-
In typical operation, a conducting tip is scanned in con- sured by these probes is proportional to ion concentration.
tact at constant applied force over the dielectric layer on There are many possible scanning modes in SECM;
the sample, generating a topographic image. While the three of the most common are shown in Fig. 18B. In feed-
probe is scanned, a set of electronics (the capacitance sen- back mode, an amperometric probe scans the sample under
sor) is used to apply an oscillating voltage to the probe and an electrolyte solution containing a redox mediator, i.e., a
to measure the tip–sample capacitance. The sensor mea- soluble molecule that can exist in two states of charge (R,
sures dynamic capacitance change rather than static capac- reduced, and O, oxidized). A potential is applied to the
itance. A sensor typically consists of a high-frequency os- probe such that mediator molecules that contact the probe
cillator and associated circuitry whose resonant frequency are electrochemically oxidized,
is determined by the total capacitance of the tip–sample
R ⇒ O + e− ,
assembly. The instrument signal arises from the variation
in resonant frequency of the circuit due to tip–sample generating a tip current Itip . The current Itip is a function
capacitance changes as the tip scans over the sample. of how close the tip is to the sample surface. For example,
Commercial sensors can detect capacitance changes of
10−21 F. The tip and sample must have an insulator be-
tween them, and surface preparation is very important to
obtaining good SCM images. Resolution is now regularly
10–20 nm.
The most important application of SCM is two-
dimensional dopant concentration profiling in semi-
conductors. However, the raw capacitance–voltage data
obtained from SCM must be converted by a mathematical
model into a dopant concentration. Therefore, develop-
ment and validation of appropriate models represents a
large part of SCM methodology. To validate the various FIGURE 17 Scanning capacitance images of a silicon wafer
models, other experimental techniques must be employed with an SiO2 overlayer. Three different dopant regions (contrasts)
to measure and verify the dopant profiles independently. are visible. Region I: n-type; region II: p-type; region III: highly
doped n-type. The images were taken with a 80-kHz, 4-V mod-
SCM data are usually compared to secondary ion mass
ulation voltage. The left image is 20 µm × 20 µm. The right im-
spectrometry (SIMS) measurements of dopant concentra- age is 80 µm × 80 µm. [Reprinted with permission from Nakagiri,
tions. Spreading resistance profiling (SRP) and computer N., Yamamoto, T., Sugimura, H., Suzuki, Y., Miyashita, M., and
simulations are also employed to check model validity. Watanabe, S. (1997). Nanotechnology 8, A32–A37.]
P1: GPB/GLE P2: GNH/GLT P3: GTY Final Pages
Encyclopedia of Physical Science and Technology EN014G-675 July 30, 2001 16:29
Dunn, R. C. (1999). “Near-field scanning optical microscopy,” Chem. Mirkin, M. V., and Horrocks, B. R. (2000). “Electroanalytical measure-
Rev. 99, 2891–2927. ments using the scanning electrochemical microscope,” Anal. Chim.
Hamers, R. J. (1996). “Scanned probe microscopies in chemistry,” Acta 406, 119–146.
J. Phys. Chem. 100, 13103–13120. Paesler, M. A., and Moyer, P. J. (1996). “Near-Field Optics: Theory,
Hartmann, U. (1999). “Magnetic force microscopy,” Annu. Rev. Materi- Instrumentation, and Applications,” Wiley, New York.
als Sci. 29, 53–87. Takano, H., Kenseth, J. R., Wong, S.-S., O’Brien, J. C., and Porter,
Jacobs, H. O., and Stemmer, A. (1999). “Measuring and modifying M. D. (1999). “Chemical and biochemical analysis using scanning
the electric surface potential distribution on a nanometre scale: A force microscopy,” Chem. Rev. 99, 2845–2890.
powerful tool in science and technology,” Surface Interface Anal. Wiesendanger, R. (1994). “Scanning Probe Microscopy and Spec-
27, 361–367. troscopy,” Cambridge University Press, Cambridge.
Majumdar, A. (1999). “Scanning thermal microscopy,” Annu. Rev. Ma- Williams, C. C. (1999). “Two-dimensional dopant profiling by scanning
terials Sci. 29, 505–585. capacitance microscopy,” Annu. Rev. Materials Sci. 29, 471–504.
P1: GNH Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN015L-953 August 2, 2001 12:43
Sonoluminescence and
Sonochemistry
Kenneth S. Suslick
University of Illinois
I. Introduction to Cavitation
II. Sonoluminescence
III. Sonochemistry
IV. Summary
. 363
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN015L-953 August 2, 2001 12:43
ultrasound to enhance chemical reactivity has been ex- to 10−4 cm. These are not molecular dimensions. Con-
plored, with important applications in mixed phase syn- sequently, the chemical effects of ultrasound do not arise
thesis, materials chemistry, and biomedical uses. For from a direct interaction with molecular species: no di-
example, the sonochemical decomposition of volatile rect coupling of the acoustic field on a molecular level is
organometallic precursors in low-volatility solvents pro- responsible for sonochemistry or sonoluminescence. In-
duces nanostructured materials in various forms with high stead, sonochemistry and sonoluminescence derive prin-
catalytic activities. Nanostructured metals, alloys, car- cipally from acoustic cavitation, which serves as an effec-
bides and sulfides, nanometer colloids, and nanostructured tive means of concentrating the diffuse energy of sound.
supported-catalysts can all be prepared by this general Compression of a gas generates heat. When the compres-
route. Another important application of sonochemistry to sion of bubbles occurs during cavitation, it is more rapid
materials chemistry has been the preparation of biomate- than thermal transport and consequently generates a short-
rials, most notably protein microspheres. Especially for lived, localized hot-spot. There is a general consensus that
liquid–solid reactions, the rate enhancements that ultra- this hot-spot is the source of homogeneous sonochemistry.
sound can provide have proved extremely useful for the In 1917, Rayleigh’s mathematical model for the collapse
synthesis of organic and organometallic compounds. Be- of cavities in incompressible liquids predicted enormous
cause cavitation can only occur in liquids, chemical reac- local temperatures and pressures. Ten years later, Richards
tions are not generally seen in the ultrasonic irradiation of and Loomis reported the first chemical and biological ef-
solids or solid-gas systems. fects of ultrasound.
If a moderately intense acoustic field (greater than
≈0.5 MPa) is applied to a liquid, the liquid can fail during
I. INTRODUCTION TO CAVITATION the expansion (i.e., tensile or negative pressure) portion of
the sound field; weak sites within the liquid (e.g., preex-
When a liquid is irradiated with high-intensity sound or ul- isting gas pockets, called “cavitation nuclei”) are caused
trasound, acoustic cavitation (the formation, growth, and to rapidly grow, thereby producing vapor and gas-filled
implosive collapse of bubbles in liquids irradiated with cavities (i.e., bubbles). These bubbles continue to grow
sound) generally occurs. This is the phenomena responsi- during the negative pressure portion of the sound field,
ble for sonochemistry and sonoluminescence. During cav- until the sound field pressure turns positive. The result-
itation, the collapse of bubbles produces intense local heat- ing inertial implosion of the bubbles (now mostly filled
ing and high pressures, with very short lifetimes. In clouds with vapor and thus unable to provide stiffness) can be
of cavitating bubbles, these hot-spots have equivalent tem- extremely violent, leading to an enormous concentration
peratures of roughly 5000 K, pressures of about 1000 at- of energy within the small residual volume of the col-
mospheres, and heating and cooling rates above 1010 K/s. lapsed bubble (Fig. 1). This violent cavitation event has
In single bubble cavitation, conditions may be even more been termed “transient cavitation.” A normal consequence
extreme. Cavitation, then, can create extreme physical and of this unstable growth and subsequent collapse is that the
chemical conditions in otherwise cold liquids. cavitation bubble itself is destroyed. Gas-filled remnants
If liquids containing solids are irradiated with ul-
trasound, related phenomena can occur. Near an ex-
tended solid surface, cavity collapse becomes nonspher-
ical, which drives high-speed jets of liquid into the
solid surface. These jets and associated shock waves
can cause substantial surface damage and expose fresh,
highly heated surfaces. In addition, high-velocity inter-
particle collisions will occur during ultrasonic irradiation
of liquid–powder suspensions through cavitation and the
shockwaves it creates in such slurries. The resultant colli-
sions are capable of inducing dramatic changes in surface
morphology, composition, and reactivity.
A. Acoustic Cavitation
Ultrasound spans the frequencies of roughly 15 kHz
to 1 GHz. With typical sound velocities in liquids of FIGURE 1 Transient acoustic cavitation: the origin of sonochem-
≈1500 m/s, acoustic wavelengths range from roughly 10 istry and sonoluminescence.
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN015L-953 August 2, 2001 12:43
II. SONOLUMINESCENCE
A. Types of Sonoluminescence
Ultrasonic irradiation of liquids can also produce light,
termed “sonoluminescence,” as first observed from water
in 1934 by Frenzel and Schultes. As with sonochemistry,
sonoluminescence derives from acoustic cavitation. There
are two classes of sonoluminescence: multiple-bubble
sonoluminescence (MBSL) and single-bubble sonolumi-
nescence (SBSL). Since cavitation is a nucleated process
and liquids generally contain large numbers particulates
that serve as nuclei, the “cavitation field” generated by a
propagating or standing acoustic wave typically consists FIGURE 5 Typical MBSL spectrum from a metal carbonyl solu-
of very large numbers of interacting bubbles, distributed tion in silicone oil. [From Suslick, K. S. et al. (1999). Phil. Trans.
over an extended region of the liquid. Such cavitation can Roy. Soc. London A, 357, 340. With permission.]
be sufficiently intense to produce MBSL.
For rather specialized but easily obtainable conditions, originated from microdischarge, whereas CN emission is
it is now established that a single, stable gas bubble can typically observed from thermal sources. When oxygen is
✉
be forced into such large amplitude pulsations that it pro- present, emission from excited states of CO2 , CH , and
✉
duces sonoluminescence emissions on each (and every) OH is observed, again similar to flame emission.
acoustic cycle. This phenomenon is called single-bubble Ultrasonic irradiation of volatile organometallics (such
sonoluminescence (SBSL). Under the appropriate condi- as Fe(CO)5 or Cr(CO)6 ) in a low volatility organic liq-
tions, the acoustic force on a bubble can be used to bal- uid produces intense sonoluminescence that corresponds
ance against its buoyancy, holding the bubble stable in to the known atomic emission lines of the metals, again
the liquid by acoustic levitation. This permits examina- analogous to flame emission. Hot-spot temperatures are
tion of the dynamic characteristics of a single cavitating sufficient not only to dissociate all the CO ligands from
bubble in considerable detail, from both a theoretical and the metal complex, but also to produce excited state metal
an experimental perspective. Such a bubble is quite small, atoms. Figure 5 shows a typical MBSL spectrum from a
compared to an acoustic wavelength (e.g., at 20 kHz, the metal carbonyl solution (Cr(CO)6 in this example). Note
maximum bubble size before collapse is ∼50 µm and at the intense line emission from the metal atom excited
minimum during collapse <1 µm). states as well as bands from excited states of the diatomics,
C2 and CH. This metal atom emission provides a useful
spectroscopic thermometer, as described later.
B. Multiple-Bubble Sonoluminescence For both aqueous and nonaqueous liquids, the emission
spectra from MBSL suggests that the principal source of
The sonoluminescence of aqueous solutions has been
light emission is from chemical reactions involving high-
studied extensively over the past 30 years. The spectrum
energy species formed during cavitation by bubble col-
of MBSL in water consists of a peak at 310 nm and a
lapse. MBSL is principally a form of chemiluminescence,
broad continuum throughout the visible region. An inten-
just as flame emission is.
sive study of aqueous MBSL was conducted by Verrall
and Sehgal and later by Didenko. The emission at 310 nm
✉ C. Single-Bubble Sonoluminescence
is from excited-state OH , but the continuum is difficult
to interpret. MBSL from aqueous and alcohol solutions of At the time of this writing, SBSL remains under active
many metal salts have been reported and are characterized investigation with unsettled controversies. Theoretical
by emission from metal atom excited states. interpretations of the experimental findings continue to be
Flint and Suslick reported the first MBSL spectra of refined. It is not yet possible to provide a definitive mech-
organic liquids. With various hydrocarbons, the observed anism for the light emission process, although the most
emission is from excited states of C2 (d3
g − a3
u , the favored model involves compressional heating (though
Swan lines), the same emission seen in flames. Further- probably without a convergent shockwave) of the bubble
more, the ultrasonic irradiation of alkanes in the presence contents, similar to MBSL. The spectra of MBSL and
of N2 (or NH3 or amines) gives emission from CN ex- SBSL, however, show some dramatic differences. While
cited states, but not from N2 excited states. Emission from MBSL is generally dominated by atomic and molec-
N2 excited states would have been expected if the MBSL ular emission lines, SBSL is an essentially featureless
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN015L-953 August 2, 2001 12:43
A. Experimental Design
A variety of devices have been used for ultrasonic irradi-
ation of solutions. There are three general designs in use
presently: the ultrasonic cleaning bath, the direct immer-
sion ultrasonic horn, and flow reactors. The originating
source of the ultrasound is generally a piezoelectric mate-
rial, usually a lead zirconate titanate ceramic (PZT), which
is subjected to a high AC voltage with an ultrasonic fre-
quency (typically 15 to 50 kHz). For industrial use, the
more robust magnetostrictive metal alloys (usually of Ni)
can be used as the core of a solenoid generating an alter-
nating magnetic field with an ultrasonic frequency. The FIGURE 10 A typical sonochemical apparatus with direct immer-
vibrating source is attached to the wall of a cleaning bath, sion ultrasonic horn. Ultrasound can be easily introduced into a
chemical reaction with good control of temperature and ambient
to an amplifying horn, or to the outer surfaces of a flow-
atmosphere. The usual piezoelectric ceramic is PZT, a lead zir-
through tube or diaphragm. conate titanate ceramic.
The ultrasonic cleaning bath is clearly the most acces-
sible source of laboratory ultrasound and has been used
B. Homogeneous Sonochemistry: Bond
successfully for a variety of liquid-solid heterogeneous
Breaking and Radical Formation
sonochemical studies. The low intensity available in these
devices (≈1 W/cm2 ), however, means that even in the case The chemical effect of ultrasound on aqueous solutions
of heterogeneous sonochemistry, an ultrasonic cleaning have been studied for many years. The primary products
bath must be viewed as an apparatus of limited capability. are H2 and H2 O2 ; there is strong evidence for various
✉ ✉
The most intense and reliable source of ultrasound gener- high-energy intermediates, including HO2 , H , OH , and
ally used in the chemical laboratory is the direct immersion perhaps e−(aq) . The work of Riesz and collaborators used
ultrasonic horn (50 to 500 W/cm2 ), as shown in Fig. 10, electron paramagnetic resonance with chemical spin-traps
✉ ✉
which can be used for work under either inert or reactive to demonstrate definitively the generation of H and OH
atmospheres or at moderate pressures (<10 atmospheres). during ultrasonic irradiation, even with clinical sources
These devices are available from several manufacturers at of ultrasound. The extensive work in Henglein’s labora-
modest cost. Commercially available flow-through reac- tory involving aqueous sonochemistry of dissolved gases
tion chambers that will attach to these horns allow the pro- has established clear analogies to combustion processes.
cessing of multi-liter volumes. The acoustic intensities are As one would expect, the sonolysis of water, which pro-
easily and reproducibly variable; the acoustic frequency duces both strong reductants and oxidants, is capable of
is well controlled, albeit fixed (typically at 20 kHz). Since causing secondary oxidation and reduction reactions, as
power levels are quite high, counter-cooling of the reac- often observed. Most recently there has been strong in-
tion solution is essential to provide temperature control. terest shown in the use of ultrasound for remediation of
✉
Large-scale ultrasonic generation in flow-trough configu- low levels of organic contamination of water. The OH
rations is a well-established technology. Liquid process- radicals produced from the sonolysis of water are able to
ing rates of 200 L/min are routinely accessible from a attack essentially all organic compounds (including halo-
variety of modular, in-line designs with acoustic power carbons, pesticides, and nitroaromatics) and through a se-
of ≈20 kW per unit. The industrial uses of these units ries of reactions oxidize them fully. The desirability of
include (1) degassing of liquids, (2) dispersion of solids sonolysis for such remediation lies in its low maintenance
into liquids, (3) emulsification of immiscible liquids, and requirements and the low-energy efficiency of alternative
(4) large-scale cell disruption. methods (e.g., ozonolysis, UV photolysis).
P1: GNH Final Pages
Encyclopedia of Physical Science and Technology EN015L-953 August 2, 2001 12:43
IV. SUMMARY
Luche, J.-L., and Bianchi, C. (1998). “Synthetic Organic Chemistry,” Catalysis” (G. Ertl, H. Knozinger, and J. Weitkamp, eds.), Wiley-VCH,
Kluwer Publishers, Dordrecht, Netherlands. Weinheim, vol. 3, ch. 8.6, pp. 1350–1357.
Mason, T. J. (ed.) (1990, 1991, 1993, 1996, 1999). “Advances in Sono- Suslick, K. S. (1998). Sonochemistry. In “Kirk-Othmer Encyclopedia of
chemistry,” vols. 1–5, JAI Press, New York. Chemical Technology,” 4th edition, Wiley, New York, vol. 26, pp. 517–
Putterman, S. J., and Weninger, K. R. (2000). Sonoluminescence: How 541.
bubbles turn sound into light. Annu. Rev. Fluid Mech. 32, 445– Suslick, K. S., Didenko, Y., Fang, M. M., Hyeon, T., Kolbeck, K. J.,
476. McNamara III, W. B., Mdleleni, M. M., and Wong, M. (1999). “Acous-
Suslick, K. S., and Crum, L. A. (1997). Sonochemistry and sonolumi- tic cavitation and its chemical consequences,” Phil. Trans. Roy. Soc.
nescence. In “Encyclopedia of Acoustics” (M. J. Crocker, ed.), Wiley- London A 357, 335–353.
Interscience, New York, vol. 1, ch. 26, pp. 271–282. Suslick, K. S., and Price, G. (1999). “Applications of ultrasound to ma-
Suslick, K. S. (1997). Sonocatalysis. In “Handbook of Heterogeneous terials chemistry,” Annu. Rev. Matl. Sci. 29, 295–326.
P1: GTY/GRD P2: GRB Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN015K-957 August 2, 2001 15:47
I. History
II. Forensic Spectroscopy Techniques
III. Conclusion
637
P1: GTY/GRD P2: GRB Final Pages
Encyclopedia of Physical Science and Technology EN015K-957 August 2, 2001 15:47
(MS), nuclear magnetic resonance (NMR), or energy dis- detector and computer that evaluate a spectrum of visible
persive x-ray spectrometry (EDX or EDS). radiation.
Forensic is a term generally associated with public Visual spectroscopy can also be semiquantitative. A
speaking or formal debate. It refers to speaking before simple example is the assessment of swimming pool water
the law or court as it is used in this article and is related quality and its chemical content using well-known color
to both civil or tort and criminal law. It has become the reactions and simple sets of colored concentration refer-
habit of some to restrict the use of the terms such as foren- ence standards. This evaluation of light or color intensity
sic science or forensic spectroscopy to criminal cases. In is called photometry. Visual photometry saw one early use
fact, criminal forensic science is one of the smaller divi- in criminal law with the evaluation of breath alcohol con-
sions within the forensic science community. The major- centrations. The test was based on the ability of breath
ity of forensic science is practiced in other fields. Some alcohol to chemically reduce the colored chromate ion to
of them include occupational health and safety, environ- a colorless form in aqueous solutions. As more alcohol
mental protection and remediation, personal injury claims vapor was bubbled through a chromate solution it became
resolution, medical malpractice investigations, and envi- more bleached. The resulting solution was then visually
ronmental toxicology studies, fine art validation and valu- compared to a calibrated, standard set of diluted chromate
ation, nuclear safety violation analysis, nuclear materials solutions to determine the subject’s breath alcohol con-
diversion investigations, and also commercial product fail- tent. The test was soon modified to include instrumental
ure analysis. These topical areas do not begin to address measurement of the decolorizing reaction and is referred
the rather large areas of social and psychological forensic to as a form of spectrophotometry.
science and medicine. Any of these fields can find some Photographic film was also used as a detector in spec-
form of spectroscopy or spectrometry to be of use. troscopy in order to extend spectral evaluations beyond the
limits of the eye. Photographic film is an effective detec-
tor but its spectral sensitivity variations, processing, and
I. HISTORY dimensional instability make it somewhat cumbersome to
use. The inclusion of photomultiplier detectors and new
Many of the initial methods scientists used to evaluate radiation sources in the evaluation of spectra opened the
materials relied on color development or color changes way to more easily evaluating materials beyond the vis-
to assist them in the identification of drugs, poisons, ex- ible light region of the radiation spectrum by also using
plosives, metals, body fluids, plant extracts, and unknown invisible ultraviolet (UV) and infrared (IR) radiation. A
organic and inorganic compounds in general. A detailed very general block diagram of a spectroscopy system is
summary of such tests was included as a 368-page adden- given in Fig. 1.
dum in the 1940, fifth edition of the Merck Index. Quite a Visible and UV radiation is of relatively short wave-
few of those tests were developed in the 1800s and many length and high energy and generally interacts with the
find use today. electrons around atoms. Infrared radiation is longer wave-
Spectroscopy is defined as the evaluation or study of length and weaker. It generally effects the bonds between
the production and measurement of electromagnetic radi- atoms in molecules. Information can be gathered about
ation spectra, associated with the emission or absorption the nature or environments of those electrons and bonds
of radiant energy (light) by different materials. It is a di- by evaluating the energy they absorb or emit. The fact that
verse topic and can be as simple as the assessment of the elements, atoms, and molecules can absorb or emit differ-
freshness or ripeness of fruit by visual inspection of color. ent wavelengths of light or radiation has led scientists to
Such an assessment might be the foundation of a produce develop a wide array of spectroscopy instruments. Each
merchant’s insurance claim. More complex visual spec- instrument provides an additional bit of information to the
troscopy might include the evaluation of chemical color forensic spectroscopist about the material that they may
tests that absorb light or flame color tests of materials that be testing.
emit light when held in flames on platinum wires. It could
also include the colors of materials fused in beads of borax,
or the color observations of fine powders formed by scrap-
ing a material across an unglazed, white porcelain plate.
Both are types of tests which selectively absorb some col-
ors of light and leave other colors to be seen. Some of
these tests are still used to evaluate minerals, drugs, met-
als, and general classes of organic and inorganic mate- FIGURE 1 Basic components of a spectroscopy system or
rials. In these cases, the eye and brain are acting as the instrument.
P1: GTY/GRD P2: GRB Final Pages
Encyclopedia of Physical Science and Technology EN015K-957 August 2, 2001 15:47
Most spectroscopical analysis is a data intensive pro- lations where the fraudulent material has been painted or
cess. A single sample scan across a wavelength region of dyed to mimic the protected or original product. Different
interest can produce a two- or three-dimensional set of colors or color mixtures that might fool the eye may be
data containing 2000 or 3000 data points. A few dozen easily distinguished with the instrument.
scans of that sample can yield a data set that was quite
cumbersome to handle just a few years ago. It is essential
B. Ultraviolet Spectroscopy (UVS)
to acknowledge that all forms of spectroscopy and spec-
trometry have been greatly enhanced by the development Ultraviolet spectroscopy requires somewhat more costly
of personal computers. When Fourier transform IR spec- quartz glass sample holders and quartz or fluoride optical
trometers were first commercialized they filled a room. components as well as a different light source than VISS
Today they can fit into a suitcase and be operated in the but has become equally accessible and easy to use. The
heart of the wilderness. To be sure, there are trade-offs to technique is applied to the evaluation of samples that may
be made in instrument performance within these extremes, contain aromatic compounds such as benzene derivatives
but computer miniaturization has played a significant part or to a wide variety of UV absorbing compounds that are
in this development. used to protect painted coatings, human skin, and fabrics.
The balance of this article will offer a look at a va- Other compounds that are used to whiten fibers and fabrics
riety of spectroscopical instruments, the materials they by producing blue-white fluorescence under outdoor illu-
can be applied to, and the types of information they can mination are called fluorescent brightners and may also be
provide. analyzed using UVS and UV fluorescence spectroscopy.
UVS is also used to detect and quantify compounds that
are separated from complex mixtures during liquid chro-
matography (LC). This method can separate and detect
II. FORENSIC SPECTROSCOPY
compounds that are easily destroyed by other techniques.
TECHNIQUES
LC is also applied to the evaluation of environmental sam-
ples for contamination, drug mixtures including steroids,
A. Visible Spectroscopy
foods, dye mixtures, and complex flavoring agents.
Instrumental visible spectroscopy (VISS) is a logical ex- The application of UVS has declined from its earlier use
tension of the color tests and visual observations that were in the criminal analysis of drugs as other more powerful
mentioned earlier. It is a sensitive, accurate method of instruments such as gas chromatograph–mass spectrome-
measuring the colors or mixtures of colors that our eyes ters (GC–MS) have become less expensive and simpler to
perceive and removes the obvious subjectivity and visual operate.
anomalies associated with human vision. It found early
and continued use in part because it used simple light
C. Fluorescence Spectroscopy (FS)
sources and its sample containers and optical systems
could be made of common glass. Fluorescence spectroscopy instruments illuminate sam-
Scientists have known since the early 1900s that they ples with short wavelength light in the visible or ultravi-
could measure color differences that were too small for the olet wavelength regions. A sample atom’s electrons may
eye to distinguish. Since then, VISS has found wide use in absorb a portion of the incident light and move to higher
the forensic analysis of materials and systems as diverse energy levels. As that empty energy level is refilled with
as pigments, dyes, and chemical reaction color changes. another electron, the atom may emit a lower energy/longer
It is also used in studies of fluorescence, atmospheric con- wavelength light photon that is characteristic of the sam-
tamination, video monitor color correction, toxicological ple molecule’s structure. The emitted light is called fluo-
variations in red blood cells, photographic color control, rescence. Some of the uses of FS were described in the
and window film color density measurements, to mention previous section.
just a few. This versatile technique can be applied to a wide FS instrumentation tends to be more complex than UV
range of sample types from solids and liquids to pastes or systems, with monochromators in both illuminating and
gels and the samples can be viewed by either transmitted detector beam paths and, sometimes, dual detectors. These
or reflected light. features are necessary to separate rather intense illuminat-
VISS is an essential element of color comparisons in ing beams scattered radiation from the sometimes weaker
criminal cases involving paint and fiber transfers due to its molecular fluorescence.
power to distinguish colors the human eye might mistak- Fluorescence spectroscopy was once used to identify
enly find indistinguishable. Civil investigations also find and quantify difficult to analyze materials like morphine
use of VISS in establishing tampering or trademark vio- that was extracted from body fluids or lysergic acid diethyl
P1: GTY/GRD P2: GRB Final Pages
Encyclopedia of Physical Science and Technology EN015K-957 August 2, 2001 15:47
amide (LSD) recovered from paper or tablet dosage units. will not be detectable. By comparison, the motion of a
It still provides methods for evaluating the effluent of carbon–hydrogen bond is generally easily detected. A sec-
manufacturing processes such as fiber/fabric dyeing, pa- ond limitation lays in the fact that components of a mixture
per production, waste recovery, landfills, and fuel storage need to be present in excess of about 1–5% if they are to
facilities. be detectable.
The instruments tend to be sensitive to temperature, Finally, an IR instrument must be able to “see” the item
sample acidity and dissolved ion content, agents that under test. Waves of radiation are not affected by particles
quench or suppress fluorescence and need to operated or samples that are much smaller than the wavelength of
under laboratory conditions. Newer instruments have re- the radiation. One might note that a beach ball or a sea gull
placed FS in many of its historical applications but it is has little affect on an ocean wave and a vacationer stand-
reemerging and gaining in use within the forensic science ing in shore counting waves would not know if either of
community. This is due to the fact that DNA-based analy- those items was floating a hundred yards off shore. When
sis technology is being applied to an ever-increasing group a scientist analyzes radiation that has passed by a sample,
of materials, and fluorescent dyes are used to detect those he or she can generally not detect (“see”) the sample if
material’s presence in a variety of samples. the illuminating radiation did not interact with it. Much
It is common to assume that DNA technology is only of the important information in an IR spectrum resides in
applied to identifying biological individuals. In fact, that the 10- to 40-µm or 10,000- to 40,000-nm region of the
same technology can be, and is being, modified for the light spectrum. These facts generally limit IRS analysis to
analysis of drugs, pollutants, and toxicology samples. Flu- samples that are at least 10 µm in size and are mentioned
orescent compounds are used as tags that make sample here because they are not usually a factor in other types
components detectable and a wide variety of spectroscopy, of spectroscopy.
photography and photometry systems are used to record Infrared spectra can be produced from samples that are
the analytical results. imaged by microscopes at distances of less than a millime-
ter or by telescopes at distances of a kilometer or more. It
is used to analyze samples ranging from drug crystals and
D. Infrared Spectroscopy (IRS)
thin paint layers to atmospheric pollutants.
Infrared spectroscopy is also called molecular spec-
troscopy because it provides information about a
2. Raman Spectroscopy
molecule’s structure or of a mixture of molecules that
are present in the sample. It is one of the most versa- Raman spectroscopy (RS) can be considered a second
tile and easily applied forms of spectroscopical analysis. form of infrared analysis that illuminates samples with
It has been advanced by the development of a host of sam- short wavelength (not infrared) light and then analyzes
pling assessories, detector types, and technological ad- the light that is scattered at right angles to the illumi-
vances that allow its application to all four of the common nating beam. A laser generally provides illumination and
forms of matter: solids, liquids, gases, and plasmas. Its use molecules in the laser beam absorb light as they vibrate
in plasma sample analysis in forensic cases is not routine and shift the wavelength of some of the scattered illumi-
but it is one of the more popular forms of spectroscopy nating light by wavelengths comparable to IR absorption
used in forensic science. bands. These are referred to as Raman shifts or Raman
spectra. These spectra are not to be confused with fluo-
rescence spectra that are produced by a different process.
1. Infrared Absorption Spectroscopy
This form of IR spectroscopy has the advantage of being
Infrared spectroscopy (IRS) is the study or analysis of how able to analyze symmetrical molecules and bonds that do
the bonds between atoms of a molecule absorb energy and not exhibit permanent charge separations or dipoles and
vibrate or bend (deform) much like springs. The energy is quite complementary to IRS. It also allows the analysis
necessary to cause motions in these bonds is quite specific of Raman shifts and IR spectra that may be outside the
and provides the forensic scientist with information about detection range of a particular IR detector.
how atoms and combinations of atoms are associated in a Early Raman spectroscopy instruments saw little foren-
molecule. sic use. Illumination sources were mercury vapor lamps
There are some basic requirements for absorption IRS. with filters that isolated specific mercury radiation wave-
One is that a vibration or deformation must be associated lengths. Sample sizes were rather large and Raman spec-
with a permanent charge separation or electric dipole if it tra were captured on photographic film. The detection of
is to be detected. Vibrations between atoms in molecular weak spectral lines sometimes required days of opera-
groups such as the eight-atom sulfur ring in pure sulfur tion and film exposure. Today’s instruments’ use of laser
P1: GTY/GRD P2: GRB Final Pages
Encyclopedia of Physical Science and Technology EN015K-957 August 2, 2001 15:47
illumination, Fourier transform dispersive systems, new an investigator to characterize the materials that may have
detectors, a variety of sampling devices, and computers contributed to their presence. The primary forensic spec-
have all combined to make Raman spectroscopy a versa- troscopy advantage of AS lays in its improved sensitivity
tile analytical tool. over molecular spectroscopy and its ability to detect very
low concentrations of elements in a complex mixture such
as soil or paint.
E. Light Detection and Ranging
This sensitivity has led to AS use for the detection and
(LIDAR) Spectroscopy
identification of minor elements in samples of glass and to
The fluorescence analysis technique called LIDAR cur- the confirmation or exclusion of tracer elements. Tracer
rently requires that specific sample species be identified elements are added to materials such as polymers, pig-
for analysis. Selected wavelengths of illumination are sent ments, or glasses and serve as atomic trade marks for a
out through a telescope toward a target area and used to specific manufacturer’s product. Tracers are typically low
excite molecules of the selected compounds. The resulting concentrations of rare earth elements. These tests can be
fluorescence of excited molecules/compounds/species is essential in helping forensic scientists discriminate other-
then collected by a high-performance telescope and de- wise similar materials in product based lawsuits where a
tector system. The resulting spectrum is analyzed to de- manufacturer claims that he or she did not make the prod-
termine their concentration of fluorescent species and can uct in question. A few basic techniques of AS are in use
be used to evaluate atmospheric pollution levels. and each can have a few variations.
A second form of LIDAR, known as Differential Ab-
sorption LIDAR (DIAL), uses two laser beams, one se-
1. Atomic Emission Spectroscopy (AES)
lected to be absorbed by a particular gas or vapor and
a second as an unabsorbed reference. Pulses of the two Early AES instruments were rather straightforward and
lasers are then aimed and focused at a distant area through excited the atoms in a sample by placing the sample in
a telescope. Gas molecules in that area reflect some of a carbon arc. The resulting light was passed through a
both of the laser beams back to the telescope where they diffracting prism, or across a diffraction grating, and the
are detected and analyzed to determine if the gas or vapor resulting light spectrum was recorded on a strip of photo-
of interest is present and, if it is, in what concentration. graphic film. This film was developed and placed in a pro-
Areas of pollution, clandestine laboratory operations jection system where the recorded spectral lines were com-
and the wind drift of pollutants can be mapped by this pared with a master film to identify the elements present
technique. It can be a particularly powerful tool in foren- in the sample.
sic investigations because it is a non-intrusive device These systems were effective and some are still in use.
that can search the atmosphere around a suspected cite Their major drawbacks lay in some tedium in sample
of illicit chemical manufacturing, storage, pollution or preparation, film development and spectral analysis. The
disposal without requiring direct access to the subject recording film also limited the dynamic range of the sys-
site. tem (the range of concentrations that could be detected in
a single sample) and trace elemental analysis was dif-
ficult. These early systems have largely been replaced
F. Atomic Spectroscopy (AS)
by a variety of more sensitive, wider range, computer-
Atomic spectroscopy refers to a wide range of techniques based equipment. All of the newer equipment is also
that are applied to materials that have been reduced to much more expensive to purchase and operate and requires
atomic or ionic forms or species in an electric arc, flame, careful handling and preparation of samples to avoid
or plasma. When one speaks of atomic spectra, molecules contamination.
and compounds that have been of discussion up to now are
no longer an issue. AS is in fact a monumental extension
2. Inductively Coupled Plasma-Optical Emission
and improvement upon the flame color tests that were
Spectroscopy (ICP-OES)
mentioned at the beginning of this article.
Molecules and compounds that have been irradiated ICP–OES (or ICP-Atomic Emission Spectroscopy, ICP-
with a burst of sufficient energy will decompose into their AES) systems excite atoms to emit analytically useful ra-
constituent atoms. This energy can also cause the elec- diation using plasmas with temperatures of about 1000◦ C.
trons around the atoms to move into higher energy states Samples are first fused, dissolved in a solvent and then
with a resulting absorption and later emission of specific aspirated into the plasma flame. The element specific
wavelengths of light. Atomic spectroscopy uses these ex- light emitted by the plasma is then viewed either radially
changes of energy to identify atomic species and to allow (through the side of the plasma) or axially (through the
P1: GTY/GRD P2: GRB Final Pages
Encyclopedia of Physical Science and Technology EN015K-957 August 2, 2001 15:47
length of the plasma) plume and viewed or analyzed by a tive absorption or emission of different wavelengths of
multi-element spectrometer and detector. light. Photometry is used to determine either the absolute
The technique is similar to atomic absorption spec- amount or the relative amounts of two or more compounds
troscopy in sensitivity but its ability to detect several ele- or elements in a sample or set of samples. In either case,
ments at a time improves its utility in forensic analysis. A SP requires the careful calibration and validation of the
forensic scientist seldom knows what possible elements spectroscopy system and detector.
that might be found in a sample and multiple element SP is an essential element of many forensic comparisons
detection is an asset. ICP has been made even more sen- of spectra, especially those involving mixtures of light
sitive, especially to metals, by coupling the plasma plume absorbing materials such as dyes and pigments. Fabrics
or flame to a mass spectrometer (ICP-MS). and carpets are an example of materials that are often
colored with a mixture of dyes and the relative proportion
of dyes absorbed by a fabric may vary from dye lot to dye
3. Flame Atomic Absorption Spectroscopy (AAS)
lot. It is necessary to establish not only that the dyes in
Atomic absorption spectroscopy instruments place a sam- two items are similar but also that they are present in the
ple in a high temperature flame that yields atomic species same amounts if the two items are to be associated.
and passes selected, element specific, illumination through Environmental and occupational laws, regulations, and
the flame to detect what wavelengths of light the sample penalties are often based not only on the presence of dan-
atoms absorb. Either acetylene or nitrous oxide fuels the gerous materials but also on the amounts of those mate-
analytical flame. This process again demands that solid rials that are present, making SP and essential part of an
samples be digested (dissolved in an acid or fused with a analysis.
salt) and dissolved to form a solution that can be aspirated
or sprayed into the instrument’s flame, while protecting
H. Microscopical Spectrophotometry (MSP)
the sample material from contamination or adulteration.
There are some constraints that have limited the ap- Microscopy has been applied to a number of spectroscopy
plication of AAS to general unknown analyses in forensic techniques including IR, UV, VIS, and fluorescence spec-
laboratories. The primary one is that the analyst must have troscopy that have been previously discussed. MSP is often
some knowledge of the sample’s makeup and which ele- referred to as “microspectroscopy” and involves concen-
ments are of analytical interest so that the proper lamp trating the illuminating beam onto the sample through a
can be selected to illuminate the sample. Just one or a few microscope and then using another microscope objective
elements can be analyzed in any given sample run and a to collect the radiation leaving the sample. It is not a suit-
crucial element might be missed. able sampling system for techniques like AES, AAS, or
The power of AAS lays in its sensitivity and ability to ICP-OES that have large analytical volumes such as flames
detect trace elements and make it possible to distinguish or plasmas.
otherwise identical materials. This can be a critical issue MSP is especially useful when sample sizes are ex-
in associating an individual with a crime scene or a cor- tremely small, including samples such as a single fiber,
poration with a damaged environmental site. the edge of a multilayered paint film, a single line of ink,
or an individual pigment particle. The application of spec-
troscopy to these types of samples has greatly increased
4. Graphite Furnace Atomic Absorption
scientists’ ability to discriminate samples that are difficult
Spectroscopy (GFAAS)
or impossible to visually distinguish. This technique adds
GFAAS is similar to AAS in that the instrument measures cost and complexity to the analytical instrument and forces
light that the sample’s atoms absorb. Samples are raised to the analyst to have a basic and functional understanding
analytical temperature in a heated graphite tube or furnace. of microscopy and its effects on light beams if he or she
The technique is somewhat easier and less costly to operate is to produce useful results.
than flame AAS and some instruments can be set up to
analyze multiple elements simultaneously, but GFAAS is
about 10 to 100 times less sensitive than flame AAS. III. CONCLUSION
so the analytical results assist the courts and juries in their Brode, W. (1943). “Chemical Spectorscopy,” Wiley, New York.
search for facts and truth. Candler, C. (1949). “Practical Spectroscopy,” Hilger and Watts,
Glasgow.
Horwitz, W. (ed.) (2000). “Official Methods of Analysis of the Associa-
SEE ALSO THE FOLLOWING ARTICLES tion of Official Analytical Chemists,” Association of Official Analyt-
ical Chemists (AOAC), Washington, DC.
COLOR SCIENCE • DNA TESTING IN FORENSIC SCIENCE Humecki, H. (1995). “Practical Guide to Infrared Microspectroscopy,”
• GAS CHROMATOGRAPHY • INFRARED SPECTROSCOPY Mercel Dekker, New York.
Saferstein, R. (ed.) (2001). “Forensic Science Handbook, Volume I,”
• LIQUID CHROMATOGRAPHY • MASS SPECTROMETRY IN Regents/Prentice Hall, Englewood Cliffs, NJ.
FORENSIC SCIENCE • RAMAN SPECTROSCOPY • TOXICOL- Saferstein, R. (ed.) (1993). “Forensic Science Handbook, Volume III,”
OGY IN FORENSIC SCIENCE Regents/Prentice-Hall, Englewood Cliffs, NJ.
Siegel, J., Saukko, P., and Knupfer, G. (eds.) (2000). “Ency-
clopedia of Forensic Sciences,” Academic Press, San Diego,
BIBLIOGRAPHY CA.
Sibilia, J. (1988). “A Guide to Materials Characterization and Chemical
American Society of Testing and Materials (2000). “Metals Test Meth- Analysis,” VCH Publishers, New York.
ods and Analytical Procedures,” Annual Book of ASTM Standards, Williams, D., and Fleming, I. (1995). “Spectroscopic Methods in
Volume 03.06, ASTM, PA. Organic Chemistry,” McGraw-Hill, London.
American Society of Testing and Materials (2000). “General Meth- Workman, J., and Springsteen, A. (eds.) (1998). “Applied Spectroscopy:
ods and Instrumentation,” Annual Book of ASTM Standards, Volume A Compact Reference for Practitioners,” Academic Press, San Diego,
14.02, ASTM, PA. CA.
P1: GPJ/GJK Final Pages P2: GSS Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
Thermal Analysis
David Dollimore, deceased
University of Toledo
I. Introduction
II. Differential Scanning Calorimetry and
Differential Thermal Analysis
III. Thermogravimetry
IV. Evolved Gas Analysis
V. Thermomechanical Methods
VI. Modulation Techniques
VII. Simultaneous Techniques
VIII. The Role of the Computer
IX. Application of Thermal Analysis Techniques
591
P1: GPJ/GJK Final Pages P2: GSS
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
should be noted that this group must by definition include TABLE II Recommended Terminology for Some Thermal
the measurement of the density of samples subjected to a Analysis Techniques
programmed temperature variation. A recently introduced Property
related technique is dynamic mechanical analysis, used measured Technique name Abbreviation
to study the viscoelastic response of a sample under an
Mass Thermogravimetry TG
oscillatory load. Other techniques are also used where
Isobaric mass-change
the sample is under an oscillatory load. determination
A recent sophisticated approach is to modulate the tem- Isothermal mass-change
perature about a predetermined overall heating rate. Those determination
who use these techniques must be subjected to a disci- Evolved gas detection EGD
pline which needs definitions and conventions so that the Evolved gas analysis EGA
reader may understand the technique being utilized. In Derivative thermogravimetry DTG
commercial equipment these techniques are often com- Emanation thermal analysis
bined so that, in these simultaneous techniques, one mate- Thermoparticulate analysis TPA
rial is subjected to two or even three measurement probes. Temperature Cooling curvea
Generally, thermal analysis techniques may be classified Heating curve
into three groups depending upon the way in which the Differential thermal analysis DTA
physical property needs to be recorded. Enthalpy Differential scanning DSC
calorimetry
1. The absolute value of the property itself can be Dimensions Thermodilatometry TDA
measured, for example, the sample mass. Mechanical Thermomechanical analysis TMA
2. The differential method measures the difference Dynamic mechanical analysis DMA
between some property of the sample and that of a a Cooling curve was initially called thermal analysis. It is still called
standard material, for example, their temperature thermal analysis in many physical chemistry textbooks. Other techniques
difference. of thermal analysis measure acoustic, optical, electrical, and magnetic
3. The rate at which the property changes with characteristics versus temperature. The list is not complete, as new tech-
temperature can be measured. These form the basis of niques are constantly being added.
derivative measurements and very often may be
interpreted on a kinetic basis. measured against temperature is indicated in Table III.
This list is not complete, as new techniques are continually
There exist national and international organizations that appearing.
recommend nomenclature abbreviations and standards, The definitions follow from Tables II and III.
and these organizations have set up committees that have Thermogravimetry (TG) is the most widely used tech-
formulated, in particular, a system of nomenclature which nique of thermal analysis, in which the mass of a substance
is adhered to in this chapter but is not always adhered to is measured as a function of the temperature while it is sub-
in journals and in certain other fields of science. These jected to a controlled temperature program. The record on
recommended nomenclatures and abbreviations are con- the thermogravimetric, or TG, curve is the mass plotted
stantly under review, so the reader is advised to seek the against the temperature (T ) or time (t) if the variation of
most up-to-date recommendations in the literature. temperature with time can be indicated as well on the same
graph. In solid-state decomposition reactions, the reactant
material degrades, often to be replaced by the solid prod-
B. Nomenclature
uct. An example of this is the decomposition of limestone
The recommended nomenclature has been put forward to quicklime.
by the International Nomenclature Committee of the In the record of the mass of the solid residue against
International Confederation for Thermal Analysis and the temperature, the decomposition of the material can be
Calorimetry (ICTAC). These recommendations are widely followed. This may be plotted in alternative ways, as the
circulated in publications of the confederation and the percentage mass loss or the fractional mass loss versus
International Union of Pure and Applied Chemistry the temperature or as the fractional decomposition versus
(IUPAC). The most widely publicized report is the book- the temperature.
let “For Better Thermal Analysis,” which is continually Derivative thermogravimetry (DTG) is not really a sep-
updated as required. arate technique but involves plotting the first derivative
The recommended abbreviations of various techniques of the TG with respect to either time or temperature.
in thermal analysis are listed in Table II. The property The DTG curve is plotted with the rate of mass loss on
P1: GPJ/GJK Final Pages P2: GSS
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
TABLE III Property Measured versus Temperature in Some Thermal Analysis Techniques
Technique Property measured vs temperature Instrument
Thermogravimetry (TG) and derivative Mass and derivative of mass with respect to temperature Thermobalance
thermogravimetry (DTG)
Differential thermal analysis (DTA) Difference in temperature between reference and sample cell DTA apparatus
Differential scanning calorimetry (DSC) Enthalpy (heat) flow DSC apparatus
Evolved gas detection and analysis (EGD and EGA) Various properties of gas Method varies but should
always be indicated
Thermodilatometry (TDA) Length of volume Dilatometer
Thermomechanical analysis (TMA) Expanion under applied stress (load) Adapted dilatometer
Dynamic mechanical analysis (DMA) Frequency response under oscillatory stress DMA apparatus
the ordinate, plotted downward, and the temperature or perature difference between a substance and a reference
time on the abscissa, increasing from left to right. This material is measured as a function of temperature while
can be achieved by an analysis of the TG curve as a the substance and reference material are subjected to a
separate operation in the dedicated computer part of the controlled program. The plot is called a DTA curve; the
equipment. temperature difference T should be plotted on the or-
Isobaric mass-change determinations refer to the equi- dinate, with the endotherm processes shown downward
librium mass of a substance at a constant partial pressure and the exotherm processes in the opposite direction, and
of the volatile product(s) measured as a function of tem- the temperature or time on the abscissa, increasing from
perature while the substance is subjected to a controlled left to right. This technique is often applied quantitatively
temperature program. when the area of the peaks can be made proportional to the
Evolved gas analysis (EGA) is a technique in which quantity of the material decomposing or to the enthalpy of
the gas evolved from a substance subjected to a controlled the process. In this respect the equipment then serves as a
temperature program is analyzed. The method of analysis calorimeter. The term DSC is applied to such experiments.
should always be noted. There are two types of equipment in which the back-
Emanation thermal analysis is a technique in which the ground temperature of the calorimeter is raised through
release of radioactive emanation from a system is mea- a programmed temperature regime being imposed on the
sured as a function of temperature while the system is system. These are power-compensation DSC and heat-
subjected to a controlled temperature program. flux DSC. The method identified as power-compensation
Thermoparticulate analysis (TPA) is a technique in DSC was originally a copyright term employed by one of
which the evolved particulate material in the evolved gases the instrument manufacturers, and often the term DSC is
is measured as a function of temperature. found in the literature applying just to power compensa-
Another group of techniques involves the measurement tion equipment.
of enthalpy changes. It has already been mentioned that, The members of the Nomenclature Committee have
historically and in physical chemistry textbooks, the term considered the distinction between quantitative DTA and
thermal analysis applies to the determination of cooling or heat-flux DSC: in their opinion, a system with a multiple
heating curves: these are techniques in which the temper- sensor (e.g., a Calvert-type arrangement) or with a con-
ature of a substance is measured as the substance is either trolled heat leak (a Boersma-type arrangement) should
cooled down or allowed to heat up. The cooling curve is be called heat-flux DSC. In practice, if the instrument
the normal technique, and again, it may be reported as the manufacturers can show that the system operates as a
first derivative of the heating curve with respect to time calorimeter over a programmed temperature range, then
from the raw experimental data while the substance is sub- they describe the equipment as DSC.
jected to a controlled temperature regime against time. The The usual method of plotting DTA results is, as already
function dT /dt should be plotted on the ordinate, and T noted, with endothermic peaks shown downward on the
or t on the abscissa, increasing from left to right. It is im- plot and exothermic peaks shown in an upward direction.
portant to state whether the technique involves a cooling However, because a DSC is considered to measure thermo-
process or a heating process. dynamic quantities directly, the DSC plots are often found
The two techniques differential thermal analysis (DTA) with the endothermic plots in an upward direction and
and differential scanning calorimetry (DSC) should be the exothermic plots in a downward direction. This con-
considered together. DTA is a technique in which the tem- forms with the IUPAC requirements for the presentation
P1: GPJ/GJK Final Pages P2: GSS
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
represents the glass transition temperature, Tc TABLE IV Materials that Can Be Used for Temperature Cali-
represents the temperature of crystallization, bration in Thermal Analysisa
Tm represents the melting temperature, and Tt Peak temperature
represents the temperature of a solid-state Material Transition type (◦ C)
transition.
Polystyrene Glass transition ∼101
c. Where the subscript relates to a specific point in
1-2 Dichloroethane m.p. −32
time or a point on the curve, it should be a
Cyclohexane Transition point −83
lowercase letter or a number, e.g., Ti represents the
m.p. +7
initial temperature, t0.5 represents the time at
Phenyl ether m.p. 30
which the fraction reacted is 0.5, T0.3 represents
o-Terphenyl m.p. 58
the temperature at which the fraction reacted is
Potassium nitrate Transition point 128
0.3, Tp represents the temperature of the peak in
Indium m.p 157
DTA or DSC, and Tc represents the temperature of
Tin m.p. 232
the extrapolated onset. This can also be applied to
Potassium perchlorate Transition point 300
DTG techniques.
Silver sulfate Transition point 430
Quartz Transition point 573
D. Standardization Potassium sulfate Transition point 583
Potassium chromate Transition point 665
No single instrument design or set of experimental con-
Barium carbonate Transition point 810
ditions is optimum for all studies. The techniques are
Strontium carbonate Transition point 925
dynamic in nature and flexible in use but produce data
which may be highly dependent upon the procedure. The a These temperature calibration materials are supplied by instrument
prime requirements for standardization can be listed as manufacturers on request. The essential condition is that the material
follows. should be pure. The above data are generally available for determination
of temperature on DTA or DSC equipment.
TABLE V Enthalpies of Fusion for Selected TABLE VII Fusible Link Method for Calibrating
Materials a TG Unit
equipment very often modify the commercial units to suit The temperature control unit can be a simple furnace
their particular needs. and programmer. The fact that in some cases the instru-
mentation operates a control of the furnace temperature,
There are, of course, other data that should be reported rather than the sample temperature, is an important design
for specific techniques, and these are mentioned when feature which has already been noted. The recording unit
dealing with these techniques. receives a signal from the measuring unit and the tem-
perature thermocouple. In most cases this is fed into a
F. Components of Thermal dedicated computer, and this allows the signals to be dis-
Analysis Instrumentation played in a variety of different ways, depending on the
requirements of the operator.
The thermal analysis equipment enables the sample to be In general, thermal analysis instruments may be divided
heated at a predetermined rate so that its temperature and into two groups: differential instruments, which contain
one or more of its physical properties can be continuously the sample and a reference cell in similar environments and
measured and recorded. There are, therefore, three basic provide a difference signal of the properties; and derivative
units involved: first, the measuring unit; second, the tem- instruments, which note the change in the property signal
perature control unit; and third, the recording unit. The versus the temperature as a proper derivative signal. How-
simple arrangement of these components is given in Fig. 1. ever, it must be noted that the latter type of device is no
There is a measuring unit in all these systems and this must longer required, as the property signal can be fed into a
be in a particular position in the furnace. There must also dedicated computer and the dedicated computer can now
be a system for passing a controlled atmosphere around do the derivative calculations and provide a curve of both
the sample and a thermocouple proximate to the sample the property versus the temperature and the rate of change
so that the temperature is accurately measured. The sam- of that property versus the temperature.
ple atmosphere is very important in liquid gas systems or
in chemical reactions, and therefore, the measuring unit
should be capable of operating under inert reactive gas 1. Furnace Design
or vacuum conditions. One point that is missed by most Furnaces have to be designed for particular applications
instrument manufacturers is the fact that the so-called vac- since they have to be compatible with the measuring sys-
uum systems require continuous pumping to eliminate the tem and also with problems associated with each tech-
leak of gas into the system through various parts of the nique, for example, convection currents in TG. However,
equipment. Units which hold their vacuum or which op- there are some general observations that can be made,
erate at controlled pressures other than atmospheric are first, regarding the thermal capacity of the furnace. If a
built only by specialist instrumentation companies. large furnace is used, then one is going to have a range
within the furnace at which a uniform temperature can be
recorded. Small furnaces, however, will not have this uni-
form range of temperature, and the positioning of the sam-
ple in smaller furnaces becomes quite important. The sec-
ond point about the size of the furnace is that large furnaces
will take considerable time to reach a particular tempera-
ture and also take some time to cool down. The smaller the
furnace, the easier it is to cool the temperature of the fur-
nace back to ambient. The sample and reference material
in differential measurements must also be subjected to the
same temperature change. This generally involves a design
feature involving both the unit in which the measurements
are made and the furnace itself. One further point about
furnace size and shape is that a long narrow furnace will
generally give a larger uniform hot zone than a short wide
furnace of similar volume. All instrumentation furnaces
are electrically powered (although high-frequency induc-
tive and infrared heating furnaces have been put on the
market). The resistance wire is generally coiled around an
insulating packing. The outside of the furnace is generally
FIGURE 1 Component parts of a thermal analysis unit. well insulated, although this has a bearing on the rate at
P1: GPJ/GJK Final Pages P2: GSS
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
TABLE VIII Upper Temperature Limits for Common gramming from ambient to a preset maximum temper-
Furnace Resistance Elements ature followed by the equipment being either switched
Furnace Temperature off or programmed to cool. In some units natural cooling
winding(s) limit (◦ C) Atmospherea is used; in others there is the additional choice of main-
taining the maximum temperature isothermally. The pro-
Nichrome 1000 A
grammed linear cooling process can often be allowed to go
Chromel A 1100 A
to a preset minimum temperature, followed by the equip-
Tantalum 1330 B
ment being switched off, by programmed heating, or by
Kanthal 1350 A
maintenance this minimum temperature isothermally. In
Platinum 1400 C
most modern equipment, cycling at a given rate between
Platinum–10% rhodium 1500 C
two preset temperatures can be achieved. It has also been
Platinum–20% rhodium 1500 C
found to be convenient to operate the equipment isother-
Kanthal super 1700 A
mally at a rapidly preset temperature or to temperature
Rhodium 1800 C
jump so that a number of isothermal intervals for a pre-
Molybdenum 2200 D
determined period are imposed upon the system or the
Tungsten 2800 D
sample. In other units very rapid quench cooling from a
a A—an oxidizing atmosphere can be used (oxygen or preset temperature can be achieved. It should be noted that
air); B—a nonoxidizing atmosphere can be used (inert or the isothermal treatment of materials does not fall within
vacuum); C—in these cases oxygen or air can be used at lower the obvious definition of thermal analysis unless this is
temperatures, but at higher temperatures an inert atmosphere
taken to mean that the material is heated at a zero rate,
is recommended; D—hydrogen should be used.
but most operators find it important because they wish to
refer their heat treatment at a determined rate against more
which the furnace will cool down. Because of the danger
classical studies involving isothermal operation. In other
of a magnetic field interfering with the measurements of
applications of the work, for example, simulation of indus-
certain physical properties of the sample or system, the
trial processes, more complicated program temperatures
furnace should be noninductively wound: that is, it should
are required, and with the aid of computer programming,
have two similar windings carrying current in opposite di-
this is generally possible. Since most thermocouples need
rections so that their magnetic fields cancel, and also, the
a reference cold junction, this is now generally provided
spacing of the windings should be decreased toward the
in the form of an electronic ice point reference unit.
end of the furnace to compensate heat losses. These points
are made because, for special purposes, the operator might
3. Recording Systems
well require the construction of furnaces to a specific de-
sign. The resistance elements of the furnace control the In the current design of thermal analysis equipment the
temperature that can be obtained. Table VIII gives a range recording system is such that the signals are received via
of resistance elements which, in theory, allows a tempera- the computer system. These signals can be obtained in the
ture of 2800◦ C to be reached. Furnaces operating above a form a of a digital readout or in the form of an X –Y plot.
temperature of 1350◦ C should do so in an inert or reducing These data can be provided using software provided by
atmosphere. Graphite or silicon carbide resistance bars are the instrument manufacturer, by the dedicated computer
used instead of metal bars in furnaces in which tempera- manufacturer, or by the laboratory using the equipment.
tures can go over 2000◦ C, and these are simple to operate, This enables the material to be subjected on the spot to
provided the restrictions on atmosphere control are noted. various analytical processes which indicate at once the
The furnace should generally be mounted so that it can observations and the changes taking place.
be moved easily and also put back in exactly the same posi-
tion each time to minimize any errors due to the geometry G. Publications and Books
of the system being altered by the movement which is nec- on Thermal Analysis
essary to load the sample in and to generally inspect the
The literature on thermal analytical methods is widely
equipment. Where large furnaces are used, it is essential
scattered because it is generally abstracted under the
to have some kind of cooling system to allow the furnace
application heading rather than under the method heading.
to be programmed rapidly back to ambient temperature.
The American Chemical Society publishes its Chemical
Abstracts, which embrace most published research papers,
2. Temperature Programmers
and all the articles which contain thermal analysis data
Most thermal analysis results require a simple linear heat- then appear in C.A. Select Abstracts on Thermal Analysis.
ing program, so that a typical instrument will allow pro- There are two international journals that publish papers on
P1: GPJ/GJK Final Pages P2: GSS
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
3. The geometry and materials of thermocouples and the diluent reference material and the sample under investiga-
location of the various thermocouples should be tion. However, as noted above, in most modern units dilu-
given. tion is not necessary and the reference cell is used empty.
4. The ordinate scale should indicate the deflection per This is because the high sensitivity of the equipment re-
degree Celsius at a specified temperature. quires only a few milligrams of sample to be studied.
Most commercial firms issue concise operating instruc-
tions and many application briefs. These should be re-
C. Position of the “Temperature” ferred to as often as needed before a deeper search of the
Thermocouple literature is attempted. The possibility of using a single-
It is normal to program the temperature to produce a lin- cell model in the future is real, for the temperature of the
ear heating rate. This, however, is not a sufficiently precise “inert” could be calculated by some form of computer
statement. One can use either the thermocouple in the ref- program and subtracted from the single-cell temperature
erence material or the thermocouple in the sample as the reading to given the T signal.
indicator of temperature. In many systems the sample ther-
mocouple is advocated as the indicator of the temperature.
In the model set up by Cunningham and Wilburn there is D. Theory of DTA
a separate thermocouple to generate the T signal and two The main applications of DTA are (i) to describe the ther-
further thermocouples to give the T signal. Using this mal decomposition and transitions occurring on heating
model the same authors came to the conclusion that the a material through a programmed temperature range, (ii)
temperature-measuring the thermocouple junction is to be to measure the heat of reaction, and (iii) to determine the
placed in the sample. This point is also made by Mackenzie kinetic parameters. The first is reasonably clear, the sec-
an Mitchell (1970) with reference to the above theoretical ond requires an explanation, and the third, once the energy
work of Wilburn and that of Grimshaw et al. (1945). terms have been related to the weight changes, is similar
Two points can now be made. First, although it is gen- to the treatment of kinetic data on the thermobalance.
erally recommended that the sample cell is the proper lo- In dealing with the measurement of the heat of reaction,
cation for the temperature thermocouple, this means that the theories of DTA can be placed in two categories: (i)
a linear response is possible only while no reaction or those which deal with heat transfer alone and (ii) those
phase change takes place. This is demonstrated in Fig. 6. which deal with the reaction equation, that is, take into
Second, the peak height of the T signal varies with the account the chemical nature of the reaction.
temperature at which T is measured. A simplified heat transfer theory is based on the method
In practice it is often found that the factors relating to the developed by Vold (1949).
sample which are important are the particle size, the par- The equation of heat balance for the cell containing the
ticle size distribution, the packing of the sample, the dilu- reaction is
tion of the sample with reference material (although most
modern units dispense with this), and control of the atmo- CdT1 = dH − K (T3 − T1 ) dt, (1)
sphere. The dilution of the sample represents an attempt to
make the thermal characteristics of the two cells identical. and that for the cell containing the inert material is
There is, however, the possibility of reaction between the
CdT2 = −K (T3 − T2 ) dt, (2)
should be reported. The following additional details are FIGURE 9 Schematic diagram of a thermobalance and control
also necessary in the reporting of TG data. unit.
A statement must be made concerning the mass and
the mass scale for the ordinate. The actual weight of the
sample is reported, so a downward trend represents a mass Two weighing systems need to be noted, namely, de-
loss. Additional scales, for example, mass loss, fractional flection and the null-point balances. There is a variety of
decomposition, and molecular composition, can also be deflection balances that can be constructed—beam type,
used according to the calculations which are subsequently helical spring, cantilevered beam, torsion wire, etc.—but
to be made. they suffer in that the sample under observation will not
If derivative TG is employed, the method of obtaining remain in a fixed position in the furnace. For this reason
the derivative should be indicated and the units of the most units employ a null-point balance.
ordinate specified. In the use of null-point balances, a sensor must be used
The definitions used in TG can be illustrated by ref- to detect the deviation of the balance beam from its null
erence to the single-stage process illustrated in Fig. 8. position, and a variety of methods may be used to detect
Reference to multistage processes can be considered by deviations from the horizontal or vertical norm. One com-
thinking of the multistage process as being a series of mon system would is make use of the varying intensities
single-stage processes. of a light source impinging upon a photoelectric cell. The
A plateau (AB in Fig. 8) is that part of the TG curve usual arrangement incorporates a light source, a shutter or
where the weight is essentially constant. mirror, and either single or double phototubes. The dis-
The initial temperature, Ti (B in Fig. 8) is the tem- placement of the shutter attached to the balance beam (or
perature (on the Celsius or Kelvin scales) at which the spring) intercepts the light beam, thus either increasing or
cumulative weight change reaches a magnitude that the decreasing the light intensity acting on the phototube. The
thermobalance can detect. resulting change in current magnitude from the phototube
The final temperature, Tt (C in Fig. 8) is that temper- is then used to restore the balance to its null point.
ature at which the cumulative weight change reaches a Furnace design features and temperature measurement
maximum. The reaction interval is that temperature dif- have already been discussed.
ference between Tf and Ti as defined above.
IV. EVOLVED GAS ANALYSIS
B. Design Factors
A. Introduction
The basic instrumental requirements for TG are a precision
balance, a furnace capable of being programmed through In EGA the gases evolved from the decomposition of ma-
a required regime of temperature change, and a computer terials in a thermal analysis unit are analyzed. There are
workstation capable of programming the furnace, record- many methods of analyzing gases. In the past, specific
ing the weight change, and processing the data (see Fig. 9). chemical methods have been favored, but now instru-
The essential requirements of an automatic and contin- mental methods based largely on mass spectrometric
uously recording balance are similar to those of an analyti- methods, chromatography, or infrared spectroscopy are
cal balance and include accuracy, sensitivity, reproducibil- generally practiced. In certain applications, however, spe-
ity, and capacity. In addition, a recording balance should cific chemical analysis is still used. In thermal analysis
have an adequate range of automatic weight adjustment, there must be an interface between the heat-treated sample
a high degree of mechanical and electronic stability, and and the gas detection unit. Gas analysis is rarely used in
a rapid response to weight changes and be unaffected by such cases by itself but more commonly combined with
vibration. TG or DTA.
P1: GPJ/GJK Final Pages P2: GSS
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
analysis. It is not restricted to the above gases, of course, gram. The record is termed the thermodilatometric curve.
but as noted there is special merit in applying the method The dimension should be plotted on the ordinate, increas-
to them. ing upward, and T or t on the abscissa, increasing from
In an infrared radiometer, the output from an infrared left to right. Linear TDA is then the measurement of one
source is split to pass through the sample and reference dimension of a solid “form” against altering temperature.
effluent streams, and the radiation intensities are then com- Volume TDA measures the change in volume of the solid
pared. The differential output is a measure of the extra ra- or other phase versus the temperature program to which
diation absorbed by the evolved gas in the sample stream. it is subjected. It should be noted that this means that any
Such a system is not specific to one gas, and the response measure of change of density versus temperature of treat-
factor varies with the absorption coefficients of the com- ment falls within the definition of TDA.
ponent gases. Thermomechanical analysis (TMA) is a technique in
Infrared spectroscopy may be applied to EGA in two which the deformation of a substance under a nonoscilla-
ways. tory load is measured as a function of temperature while
the substance is subjected to a controlled temperature pro-
1. Continuous monitoring of specific products. This is a gram. The mode as determined by the type of stress applied
development of the infrared radiometer, in which the (compression, tension, flexure, or torsion), which should
detector is sensitized to only one component, either always be stated.
by including filters (a nondispersion analyzer) or by As in other cases, when reporting TMA the temperature
operating at a single wavelength (a dispersion environment of the substance should be clearly stated. The
analyzer). A series of such analyzers can be used, type of deformation (tensile, torsional, bending, etc.) and
each detecting one gas. Commercial units are the dimensions, geometry, and materials of the loading el-
available to detect CO2 , CO, and H2 O. Such methods ements should be noted. The ordinate scale should also be
can be used to establish kinetic parameters associated identified in specific terms where possible. For static pro-
with the evolution of these gases. cedures, increasing expansion, elongation, or extension
2. Special analysis of evolved gases. When the and torsional displacement should be plotted upward. In-
composition or identity of the components in a gas creased penetration or deformation in flexure should be
stream needs to be established, then a spectral plotted downward. For dynamic mechanical procedures,
analysis of evolved gases is required. Obtaining such the relative modulus and/or mechanical loss should be
data is relatively easy but the method is plotted upward.
noncontinuous with respect to time. In dealing with change in the dimensions of a sample
with temperature the measurements relate to the deforma-
E. Gas Chromatographic Methods tion and strength of the sample. Expansion in a solid or a
liquid is indicative of a decrease in surface energy. Thus,
There are numerous descriptions of the coupling of gas the expansion of a solid in an adsorption process may be
chromatography with thermal analysis units. Chiu (1968, interpreted as a decrease in surface energy.
1970) describes a TG unit coupled with a gas chromato- A practical distinction between TMA and TDA is sim-
graph. In the use of gas chromatography the gas detector ply that in TMA some kind of stress or load is applied to
device used has to be suitable to match the gases evolved; the test material, while in TDA no load or stress is required.
otherwise they will escape detection. The three most com- The same basic equipment may be used. In commercial
monly employed are thermal conductivity detectors, gas equipment the temperature may go as low as −170◦ C or
density detectors, and ionization detectors. The restriction as high as 1000◦ C.
on the employment of gas chromatography coupled with
thermal analysis units is that the analysis is intermittent
and not continuous. B. Apparatus
A wide variety of equipment is available, which includes
the measurement of various properties. The list given here,
V. THERMOMECHANICAL METHODS
with brief descriptions, is illustrative rather than compre-
hensive. Logically volume changes and measurement of
A. Introduction
density should be discussed under this heading. However,
Thermodilatometry (TDA) has been defined as a technique most commercial TMA units note only a single dimen-
in which a dimension of a substance under a negligible sional change on fabricated units (it is difficult to make
load is measured as a function of temperature while the TMA measurements on powders). The measurement of
substance is subjected to a controlled temperature pro- density involves techniques which would be difficult to
P1: GPJ/GJK Final Pages P2: GSS
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
material with a high Q (quality factor), the energy incor- where AT (±K) denotes the amplitude of the temperature
porated into oscillation will be equal to that introduced by modulation, w (s−1 ) is the modulation frequency, and
deformation, with the frequency of the resultant oscillation
2π
being a function of the modulus (stiffness) of the material. w= ,
Most real materials, however, do not exhibit ideal elastic p
behavior but, rather, exhibit viscoelastic behavior in which where p (s) is the modulation period. By using those points
a portion of the deformation energy is dissipated in other which lie on the linear temperature profile, a “normal”
forms such as heat. The greater this tendency for energy signal can be obtained together with the modulated signal.
dissipation, the larger the damping of this deformation- In both sets of experimental data the total heat flow at any
induced oscillation. On the other hand, if this dissi- point is given by
pated energy is continually made up (by an in-phase
dQ
drive signal applied to the system), the sample will stay = C pβ + f (T1 t),
in continuous natural frequency (compound resonance) dt
oscillation. where Q (J) denotes the heat, t (s) the time, C p (J/K) the
The two properties measured by DMA are the resonant sample heat capacity, and f (T1 t) the heat flow from kinetic
frequency and energy dissipation, and these can be mea- processes which are dependent on both temperature and
sured over a wide range of temperatures and moduli. time.
The resonant frequency obtained is related to the The linear temperature programmed DSC measures
Young’s modulus of the sample via the following equation: only the total heat flow. However, the sinusoidal heating
3 profile gives the heat capacity data corresponding to the
4π 2 f 2 (J − K ) L
E= , rate of temperature change. The heat capacity component
2w((L/2) + D)2 T of the total heat flow, C pβ, is called the reversing heat flow
where and the kinetic component, f (T1 t), is called the nonrevers-
ing heat flow. In one experiment the heat capacity can then
E = Young’s modulus, be calculated using a discrete Fourier transformation by
f = DMA frequency, the relationship
J = moment of inertia of the arm,
(Q amp ) (P)
K = spring constant of the pivot, Cp = K ,
(Tamp ) (2π )
D = clamping distance,
w = sample width, where k denotes the heat capacity constant, Q amp the heat
T = sample thickness, flow amplitude, and Tamp the temperature amplitude.
The reversing component of the total heat flow signal
L = sample length.
(C pβ) allows the nonreversing component to be calculated
The energy dissipation obtained is related to prop- using the relationship
erties such as impact resistance, brittleness, and noise
nonreversing heat flow = total heat flow
abatement.
− reversing heat flow.
B. Temperature Modulated Differential The technique finds advantages over normal DSC in
Scanning Calorimetry (MDSC) calculation of the heat capacity and in determination of
the glass transition point.
In this technique the linear temperature regime impassed
in DSC is replaced by a sinusordal temperature modulation
superimposed on a linear (constant) heating profile. VII. SIMULTANEOUS TECHNIQUES
The program sample temperature (T (t)) in normal DSC
is given by A. Introduction
T (t) = To + βt, It is becoming a common practice to apply two (or more)
techniques of thermal analysis to the same sample at the
where To (K), β (K/min), and t (min) denote the starting
same time. The most common application is that of TG
temperature, linear constant heating or cooling rate, and
and DTA. However, it should be noted that EGA is rarely
time, respectively.
used as a separate technique and is most often used in
A sinusoidal modulation would then be represented by
a simultaneous combination. In writing, the names of si-
T (t) = To + βt + AT sin wt, multaneous techniques should be separated by the use of
P1: GPJ/GJK Final Pages P2: GSS
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
the word and when used in full and by a hyphen when ab- be to inhibit or prevent mass loss. Crucibles ideal for TG
breviated acceptably, e.g., simultaneous TG-DTA. Unless may not be ideal for DTA. Finally, the heating rate may
contrary to established practice, all abbreviations should be cited: to obtain good DTA experiments the heating rate
be written in capital letters. should be reasonably fast; otherwise the peaks are broad
The other practice of thermal analysis in which more and shallow. A slow heating rate, however, often gives the
than one technique is used is coupled simultaneous tech- best results in TG experiments.
niques. This term covers the application of two or more
techniques to the same sample when the two instruments
C. New Studies Made Possible
involved are connected by an interface, for example, si-
by Simultaneous Techniques
multaneous DTA and mass spectrometry. The term inter-
face refers to a specific piece of equipment that enables There are cases where simultaneous techniques can lead
the two instruments to be joined together. In coupled si- to interpretations not possible using convention tech-
multaneous techniques as in discountinuous simultaneous niques separately. One such study is evaporation. In con-
techniques, the first technique mentioned refers to the first ventional DTA, with the sample in a closed crucible,
in time measurement. Thus, when a DTA instrument and such a study is impracticable. However, with the open
a mass spectrometer are used together, then DTA-MS is crucible, evaporation in a TG-DTA unit shows a dis-
correct. tinctively shaped endothermic peak; the TG for one-
The name discountinuous simultaneous techniques is component systems demonstrates a zero-order process;
also found, and this term covers the application of cou- and, using the Langmuir equation for evaporation, Price
pled techniques to the same sample when sampling for and Hawkins show that a vapor-pressure curve (par-
the second technique is discontinuous, for example, dis- tial pressure plotted against temperature) can be ob-
continuous simultaneous DTA and gas chromatography tained. Calibration for temperature also becomes easy and
when selected portions of evolved volatiles are “sampled” more reliable. Combination of techniques in simultane-
on the gas chromatography unit. Another case of a simul- ous units also allows interpretations of complex degra-
taneous technique arises when two samples are subjected dations to be studied. Thus it can be established that
in the same furnace to two different thermal analysis tech- Mg(NO3 )2 . 6H2 O melts at 90◦ C and then water is lost
niques, for example, DTA and TG. from the system, leaving an anhydrous solid salt at 350◦ C.
This melts at 390◦ C and decomposes to the oxide at
600◦ C. DTA shows that all the processes are endother-
B. Advantages and Disadvantages
mic; gas analysis shows the loss of water and identifies the
of Simultaneous Techniques
gases evolved in the final dissociation to the oxide. Hot-
The advantage of using simultaneous techniques lies in stage microscopy has been used to establish the melting
the nature of the samples being investigated. They are a processes.
reflection of the fact that in certain materials there is a
distinct probability of different samples showing different
signals. One can cite impure samples where the signal VIII. THE ROLE OF THE COMPUTER
may vary with the nature and amount of the impurity. In
other samples studied by simultaneous techniques such as In some of the early designs for dedicated computers
pharmaceutical stearic acid or magnesium starate (known the computer was completely dedicated to the equipment
to contain other carboxylic acids), coal samples, and other and operated only with programs provided by the instru-
natural products, the sample may vary from one region mentation company. To appreciate the role of the com-
to another. Simultaneous tests on one sample may then puter, however, it must be noted that it should play three
represent a real advantage. roles.
It must not be imagined that the combination of tech-
niques should always produce the same results shown by 1. To control the programs that alter the temperature
the separate techniques, and this is particularly true for regime.
DTA and TG. Consider first the mass of material stud- 2. To record the property being measured and
ied. Obviously the mass of the substance used in a TG demonstrate its variation with the temperature regime.
run should be sufficient to note the change in mass accu- 3. To interface this dedicated computer with computer
rately. However, the ideal mass of a sample used in DTA programs or transfer the data to other computers and
to achieve the best results may be much smaller than in allow the operator in charge of the study to
TG. The sample holders in many DTA experiments are manipulate the data and derive appropriate
sealed. The use of such a technique in DTA-TG would parameters.
P1: GPJ/GJK Final Pages P2: GSS
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
In the case of determination of kinetic data from raw holds, where R is the gas constant, T is the temperature of
experimental data, this usually means feeding such data the phase change (K), and H is the enthalpy of fusion.
(e.g., mass and temperature) into a spreadsheet and calcu- The above phase transitions can all be investigated by DTA
lating the preexponential term and the activation energy or DSC.
in the Arrhenius equation. If an amorphous (energy-rich) solid phase undergoes
Most modern dedicated computers fulfill these three transition, however, then a stable crystal may result. The
roles and allow the operator to manipulate the basic ex- amorphorus or glassy phase is metastable, and the process
perimental data by using his own spreadsheets or computer is exothermic and irreversible. This can be represented as
programs. It must be noted that instrument manufacturers
Metastable → Stable form
supply their own programs to make sure that the use of
the equipment in industry is made as easy as possible. (energy rich) (crystalline)
Exothermic and irreversible
IX. APPLICATION OF THERMAL Such changes are shown in both inorganic and organic
ANALYSIS TECHNIQUES systems.
The above are all examples of transitions which are
A. Introduction of first order. However, second-order transitions can also
be followed and determine the position of the baseline in
The application of thermal analysis techniques is based DSC. This leads to the calculation of heat capacity from
on either thermodynamic considerations or the kinetics of DSC measurements and is the basis for the establishment
change. It is not just the fact that the temperature is being of the glass transition temperature (Tg ) in polymers. A fur-
changed that makes a choice of this kind necessary, but the ther extension based on thermodynamic factors is purity
kind of systems investigated. Some macromolecules are determination based on the lowering of the freezing point
so big that phase changes which, in simple systems, would caused by the presence of impurities.
show only thermodynamic (equilibrium) features exhibit In chemical reactions certain features may be noted.
a kinetic factor. In the preceding survey certain techniques Thus, carbonate decompositions are endothermic, and loss
have not been reported or have not been described in de- of water is also endothermic. Obviously, such processes
tail. Only the main techniques available from more than involving weight loss can be studied using both TG and
one commercial manufacturer have been reported in some DTA/DSC. It may also be advantageous to study such pro-
detail. cesses under partial pressure (P) of the gaseous products
when one may expect the relationship
B. Thermodynamic Considerations −H
ln p = + constant, (7)
RT
In a condensed system, if a phase change occurs in the
similar to the liquid–gas phase change, but here H is the
heating mode where one phase is stable over a definite
enthalpy of the reaction.
temperature range and the other phase is stable over, say, a
higher temperature range, then the process of change at the
transition temperature will be endothermic and reversible C. Kinetic Considerations
on the cooling mode with an exothermic character. This
Where kinetic factors are important in studying systems
can be represented as
using thermal analysis techniques, then the Arrhenius
A(1) −−−→ A(2) endothermic parameters in theory should be able to be estimated. The
heating
Arrhenius equation can be written
A(2) −−−→ A(1) exothermic k(T ) = Ae−E/RT , (8)
cooling
This is true where both phase (1) and phase (2) are solids where k(T ) is the specific reaction rate constant and is a
or phase (1) is solid and phase (2) is liquid. In the phase function of temperature, A is the preexponential term, E
change from a liquid to a gas, the process is reversible, is the energy of activation, R is the gas constant, and T
but because the gas is usually lost from the system, it is the temperature (K). In the rising temperature mode we
is probably best to follow the process at different partial have
pressures of the vapor ( p) when the relationship dα/dt = k(T ) f (α), (9)
−H where α is the fraction decomposed at time t, dα/dt is the
ln p = − constant (6)
RT rate of the reaction, k(T ) is the specific rate constant at
P1: GPJ/GJK Final Pages P2: GSS
Encyclopedia of Physical Science and Technology En0016b-770 July 31, 2001 17:18
temperature T , and f (α) is a score function of α describ- Blaine, R. L., and Fair, P. G. (1983). Thermochim. Acta 67, 233.
ing the progress of the reaction. The temperature regime Boersma, L. (1955). J. Am. Ceram. Soc. 38, 281.
imposed on the system can be represented by Brown, M. E., Dollimore, D., and Galwey, A. K. (1980). Reactions in
solid state. In “Comprehensive Chemical Kinetics” (C. H. Bamford
T = T0 + bt, (10) and C. F. H. Tipper, eds.), Vol. 22, p. 99, Elsevier Science.
Charsley, E. L., Rumsey, J. A., and Warrington, S. B. (1984). Anal. Proc.
where T is the temperature at time t, T0 is the starting tem- XX, 5.
perature, and b is the heating rate. Combination of all these Chiu, J. (1968). Anal. Chem. 40, 1516.
equations, noting that in linear temperature programming Chiu, J. (1970). Thermochim. Acta 1, 231.
Cunningham, A. D., and Wilburn, F. W. (1970). “Differential Thermal
b = dT /dt, gives Analysis” (R. C. Mackenzie, ed.), Vol. 1, p. 31, Academic Press,
(dα/dT )b London.
k(T ) = , (11) David, D. J. (1964). Anal. Chem. 36, 2162.
f (α) Dollimore, D., and Reading, M. (1993). Application of thermal anal-
thus allowing k to be plotted against 1/T if f (α) can be ysis to kinetic evaluation of thermal decomposition. In “Treatise on
identified, whence from the relationship Analytical Chemistry, Part I. Thermal Methods,” 2nd ed., (J. D. Wine-
fordner, D. Dollimore, and J. Dunn, eds.), Vol. 13, pp. 1–61, John
ln k = ln A − (E/RT ), Wiley & Sons, New York.
Dollimore, J., Freedman, B. H., and Quinn, D. F. (1970). Carbon 8, 587.
both A and E can be calculated. The difficulty in obtain- Dunn, J. G., and Sharp, J. H. (1993). Thermogravimetry. In “Treatise
ing dα/dT with sufficient precision has led to extensive on Analytical Chemistry, Part I. Thermal Methods,” 2nd ed. (J. D.
studies into the use of an integral method. This entails the Winefordner, D. Dollimore, and J. Dunn, eds.), Vol. 13, pp. 127–
266.
evaluation of
Flynn, J. H. (1981). In “Thermal Analysis in Polymer Characterization”
(E. A. Turi, ed.), p. 43, Heyden, Philadelphia.
e−E/RT dT, Gallagher, P. K. (1978). Thermochim. Acta 26, 175.
Galwey, A. K., and Brown, M. E. (1999). “Thermal Decomposition of
and this is impossible to do analytically. Numerous meth- Ionic Solids,” p. 597, Elsevier, Amsterdam.
ods of overcoming this difficulty have been suggested, Grimshaw, R. W., Heaton, E., and Roberts, A. L. (1945). Trans. Br.
usually based on the numerical methods of integrations Ceram. Soc. 44, 76.
Hatakeyama, T., and Quinn, F. X. (1999). “Thermal Analysis,” 2nd ed.,
or alternative similar expressions that can be analytically
p. 180, Wiley, Chichester.
evaluated. The solid state is the only phase for which one Hill, J. O. (1991). “For Better Thermal Analysis,” 3rd ed., ICTA.
has to state a prehistory, and an article by Flynn (1981) Mackenzie, R. C., and Mitchell, B. D. (1970). “Differential Thermal
deals with this aspect. Related to these studies are the pre- Analysis” (R. C. Mackenzie, ed.), Vol. 1, p. 63, Academic Press,
diction of shelf life, especially of pharmaceuticals, and the London.
Marti, E. (1972). Thermochim. Acta 5, 173.
proximate analysis of coal.
McGhie, A. R. (1983). Anal. Chem. 55, 987.
McGhie, A. R., Chiv, J., Fair, P. G., and Blaine, R. L. (1983). Ther-
SEE ALSO THE FOLLOWING ARTICLES mochim. Acta 67, 241.
Norem, S. D., O’Neill, M. J., and Gray, A. A. (1970). Thermochim. Acta
1, 29.
CRITICAL DATA IN PHYSICS AND CHEMISTRY • HEAT O’Neill, M. J. (1964). Anal. Chem. 36, 1233.
TRANSFER • KINETICS (CHEMISTRY) • PHYSICAL CHEM- Price, D., Dollimore, D., Fatemi, N. J., and Whitehead, R. (1980). Ther-
ISTRY • THERMODYNAMICS • THERMOMETRY mochim. Acta 42, 517.
Prout, E. G., and Tompkins, F. C. (1946). Trans. Faraday Soc. 43, 482.
Radecki, A., and Wesolowski, M. (1979). J. Therm. Anal. 17, 73.
Rosenvold, R. J., Dubow, J. B., and Rajeshwar, K. (1982). Thermochim.
BIBLIOGRAPHY Acta 53, 321.
Shoemaker, D. P., and Garland, C. W. (1967). “Experiments in Physical
Adamson, A. A. (1986). “A Textbook of Physical Chemistry,” 3rd. ed., Chemistry,” 2nd ed., p. 177, McGraw–Hill, New York.
p. 401, Academic Press, Orlando, FL. Vold, M. J. (1949). Anal. Chem. 21, 683.
Atkins, P. W. (1982). “Physical Chemistry,” 2nd ed., p. 298, Freeman, Watson, E. S. M., O’Neill, M. J., and Brenner, N. (1964). Anal. Chem.
San Francisco. 36, 1233.
P1: GSS Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography
Z. H. Cho
Korea Advanced Institute of Science
I. Introduction
II. Principles of Computerized Tomography
III. Areas of Application
IV. Recent Developments
843
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
844 Tomography
Spin echo Signal produced by the 90◦ –180◦ RF pulse annihilation photons from decaying radionuclides, trans-
sequence. The spin-echo signal is actually conjugate mitted and reflected ultrasound beams, appropriately ex-
symmetric to its center if its duration is short compared cited nuclear spins under a strong magnetic field, and
with T2 . Its amplitude is determined by the T2 of the so on. The data obtained, usually referred to as projec-
substance, excluding the effects of field inhomogeneity. tion data, are processed by a mathematical image recon-
Spin–lattice relaxation (thermal or longitudinal relax- struction algorithm using a digital computer to form an
ation) Phenomenon of spins going to the thermal equi- image or a set of images, each representing a slice or
librium state with other molecules in lattice. It is char- several slices of an object. Although direct Fourier trans-
acterized by the exponential time constant T1 . Also, it form image reconstruction is employed in some cases
determines the recovery time of the longitudinal mag- to obtain an image or a set of images, as in NMR CT,
netization Mz . most types of image reconstruction employ some forms of
Spin–spin relaxation (transverse relaxation) Exchange projection reconstruction. Those mathematical techniques
of energy of excited nuclei with other precessing nuclei. known as 3-D image reconstruction form the basis of all
It is characterized by the exponential time constant T2 . the CTs discussed in this article. Finally, mathematically
Also, it determines the envelope of the free induction formed or reconstructed images are displayed on a televi-
decay in a perfectly uniform magnetic field and the sion screen or photographed by a camera attached to the
amplitude of the spin echo. system.
Superconductive magnet Magnet that requires no elec- From the physics and engineering points of view, both
trical power once the field has been established. Su- 3-D image reconstruction and CT are new concepts and
perconductivity is a property of some materials that tools. For the first time in history, human beings are ca-
have no electrical resistance when the temperature is pable of visualizing the inner structures of an object non-
near absolute zero. Liquid helium is generally used to invasively. At present, the most widely used CT is X-
maintain a low temperature. ray CT; it is estimated that more than 5000 X-ray CT
True coincidence Event detected in coincidence without units, each costing as much as $1 million, are in opera-
scatter of either photon, generated from an annihilated tion throughout the world. In the area of ECT, two types
positron, in the object. are under active development: single photon emission CT
(SPECT) and positron emission tomography (PET). Al-
though ECT is similar to X-ray CT, it differs in functional
COMPUTERIZED or computed tomography (CT) is a form; that is, X-ray CT is usually capable of visualizing
technique of producing cross-sectional three-dimensional anatomical details, while ECT is capable of visualizing
images from multiple views or projection data obtained the functional or metabolic behaviors of an object. Ul-
with penetrating probe radiations or by other means such trasound CT is similar to X-ray CT and is capable of
as magnetic field gradients, by processing those data using visualizing cross-sectional images. The development of
a computer and mathematical image reconstruction algo- ultrasound CT has been relatively slow, due mainly to the
rithms. Major applications of the basic CT concepts are difficulties inherent in the basic properties of ultrasound,
medical diagnosis, industrial nondestructive testing, and such as scattering and diffraction. The most recent and
other areas of the physical sciences, such as geophysical probably most exciting development in the field of CT is
exploration. NMR CT. It is now capable of visualizing in three dimen-
sions the distribution of several nuclei, such as the pro-
ton, the sodium nucleus, and the phosphorus nucleus, and
I. INTRODUCTION thereby performing chemical imaging. As the resolution
and sensitivity of NMR CT improve rapidly, it is becom-
Computerized tomography is a technique by which three- ing one of the most promising medical diagnostic imaging
dimensional (3-D) imaging of an object is made possi- modalities.
ble. The basic data are most often obtained in the form Computerized tomography is believed to be the most
of projection data. Since the first tomographic system, important development in diagnostic imaging history
known as X-ray CT, was designed by G. N. Hounsfield since the discovery of X rays in 1895 by W. K. Roentgen.
in 1972, many applications of the technology, based on Applications of CT concepts are rapidly expanding from
3-D image reconstruction from projection, have been de- medical imaging to many branches of the physical sci-
veloped. These include X-ray CT, radionuclide (isotope) ences. The principles of X-ray CT are now being applied
emission CT (ECT), and nuclear magnetic resonance to many diverse areas, ranging from the examination of
(NMR) CT. The basic forms of projection data are ob- defects in nuclear reactor cores to the rapid inspection of
tained from focused and collimated X rays, γ rays, and automobile tires on the production line. The principles
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 845
A. Mathematics and Algorithms Since 1970, active research on image reconstruction from
for Image Reconstruction projection has been initiated mainly through the develop-
ment of CT. This projection reconstruction is now applied
Several mathematical image reconstruction algorithms to many areas of science. The most important area of its
used for CT have been developed since the basic con- application has probably been CT; projection reconstruc-
cept of CT was established. These algorithms are based tion has remained the basic algorithm for CT since the
primarily on mathematical theory as described in detail inception of X-ray CT in 1972.
in this section, which presents a mathematical basis for The simplest form of projection data is illustrated in
the reconstruction algorithms currently used in many CT Fig. 1. The line integrals of a physical object are estimated
modalities, including X-ray CT, ECT, and NMR CT. Ac- along straight lines. Each line integral, in practice, repre-
cording to data processing methods, algorithms can be sents a physical property in a strip with a finite width.
classified into projection reconstruction, iterative method, Therefore, a set of line-integral data is obtained at each
and direct Fourier imaging. A classification of these three view. By repeatedly assessing the sets of data at differ-
general methods is given in Table I. ent views, that is, around 180◦ or 360◦ with a specified
angular interval θ , a complete projection data set can
be obtained. Each set of estimated line integrals is often
TABLE I Image Reconstruction Algorithms called a projection or a line-integral projection. The collec-
Class Specific method tion of all these estimated line-integral sets around 180◦ or
360◦ is referred to as line-integral projection data or sim-
Projection reconstruction 2-D Projection reconstruction ply projections. Similarly, planar-integral projection data
Filtered backprojection (FB) can also be obtained; they represent the collection of all
Parallel beam mode integral values of a physical object along planes.
Fan beam mode From a set of measured projection data or projections,
Backprojection filtering (BF) an image can be formed through the use of appropriate
3-D Projection reconstruction processing algorithms. Image reconstruction from projec-
True 3-D reconstruction (TTR) tions is the process of producing an image of a 2-D or a
Generalized TTR 3-D distribution of some physical property from the esti-
Planar-integral projection reconstruction mates of its line (or plane) integrals along a finite number
(PPR)
of lines (or planes) of known locations. For the reconstruc-
Iterative method Algebraic reconstruction technique (ART)
tion of the images from projections, several algorithms can
Maximum likelihood reconstruction (MLR)
be used depending on the imaging modalities. Theoretical
Fourier reconstruction Direct Fourier reconstruction (DFR)
aspects and characteristics of these algorithms are dis-
Direct Fourier imaging in NMR (DFI)
cussed in the following subsections. Fourier transforms,
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
846 Tomography
which constitute the main body of image reconstruction in represent the Cartesian and polar coordinates of (x, y) in
general, are also briefly explained in the following. Fourier the spatial frequency domain, respectively.
transforms are basically part of a conversion process that Equation (5) states that a 1-D Fourier transform of pro-
converts spatial domain data to spatial frequency domain jection data at a given angle φ represents the 2-D Fourier
data. They are defined as transform values of the object function f (x, y) in the spa-
∞ tial frequency domain along the radial frequency with a
F(ω) = 1 [ f (x)] = f (x) exp(−iωx) d x (1) given angle φ. This is the projection theorem that plays a
−∞ central role in 2-D image reconstruction. From this theo-
−1 rem, it can easily be shown that the object function f (x, y)
f (x) = 1 [F(ω)]
∞ can be recovered as
=
1
F(ω) exp(iωx) dω, (2) π ∞
1
2π −∞ f (x, y) = dφ d x pφ (x )h(x cos φ
√ 2π 0 −∞
where i = −1 and 1 [·] and −1 1 [·] are the 1-D forward
and inverse Fourier transform operators, respectively.
+ y sin φ − x ) , (6)
a. Two-dimensional projection reconstruction.
where −1
i. Filtered backprojection algorithm. The filtered h(x ) = 1 [|ω|].
backprojection (FB) or convolution backprojection
algorithm is the most popular and most frequently used The convolution kernel h(x ) in Eq. (6) is an inverse
reconstruction method so far employed in CT, with the Fourier transform of |ω|, but its exact form is not real-
exception of NMR CT. For the mathematical formulation izable in practice. Therefore, several modified filter func-
of this FB algorithm, there are two basic forms in tions have been suggested. The selection of a particular
existence: the parallel beam and fan beam modes. filter function will affect the characteristics of the recon-
Parallel Beam Mode. Let us assume an object dis- structed image, that is, the desired image resolution and
tribution function f (x, y) represented by the Cartesian contrast.
coordinates (x, y), with rotated coordinates expressed as Fan Beam Mode. Although the parallel beam recon-
(x , y ). The basic data to be used in the reconstruction are struction algorithm has been a basic tool for image recon-
projections that represent sets of line integrals of an object struction, the fan beam reconstruction algorithm is nev-
in various directions. The projection data pφ (x ) shown in ertheless widely used. For example, it is used in X-ray
Fig. 1 is a set of line integrals taken along the y direction, CT, due mainly to the fact that the basic data collection
that is, procedure involved is the fan beam mode. In addition,
∞ image reconstruction that utilizes the fan beam algorithm
pφ (x ) = f (x , y ) dy often provides better resolution with the same amount of
−∞ sampled data than the parallel beam algorithm because of
∞
improved sampling at the central region.
= f (x, y)δ(x cos φ + y sin φ − x ) d x d y, If the fan beam projection data set is represented as
pα1 (β1 ), where β1 and α1 represent the detector position
−∞
and rotation angle of the beam, respectively (Fig. 2), then
(3) the relation between parallel and fan beam projection data
with coordinates (x , φ) and (α1 , β1 ) can be represented as
where
x = Rd sin β1
x cos φ sin φ x (7)
= . (4)
φ = α 1 + β1 ,
y −sin φ cos φ y
The Fourier transform of projection data Pφ (ω) can be where Rd is the distance between the center point and the
related to the projection data pφ (x ) as apex of the fan. Through the use of Eqs. (6) and (7) and
∞ the rotation angle φ = 0 ∼ 2π , the fan beam analogy can
be derived as
Pφ (ω) = pφ (x ) exp(−iωx ) d x 2π β1m
−∞ 1
f (x, y) = dα1 dβ1 pα1 (β1 )J (β1 )g
= F(ωx , ω y )|φ = F(ω, φ), (5) 4π 0 −β1m
Tomography 847
848 Tomography
1. Backprojection to get 2N × 2N blurred image where (ωx , ωz ) are the spatial frequency domain coordi-
2. Two-dimensional fast Fourier transform of the full nates of (x , z ) in the direction of (θ, φ). If the uniform
2N × 2N data and filtering planes that pass through the center are superimposed from
3. Two-dimensional inverse fast Fourier transform and all possible directions, in 3-D space, the overlapped den-
adoption of N × N image in the central region sity function becomes 1/ρ. Therefore, to obtain the 3-D
4. Normalization of the image with precalculated object function in the Fourier domain, the 2-D Fourier
reference data, such as the uniform disk image transform of each projection data set should be compen-
sated with the factor ω before the superposition.
b. Three-dimensional projection reconstruction. Image function can now be obtained by
i. True three-dimensional reconstruction with line-
integral data. In the conventional reconstruction algo- π
1
rithm, projections of an object are taken transaxially and f (x, y, z) = dθ sin θ
4π 2 0
are used to reconstruct the object slice by slice. The stacked 2π
slices constitute the 3-D volume image of the object.
× dφ −1
2 [ω] ∗∗ pθ,φ (x , z ) . (15)
For example, in ECT, in which radionuclides are in- 0
jected into the patient, the photons are emitted isotropi-
cally in 4π directions. This slice-stacking method for a
volume object does not make full use of the photons be- Equation (15) shows a 3-D FB algorithm, which is in fact
cause it captures photons emitted only in the direction the backprojection of the filtered 2-D projection data in
perpendicular to the long axis of the body, constituting 3-D space. In this case, the 2-D filter function or kernel is
only a small fraction of the total emitted photons. All the simply −1 2 [ω].
emitted photons, especially in ECT, should be collected ii. True three-dimensional reconstruction algorithm for
through the use of complete spherical geometry. For im- generalized geometry. The ideal detector configuration
age reconstruction with the data collected in 4π geometry, for the ECT is a complete sphere. The reconstruction of
the direct true three-dimensional reconstruction (TTR) al- such a spherically configured emission or transmission
gorithm will be required for the maximum utilization of data consisting of sets of line-integral data was discussed
all the available photons. in the preceding section. Three-dimensional images can
Parallel Beam Mode. Let us consider a spherical ge- be reconstructed by FB of the 2-D line-integral data sets in
ometry that completely surrounds the object, and let us 3-D space. Practical system design, however, prohibits the
assume that all the emitted photons are captured and re- construction of such a configuration when one considers
arranged into 2-D parallel data sets in 4π directions. The the elongated shape of the human body as an example.
PSF of this system is 1/r 2 , and the backprojected blurred As a practical alternative, a truncated spherical con-
function is given by figuration (Fig. 3) can be considered, and an algorithm
suitable for such geometry has been developed. This
b(x, y, z) = f (x, y, z) ∗∗∗ (1/r 2 ), (12) 3-D image reconstruction algorithm is known as the TTR
where ∗∗∗ represents a 3-D convolution operator. The ob- algorithm. The algorithm eventually will lead to a gen-
ject function can be obtained as eralized algorithm for both 2-D slice reconstruction and
3-D volume image reconstruction of a complete spherical
−1
f (x, y, z) = 3 [ρ] ∗∗∗ b(x, y, z), (13) volume. Consider a sphere in which parts of the surface
−1 have been removed along the body axis to adapt to the
where ρ = (ωx2 + ω2y + ωz2 )1/2 and 3 [·]
is a 3-D inverse
shape of the human body. Although the sphere is trun-
Fourier transform operator. Equation (13) is essentially
cated, this geometry retains spherical symmetry in the re-
a 3-D convolution (deconvolution) of a simple backpro-
maining spherical surface. Therefore, it also retains the
jected (blurred) image. This 3-D convolution process, as
possibility of reconstructing a true 3-D volume image, as
well as the backprojection operation, however, would re-
explained in the following paragraphs.
quire an unusually long computation time.
The 3-D image within the reconstruction sphere, which
As an alternative approach, the FB method has been
has a radius of R0 , is obtained by summing all the im-
proposed. The 3-D version of the projection theorem states
ages reconstructed at each slice orientation in which a
that the Fourier transform of the 2-D projection represents
complete set of 2-D projection data is provided. This sum
the plane data passing through the center in the frequency
image should be divided by the number of slice orienta-
domain in the same direction as the projected plane of the
tions reconstructed for normalization. Essentially, this is
object. This statement can be written
the basis of the reconstruction algorithm developed for
2 [ pθ,φ (x , z )] = F(ωx , ωz ; θ, φ), (14) truncated spherical geometry.
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 849
FIGURE 3 Basic geometry of the truncated spherical configuration for the development of the TTR algorithm.
−→ −→
For the implementation of the algorithm, all slice ori- the x1 axis is normal to both OO p and OO q , as shown in
entations involved in each set of projection data must be Fig. 3.
identified and the corresponding composite filter function The filter kernel h(x1 , z 1 ; α) in Eq. (17) is then given as
generated. In fact, each 2-D parallel projection data set
corresponding to a certain object size has a fixed number h(x1 , z 1 ; α) = −12 H ω x 1 , ωz 1 ; α
of slice orientations. This allows us to treat a certain pro-
and
jection data set in a unified way; that is, a 2-D projection
data set can be processed with one filter function. Let us H ωx1 , ωz1 ; α = |ωx2 | = ω| cos(ξ − α)|, (18)
consider a 2-D projection data set projected on a direction
where ωx2 is the rotated axis from ωx1 with angle α and
parallel to the line OOp in Fig. 3. Here, Op is the center of
(ω, ξ ) represents the polar coordinates of (ωx1 , ωz1 ).
the projection plane. As shown in the figure, these projec-
The denominator of Eq. (17) is the normalizing factor
tion data are shared by slice orientations rotated around
that represents the sum of all the weighting coefficients of
the line OOp from −β to β, where
projection data sets. The projection data set pθp ,φ (x1 , z 1 ) is
β = cos−1 (cos θu / cos θp ). (16) indepenent of α, and the convolution is a linear operation.
By use of the composite filter concept, object function Therefore, Eq. (17) can be further simplified as follows:
f (r ) can be obtained with the following equation: f (r) =
f (r) = π θu −1
π θu β 0 dφ −θu dθp cos θp 2 Pθp ,φ ωx1 , ωz1 Hθp ωx1 , ωz1
,
0 dφ −θu dθp cos θp −β dα[ pθp ,φ (x 1 , z 1 ) ∗∗ h(x 1 , z 1 ; α)] 4π 2 (1 − cos θu )
2π θu β ,
0 dφ −θu dθ p cos θ p −β dα (19)
(17) where
where pθp ,φ (x1 , z 1 ) is the parallel projection data set at po- 2ω cos ξ sin β,
lar angle θp and the azimuthal angle of φ, and h(x1 , z 1 ; α) 0≤ξ ≤ π −β
is the rotated form of the conventional filter function with Hθ (ω, ξ ) = 2
2ω(1 − sin ξ cos β), (20)
angle α, to be applied to projection data pθp ,φ at a slice
π π
orientation corresponding to the rotation angle variable α. −β <ξ ≤
−→
Here the z axis lies in the direction of the line OO q , and 2 2
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
850 Tomography
iii. Planar-integral projection reconstruction. was the algorithm first used in the EMI brain scanner de-
Fourier NMR imaging techniques suggested the pos- veloped by Hounsfield.
sibility of exciting the entire volume of an object, thereby The ART is not the major algorithm used in commercial
obtaining planar-integral data sets with which an efficient scanners today, however, because for practical purposes it
volume image reconstruction could be formed. is inefficient. It is generally agreed that the ART is slow
One-dimensional planar-integral projection data along in computation and also that iteration cannot begin before
the direction T at an angle (θ, φ) of a 3-D object function the completion of projection data collection. The use of
f (x , y , z ) can be given as ART based on iterative techniques, in general, is limited,
∞ with the exception of a few special cases, such as that of
limited view or limited scan angle. The basic algorithm is
pθ,φ (z ) = f (x , y , z )δ(T − z ) d x dy dz , (21)
simple and will not be discussed in detail here.
−∞
Tomography 851
scanning represents the radial set of the 2-D Fourier do- gyromagnetic ratio, and G x and G y are the x and y direc-
main object function, it is possible to form a set of polar tional gradient fields, respectively. As shown in Eq. (26),
rasters that represent the Fourier transform of the object the measured signal s(tx , t y ) is the 2-D inverse Fourier
function through 1-D Fourier transform of the projection transform of spin density f (x, y). Therefore, the spin den-
data. In the direct Fourier reconstruction (DFR) technique, sity image f˜(ωx , ω y ) can be reconstructed by taking the
therefore, the reconstruction procedures involve the inter- 2-D Fourier transform of the measured signal s(tx , t y ) as
polation problem as well as 2-D Fourier transform, con-
∞
version of data from polar form to Cartesian form, and 2-D
inverse Fourier transform of the interpolated data with the f˜(ωx , ω y ) = s(tx , t y ) exp(−itx ωx − it y ω y ) dtx dt y .
fast Fourier transform algorithm. This algorithm would −∞
852 Tomography
reconstructed to create the desired images, which in turn cess inevitably leads to quantization errors (noise), which
provide information on particular structures or metabolic can be expressed as
functions of the object under examination. The manipula-
tion and processing of data are usually performed by gen-
σq = √ , (29)
eral computers with array processors or by similar types of 2 3
image processors, such as backprojectors, depending on
where σq is the standard deviation of the quantization error
the data acquisition method and reconstruction algorithm
and is the step size of the ADC. It is important that
being used.
the quantization noise of an ADC be kept well below the
The third component is the display part, on which the
level of other sources of noise within the data acquisition
observer examines and analyzes the object. A multiformat
system.
camera is commonly attached to the display console to
Integral and differential linearities are other measures
record the pictures or images on film.
of imperfection arising from the component limitations in
In some systems these parts are independent modular
the circuitry and circuit design of the ADC.
components, and in others they are integrated as a single
At present, high-accuracy and high-speed ADCs of
package. A general system block diagram is shown in
16-bit resolution with conversion times of less than 20
Fig. 4.
to 30 µs are widely available. Ultra-high-speed conver-
sion with high resolution is still limited, however, by the
lack of high-speed and wide-range linear ADCs.
2. Data Acquisition
To make full use of the dynamic range of the ADC, the
The source and sensor system with controller compose input signal level should be scaled and shifted to the range
the main body of a CT system. System configuration, the of the ADC. This is usually achieved through the use of a
related physics, and the principles of particular CTs are preamplifier, attenuator, and/or level shifter. It is corrected
detailed in later sections. later or during data processing through the use of appro-
It is assumed that signals containing diagnostic infor- priate software or fixed hardware devices. In some cases,
mation can be measured in the form of electrical voltage, special converters such as log amplifiers or integrators are
current, or number of photons counted. If the detected used depending on the detected signal and the related pro-
signal is not in digital form, it is converted to digital form cessing algorithms. The output of the ADC can be stored
for computer use. This process is most often performed by two methods depending on the interface technique em-
rapidly by an analog-to-digital converter (ADC). Since all ployed: (1) It can be put into the memory of the data ac-
the data manipulations that follow are based on this digital quisition system until a set of data has been collected,
representation, the performance of the ADC is vital to the then transmitted to the computing system. (2) It can be
overall system performance and to the final results. continuously transmitted to the computing system as the
An ADC is characterized by a number of parameters, conversion operation occurs. In high-speed data transmis-
including resolution, integral and differential linearities, sion, the direct memory access technique is often used.
internal noise, and operating speed. The resolution of an An auxiliary function of the data acquisition system is
ADC is determined by the number of binary bits to which buffering, in which data are temporarily stored before be-
analog data can be digitized. For example, a 10-bit ADC ing sent to the computing system. Buffering also protects
has 1024 possible digitizations or output levels. Similarly, the computer system from the detector electronics, which
a 12-bit ADC has 4096 possible levels, while a 16-bit often involve high voltages.
converter has 65,536 possible levels, and so on. The finite The source and detectors are usually housed in what
number of levels or steps involved in the conversion pro- is referred to as the scanning gantry. The gantry is a
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 853
854 Tomography
4. Display
The display of the CT system provides a means of visual
examination of the scan information obtained through the
data acquisition system and computer. The CT display
system is the fundamental link between the CT image and
the human visual system, which provides the radiologist
with maximal diagnostic information. Display monitors
are usually attached to an operator console through which
all the commands are made.
To facilitate visualization of small differences in tis-
sue density, the value from each address of the display
FIGURE 6 Block diagram of the internal structure of an array
processor. memory is first modified by some transforms before be-
ing changed to an analog signal by a digital-to-analog con-
verter. Most present-day displays utilize 256 gray levels in
performs high-speed computation). In addition, an input–
accordance with human visual perception characteristics.
output interface can be used to store measured data and
In conventional CT display systems, a selected range of
to transfer the reconstructed image directly to the display
CT values is displayed uniformly over the range, which is
system without using the host computer. The speed ad-
preselected by setting window level and width. The win-
vantage gained by the array processor is made through the
dow level control determines the midpoint of the density
parallel processing of a large number of data read from
range, while the window width determines the width of
the memory of the processor, which uses its own bus, and
the range. By changing the window width and level, the
through the use of a pipeline structure in the arithmetic
observer can enhance the visualization of the subject struc-
unit. Distributive processing techniques can also be used
tures. In essence, the window width controls the contrast
to divide the processing load between the host and array
of the image, and the window level changes the density of
processor, maximizing the efficiency of both systems.
the tissue to be displayed. Decreasing the window width
The backprojector performs high-speed back-
usually results in a high-contrast image, but it also en-
projection. For example, many CT systems, especially
hances noise.
X-ray CT, are currently hindered by the backprojection
In display systems, image processors are usually uti-
operation, and therefore additional hardware computing
lized to facilitate the manipulation of image enhancement
devices like backprojectors are usually added to allow
and restoration (including filtering, selection of regions
the whole image reconstruction process to be carried out
of interest, cut-view display, zooming, and distance mea-
almost instantaneously.
surement) without interrupting the host computer.
The computational speed of the special computing pro-
cessor is often measured in units of million (mega) floating
point operations per second (MFLOP). Through the use
III. AREAS OF APPLICATION
of array processors, more than 100 MFLOP can be easily
attained. The entire processing of the data for the recon-
A. X-Ray Computerized Tomography
struction depicted in Table II, for example, can be com-
pleted in only a few seconds. X-Ray CT is a product of X-ray technology with advanced
Finally, reconstructed images are stored in the refresh computer signal processing and capable of generating a
memory, magnetic disk, or magnetic tape, depending on cross-sectional display of the body. It is, in fact, the origin
the amount of storage, transfer rate, access time, cost, and of the entire CT evolution, which began in the early 1970s.
other factors. The basic principle of X-ray CT involves X-ray gen-
Besides reconstruction, the computer provides machine eration, detection, digitization, processing, and computer
control, pulse sequence control, display, and data han- image reconstruction. As the X rays pass through the body,
dling. The dedicated microcomputer controller elements they are attenuated at different rates by different tissues.
are therefore becoming increasingly important. They al- The attenuated X rays are then collected by detectors and
low the implementation of highly efficient input–output converted to digital impulses or data by the ADCs. These
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 855
where µ, ρ, and Z are the linear attenuation coefficient, FIGURE 7 Linear attenuation coefficient of water and contribu-
density of the absorber, and atomic composition or num- tion of each interaction to the attenuation of X rays as a function
of the energy.
ber, respectively.
In the energy region where most commercial X-ray
CT systems are being engaged for medical tomography represents the integral value of the attenuation coefficients
(∼70 keV), two types of interactions are dominant: pho- along the path, the projection data taken at different views
toelectric absorption and coherent and incoherent scatter- are the basis for tomographic image reconstruction.
ing.
In photoelectric absorption, the X-ray photon is com- b. Beam hardening. X-Ray beams generally used
pletely absorbed by transferring all of its energy to an in X-ray CT are not monoenergetic and have a finite
electron. Scattering is subdivided into two components: spectrum. When this polychromatic X-ray beam passes
Rayleigh (coherent) scattering and Compton (incoherent) through a material, X rays of different energy in the spec-
scattering. In Rayleigh scattering, the direction of the trum undergo different attenuation, and as a result the out-
X-ray photon is changed but the energy is conserved. In put energy spectrum differs from the input spectrum; that
Compton scattering, on the other hand, both direction and is, lower energy X rays are attenuated more heavily than
energy are changed. Figure 7 depicts the interactions of higher energy X rays. This trend is accentuated if the path
the X rays with water. It shows the contribution of each length is large or if material possesses components of high
photon interaction to linear attenuation coefficients of dif- atomic number. The consequence of this nonuniform at-
ferent types as a function of energy. If the absorber is not tentuation of the polychromatic X-ray beam is called the
homogeneous, µ(ρ, Z ) is simply a space-variant function beam-hardening effect. It produces a visible artifact in
dependent on the material. By directing a monochromatic the final reconstructed image. The original EMI scanner
X-ray beam in the y direction, the output X-ray intensity utilized compensatory measures to deal with the beam-
I (x) is hardening effect. A water bag was used as a compensator
by surrounding the head so that the total path lengths of
I (x) = I0 (x) exp − µ(x, y) dy , (31) the X rays were always the same for all projections. An-
other method is to preharden the X-ray beam by passing
where I0 (x) and µ(x, y) are the incident X-ray intensity it through an aluminum or copper filter so that the output
and X-ray attenuation coefficient, respectively. By taking X ray is close to the monoenergetic beam before it passes
the logarithm and rearranging Eq. (31), one can obtain through the body.
projection data p(x) as
c. X-Ray source. Two types of X-ray sources are
p(x) = −ln[I (x)/I0 (x)], (32)
currently used in X-ray CT. The most simple type is the
where p(x) is a simple integration or summation of the to- fixed-anode X-ray tube, in which the anode is cooled by
tal attenuation coefficients along the X-ray path. In X-ray oil and is continuously energized. A typical focal spot size
CT, the contrast is the difference in attenuation coefficients of this type is about 2 mm × 16 mm on a 20◦ angle tung-
of the material involved. Since each set of projection data sten target. The relatively small heat dissipation capability
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
856 Tomography
associated with this type of tube limits the amount of indicators of the energy and fluence of the incident X-ray
photon flux generation, resulting in a statistically nois- photons on the scintillation crystal. An overall gain of a
ier image than others, for example, the rotating-anode few millions is common in most PMTs. Other detector
X-ray tube, when operated continuously. The rotating- crystals, such as BGO (bismuth germanate) and CaF2 , are
anode X-ray tube, on the other hand, allows for a large also used.
photon flux because it has much greater heat capacity. Scintillation Crystal–Photodiode Coupled Detector.
The performance of this detector, which was developed
d. Detectors. X-Ray photons are collected by radi- more recently than its counterparts, has been found to be
ation detectors of various kinds, such as a scintillation satisfactory. A typical scintillation crystal–photodiode
crystal coupled with a photomultiplier tube (PMT). In gen- detector comprises a CsI(T1) scintillation crystal, a
eral, the output of the detectors consists of electrical sig- P–N junction photodiode coupled to the crystal, and a
nals that are proportional to the incident X-ray energy or preamplifier for the low-level signal amplification.
fluence. The most important parameters to be considered The incident X-ray photon is converted to visible light
in the selection of detectors for X-ray CT are efficiency, re- in the scintillation crystal, which then falls into the PN
sponse time (or afterglow), and linearity. Efficiency refers junction photodiode. Generated electron-hole pairs are
to the absorption and conversion efficiency of the incident collected at the junctions. Since the generated current is
X rays to electrical signals. Linearity refers to the dynamic usually weak, a low-noise preamplifier is required. The
range of the detector response. Response time refers to the voltage output is proportional to the energy and fluence of
speed with which the detectors can detect X-ray photons the X-ray incidence on the detector.
and recover to detect the next photon. This is determined ii. Gas ionization detector. Some commercial X-ray
by the afterglow—one of the important characteristics of CT systems use gas ionization detectors. To improve the
detector materials for X-ray CT application. detector efficiency, xenon gas is pressurized by as much as
Detector types currently in use for X-ray CT can be 20 atm and filled in a long chamber to maximize detection
divided into two classes: scintillation detectors and gas efficiency. It consists of tungsten plates, which serve as
ionization detectors. anodes for electrons. When X-ray photons are incident on
i. Scintillation detector. Scintillation crystals [such as the detector cell, gas is ionized. This ionized gas provides
NaI(T1) and CsI(T1)] produce flashes of light as they ab- current that is directly proportional to the X-ray photon
sorb X-ray photons. The light is then converted to electri- energy. The advantages of the gas ionization detector are
cal signals by subsequent electronics. The following two high spatial resolution capability and simplicity. In addi-
types of scintillation detector systems are commonly used tion, the compact detector assembly can be made on a
in X-ray CT: large scale by packing a few hundred equivalent detector
Scintillation Crystal–PMT Coupled Detector. Light elements that have detector widths as narrow as 1 mm. The
produced in the crystal is coupled to the photocathode of disadvantage of the ionization detector, even with highly
a PMT. In the PMT, photoelectrons are generated from pressurized gas, is the low detection efficiency. Character-
the photocathode as the light strikes it. These electrons istics of each detector are briefly described in Table III.
are multiplied through a series of cascaded dynodes in
which electron multiplication processes take place. Each e. Data acquisition and reconstruction algorithms.
dynode produces more electrons than incident electrons. At each view, detector signals from the detector system
The multiplied or amplified electrons constitute output are converted to digital pulses by the ADC. These sig-
signals in the form of a charge or a current, which are the nals are collected by the computer via signal processing
NaI(T1)-PMT High detection efficiency, 100% at 70 keV(1-in. crystal) Afterglow; restricted dynamic range; low packing
density; hygroscopic
CaF2− PMT No afterglow Low detection efficiency (62% with 1-in. thick)
BGO (bismuth germanate)-PMT No afterglow; high detection efficiency; nonhygroscopic Low light output
CsI(T1)-photodiode Good spectral match with available PMTs; high detection —
efficiency (94.5% at 120 keV with 5-mm thick)
Xenon gas ionization detector Simple and compact; no afterglow; high Low efficiency; possible instability; slow response time
resolution capability
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 857
electronics for image reconstruction. The steps of the incremental 1◦ rotation followed in preparation for the
view angle between successive views are normally ∼1◦ upcoming scanning. This procedure is depicted in Fig. 9a.
or less, and a few hundred views are usually taken from The collection of projection data needed for tomographic
each tomographic slice. After the projection data sets (i.e., image reconstruction of a slice took several minutes.
line-integral projection data) are obtained, image recon-
struction takes place. The time required for image recon- b. Second generation. The design of the second-
struction can be relatively long in comparison to data ac- generation CT incorporated a narrow-angle fan beam
quisition time. Therefore, the procedure usually requires X ray and an array of multiple detectors. Since the di-
special-purpose computer peripherals, such as array pro- verging fan beams passing through the patient increase
cessors or backprojectors. data collection channels, the number of angular rotations
required could be reduced. Therefore, the scan time in this
f. System configuration. Figure 8 is a block diagram second generation was shortened substantially; the nomi-
of a typical X-ray CT system currently in use. A 16- to nal scan time was ∼20 s. The second-generation scanner
32-bit minicomputer system equipped with a dedicated still entailed translational motion as well as rotational
backprojector is normally used for data acquisition, signal motion to cover the object fully, however. Figure 9b il-
processing, and system control. lustrates the configuration of the second-generation CT
scanner.
2. System Evolution
c. Third generation. In the third generation, the fan
Although remarkable progress has been made in all as- beam angle is widened, and the fan covers the entire object
pects of X-ray CT since it was introduced in 1972 by to be scanned. Each projection path is defined by a match-
Hounsfield at EMI, the most significant changes have ing detector, which can be either a small and narrow scintil-
been made in the area of data acquisition. These stages lation detector slab or a segment of a gas ionization cham-
of progress are classified into “generations.” Since 1972, ber. Because the entire object is covered or encompassed,
the X-ray CT has evolved from the first generation to the no translational movement is required. Therefore, both the
fourth and possibly to the fifth generation, which will in- X-ray tube and the detector array need only simple rota-
corporate the latest developments in dynamic scanners. tional motion around a fixed axial center. The entire 360◦
is usually scanned for whole data collection. Scan time is
a. First generation. The first generation naturally as short as 3 s. This high speed scan capability allows for
entailed the first EMI scanner developed by Hounsfield. synchronized rotational motion with physiological signals
This scanner used a single pencil beam and a single such as that obtainable from electrocardiograph (ECG),
detector, which translated and rotated synchronously. when the imaging of moving organs is required. The ma-
There was translational motion across the object being jor drawback of this configuration is that the effects of
scanned, and at the end of each translational motion an the drift of the detector are cumulative so that artifacts
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
858 Tomography
FIGURE 9 X-Ray CT system evolution from first- to fourth-generation scanners. (a) First generation; (b) second
generation; (c) third generation; (d) fourth generation.
appear on the reconstructed image. Figure 9c shows the reconstructor (DSR) and cardiovascular CT (CV CT).
configuration of the third-generation CT scanner. Almost These can be categorized as the fifth generation X-ray CT
all third-generation scanners use pulsed X-ray sources to scanners.
take advantage of significant dead time between succes- i. Dynamic spatial reconstructor. The development
sive views. of the DSR at the Mayo Clinic was completed in 1982.
It can produce real-time images of body organs in motion.
d. Fourth generation. The construction of a station- The DSR comprises 28 X-ray sources with 28 opposing
ary circular ring detector array is probably the ideal choice X-ray imaging chains, which are image intensifiers cou-
of the detector for the rotating X-ray source. A striking pled with X-ray detection phosphors mounted in the same
analogy is the circular ring PET scanners of various types gantry. The physical size of components and the required
developed during the 1970s. This stationary ring is the radiation flux determine the number of X-ray sources and
most distinct feature of the fourth-generation X-ray CT. imaging chains. As many as 240 images of adjacent slices
The X-ray source rotates, but the detector array does not. A with a thickness of 1 mm can be made from a cylindri-
wide-angle fan beam X ray encompasses the entire patient, cal volume that is 38 cm in diameter and 24-cm long. A
and 600 or more stationary detectors form a circular ring complete volume scan is achieved in 0.01 s after each of
array. With this kind of configuration, detector drift is not the 28 X-ray sources is pulsed in succession (for 0.34 ms).
cumulative and therefore can be corrected. The advantages These scans can be repeated 60 times per second. A high
of this system are similar to those of the third-generation temporal resolution image can be obtained with 28 angles
systems, but the main drawback of the third-generation of view recorded in 0.01 s if the gantry is kept stationary.
systems—drift effects—has been eliminated since detec- For stationary objects, however, high spatial and density
tor drifts no longer accumulate over successive views. The resolution images can be produced using all 240 views
fourth-generation systems are generally more expensive, for reconstruction. Trade-offs between temporal, spatial,
however, due to the large number of scintillation detec- and density resolution can be made by selecting the ap-
tors and PMTs employed. Figure 9d shows a schematic propriate subsets from the total projection data. Figure 10
diagram of the fourth-generation X-ray CT scanner. depicts a schematic diagram of the DSR scanner.
ii. Cardiovascular computerized tomography. This
e. Dynamic scanner. For the imaging of moving or- system was proposed and developed by Boyd et al. at the
gans, such as the heart, ultrafast scanners are required. University of California, San Francisco, with the same
Typical systems of this kind are the dynamic spatial goals in mind, but with a more compact and physically
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 859
860 Tomography
imaging with a resolution as high as 2 to 3 mm full width or strip joining the two detectors, the physical collimation
at half-maximum (FWHM). used in SPECT to confine the direction of incoming pho-
This section is devoted to the instrumentation aspects tons can be eliminated. For this reason, coincidence detec-
of PET and is divided into three parts. The first part in- tion is also called electronic collimation. The total number
troduces some basic principles, the second deals with var- of coincidence events detected by a given pair of detec-
ious physical factors affecting system performance, and tors constitutes a measure of the integrated radioactivity
the third discusses the evolution of PET system designs. (i.e., line-integral projection data) along the strip joining
two detectors. From a complete set of line-integral projec-
a. Basic principles. tion data obtained from several views with detectors sur-
i. Positron emitters and physics. Positron-emitting ra- rounding an object, the activity distribution within the slice
dionuclides possess several important physical properties can be reconstructed by using the algorithms discussed in
that make PET a unique imaging technique. Its most im- Section II.A.
portant property is the directionality and simultaneity of The accuracy of the spatial localization of a positron-
photons generated by annihilation. The emitted positron emitting radionuclide by the coincidence detection pro-
combines with a nearby electron, and two photons are gen- cedure is limited, however, by two physical properties of
erated by a phenomenon known as annihilation. These two positron annihilation: (1) The two annihilation photons
annihilation photons, each with energy of ∼511 keV, are are not exactly collinear, thereby creating angular uncer-
generated simultaneously (simultaneity), and they travel tainty, and (2) there is uncertainty about the annihilation
at ∼180◦ from one another (directionality). The nearly position of the emitted positrons, that is, uncertainty about
collinear direction of the two annihilation photons makes positron ranges. The latter is strongly dependent on the ki-
possible the identification of the annihilation event or the netic energy of the emitted positron. The combined effect
existence of positron emitters through the detection of two of these two factors introduces a fundamental uncertainty
photons. This is usually achieved by a coincidence detec- in locating the sources of the positrons. This uncertainty,
tion circuit that records an event only if both detectors depending on what kind of radionuclides are used and on
sense annihilation photons simultaneously (solid lines, detector separation, is typically about 2 to 3 mm FWHM.
Fig. 11). Because the two detectors record coincidence This value is accepted as a lower limit on the resolution
events only from a volume of space defined by a column that can be achieved by a positron camera. Despite these
inherent spatial resolution limits, PET could potentially
be the highest resolution nuclear imaging technique.
ii. Advantages of PET. Perhaps the greatest advan-
tage of PET over other nuclear imaging systems is the
electronic collimation with which a solid angle is extended
to virtually the entire object. The sensitivity (photon col-
lection capability) gain due to this large detection solid
angle that results from electronic collimation is greater
than that of SPECT. This is considered to be an important
advantage of PET over SPECT.
Other advantages of PET, which uses electronic col-
limation, are the uniformity of resolution and sensitivity
over the entire range. The latter, uniform sensitivity, stems
from the fact that the combined attenuation affecting a
photon pair is the same, regardless of the position of the
annihilation, as long as the annihilation points are inside
the column defined by a detector pair. This allows us to
compensate accurately for the attenuation suffered within
the object. This is again a distinct advantage of PET over
SPECT.
Another advantage of PET consists of the physiological
aspects of most of the available positron-emitting radionu-
clides, which are usually of low atomic number. Among
FIGURE 11 Principle of coincidence detection. True coincidence these radionuclides, 11 C, 13 N, 15 O, and 18 F are used
(solid line), random coincidence (dashed line), and scattered co- most often in PET because of their physiological affin-
incidence (broken line) are indicated. ity in the human body and their short physical half-lives
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 861
(11 C, 20.34 min, 13 N, 9.96 min; 15 O, 2.05 min; 18 F, TABLE IV Properties of Detector Materials Used in Positron
Cameras
110 min), which facilitate effective imaging with mini-
mal dosages to the patient. The radionuclides 11 C, 13 N, Detector material Nal(T1) BGO CsF BaF2
and 15 O are the major components of molecules in living
Density (g/cm3 ) 3.67 7.13 4.64 4.89
matter. Therefore, they are closely related to the metabolic
Linear attenuation coefficient 0.34 0.92 0.44 0.47
processes in human physiology. at 511 keV (cm−1 )
Scintillation decay time (ns) 250 300 5 0.8
b. Physical factors affecting system performance.
Emission wavelength (Å) 4100 4800 3900 2250
i. Detector and related materials. The spatial reso-
Energy resolution in FWHM >7 >10 23 13
lution of the positron camera depends on the width of at 511 keV (%)
the detection channel and the sampling interval. The for-
mer sets the resolution limit obtainable with the system,
provided that the sampling requirement or the Nyquist design, (i.e., half the detector width). In projection data
sampling criterion is met. The detection channel can be obtained with a detector width of w, a maximum spatial
characterized by the detector aperture function and can frequency is about 2/w, because the intrinsic resolution
be determined by simple ray tracing. At the center of expected is about half the detector width. The desired sam-
a detector pair, it can be represented by a triangle with pling distance, therefore, should be less than w/4.
an FWHM equal to half the detector width. The aperture In the hexagonal- or octagonal-geometry PET system,
function toward the detector becomes trapezoidal and ul- desired sampling can easily be achieved by introducing
timately rectangular at the position immediately adjacent translational and/or rotational motions (Fig. 12a). In the
to the detector. circular ring system, however, improving the sampling ar-
In most PET instruments, a standard γ -ray detector de- bitrarily has been difficult, and several sampling schemes
sign is chosen for detector assembly. It consists of a PMT for overcoming this inherent difficulty have been proposed
and a scintillation crystal coupled together. and implemented in experimental systems. For example,
Several scintillation crystals used in the past are being wobbling motion has been widely used (Fig. 12b). How-
recommended for PET design. In the early 1970s, NaI(T1) ever, two sampling schemes suitable to the circular ring
was the most commonly used scintillation detector for
PET, even though it was comparatively inefficient in high-
energy 511 keV photon detection. A more recent devel-
opment in scintillation detectors was the introduction of
BGO, which led to substantial improvement in detection
efficiency. The nonhygroscopic nature of BGO crystals
further facilitates compact detector assembly or packing,
which also leads to an overall increase in detection effi-
ciency. The design of a high-resolution PET system was
made possible by the introduction of BGO, since an ex-
tremely narrow slab of each detector crystal (4–5 mm in
width) is considered to be the essential requirement of the
high-resolution PET.
Two other crystals have been considered for PET ap-
plication, CsF and BaF2 , both of which have very rapid
detection capabilities. Therefore, these crystals have made
excellent time resolution feasible. This advantage has been
exploited in time-of-flight (TOF) PET systems.
Some of the physical properties of the detector materials
mentioned here are listed in Table IV. FIGURE 12 Sampling motions. (a) Hexagonal geometry with
ii. Sampling. Because detector width is limited by de- translational and rotational motions. (b) Wobbling motion applied
tection efficiency, the widths of the detectors are usually to circular ring systems. In this scheme, the entire ring moves in
kept as large as the resolution allows. With a given detector a circular path (small track of circle shown at the ring center); this
is 2-D motion. (c) Dichotomic motion applied to circular ring sys-
size, maximum resolution can be achieved by satisfying
tems. In this scheme, two half-arcs, which form a ring, move back
the Nyquist sampling criterion; that is, the sampling dis- and forth in opposite directions along the circular path. Note that
tance must be less than half the distance of the highest the motion is 1-D. (d) Positology. In this case, continuous rotation
spatial resolution obtainable by that particular ring PET of a ring with unevenly spaced detectors provides finer sampling.
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
862 Tomography
system have been developed, namely, dichotomic sam- dences produce a haze of background over the field of
pling and clam shell sampling. view in the reconstructed image.
Although the wobbling scheme is one of the most For a uniform distribution of activity, a formula for the
common methods, it is generally believed that the sam- random count rate has been established and given by
plings obtained by this method are usually neither uniform
Cr = τ f d Cs2 (counts s−1 µCi−1 cm−3 ), (34)
nor equally spaced. To obtain more uniform and equally
spaced samplings with the fewest possible number of where Cs is the single count rate for the entire ring, τ is
motions, a new sampling scheme, known as dichotomic the coincidence resolving time, and f d is the fraction of
sampling, has been proposed and incorporated into an ex- detectors covering the object in the whole detector ring in
perimental system (Fig. 12c). This scheme employs two coincidence with any one given detector, that is,
half-rings (from which the term dichotomic is derived),
f d = (2/π ) sin−1 (d/D). (35)
which rotate in such a way that finely sampled parallel
or fan data sets can be obtained with a minimal number Because both the single rate and true coincidence rate
of scan stops. Another scheme, developed at the Univer- are proportional to the amount of activity, it is apparent
sity of California, Berkeley, is similar to the dichotomic from Eq. (34) that the random coincidence rate is propor-
sampling scheme but employs a slightly different motion. tional to the square of the true coincidence rate.
Another method for improving the sampling scheme To reduce the random coincidences, it is imperative to
is the positology (Fig. 12d) developed by Tanaka et al. in minimize single counts. Because single counts can arise
Japan. In this scheme, rapid rotational motion is employed from both in-slice and out-of-slice annihilation events,
with a nonuniformly spaced circular detector array. With they can be suppressed by limiting the detection channels
this scheme, through the rotation of the entire ring, the by slice collimation or by the increase in the energy thresh-
desired sampling can be achieved. old so that maximum rejection of any scattered single
iii. Sensitivity. Sensitivity in PET is defined as the events can be achieved. It is common practice to reduce
capability of detecting the true coincidences (solid line, as many annihilation events as possible from out of slice
Fig. 11) with a given amount of radioactivity. Sensitivity by using annular interdetector ring collimators.
has traditionally been measured with a phantom of diame- Random coincidences can be partially corrected by soft-
ter d = 20 cm, filled with a uniform activity concentration ware, provided that the single counts of each channel and
ρ (µCi/cm3 ) (1µCi = 37,000 disintegrations per second). the coincidence resolving time are recorded.
By considering several factors, including the activity in v. Scattered coincidences. Scattered coincidences oc-
the field of view, the self-attenuation of γ rays within the cur when one or both γ rays resulting from an annihilation
phantom, the solid angle subtended by the detector ar- event are scattered in the medium and detected with the
ray, and the detection efficiency of the array, an empirical remaining energy of the γ rays above the energy threshold
formula for sensitivity measure is derived and given by (broken line, Fig. 11). They give incorrect positional infor-
mation and produce a line-spread function with long tails.
Ct = 14,500ραε2 h 2 d 2 /D (counts s−1 µCi−1 cm−3 ). The number of scattered coincidences can be reduced by
(33) setting the energy threshold level high and by using tighter
interslice collimation.
In this equation, α is the probability of no scatter of both vi. Correction of random and scattered coincidences.
annihilation photons, ε is the detector efficiency including Random coincidences can be corrected by two simple pro-
the detector packing ratio, h is the thickness of slice to be cedures. One is by using the delayed coincidence mea-
imaged, d is the diameter of the phantom, and D is the surement with the same time window and the other, men-
ring diameter. tioned earlier, is by estimating random coincidences using
If the system contains two or more detector rings, co- Eq. (34). As stated above, random coincidences can be re-
incidence can be measured between the detectors in the duced either by minimizing the coincidence time window
different rings (coincidences not in the planes that are or by reducing the activity.
perpendicular to the axis) to increase the sensitivity of a Since the scattered coincidences are an inherent physi-
given imaging plane. Sensitivity is higher in the cross-slice cal property, they cannot be easily reduced or eliminated
planes because they involve twice as many detectors. as random coincidences. Because it is prompt in nature,
iv. Random coincidences. As the source of back- differentiation between true and scatter simply by mini-
ground noise in PET images, random or accidental coin- mization of the time window is therefore difficult. In addi-
cidences occur when two photons emitted from two tion, the energy loss in small-angle scatter is so small that
independent positions are detected within the coincidence it is difficult to differentiate through the energy window. In
resolving time τ (dashed line, Fig. 11). Random coinci- the case of multilayer ring geometry in imaging multiple
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 863
slices, scatter elimination is achieved by the use of tighter geometry (the arrangement of detector arrays) is the most
interslice collimation. However, one should admit some basic design choice, because it determines fundamental
loss of observable volume when large-volume imaging is system performance. It can be categorized into three basic
of importance. types: planar, polygonal, and circular ring. To cover the
vii. Attenuation correction. Attenuation correction is imaging volume in the axial direction, multiring systems
one of the most important parts of PET imaging. A va- have appeared in which several rings are stacked on one an-
riety of correction schemes have been developed in the other. They provide high total sensitivity and offer 2N − 1
past. One of the simplest and crudest ways of correcting (N is the number of rings) image slices simultaneously.
attenuation is through the use of geometric shape, that is, The trend appears to be toward more generalized circu-
by finding edge contours and using contour information lar ring types of various forms. Among the advantages of
to determine attenuation lengths for the subsequent cor- circular ring geometry are uniformity, high sensitivity due
rection. In this case, attenuation coefficients are custom- to high packing fraction, and high angular sampling capa-
arily assumed constant. A more accurate method is to use bility. Because the requirement for angular sampling can
the transmission scan information obtained by an external be met even by the stationary ring, systems do not require
positron source surrounding the patient. This technique further rotation to improve angular sampling. Although
often suffers from statistical noise and thereby makes at- linear sampling has been a limiting factor in the circu-
tenuation correction difficult, unless a sufficient amount lar ring system, various solutions to this problem have
of data is taken. The correction can also be made through also been suggested, and their efficacy has been proved
X-ray CT by a procedure similar to that described earlier, experimentally. It is also worth noting that trends are in
but one should be aware that attenuation coefficients are the direction of volume imaging, either through multilayer
different from those obtained with 511-keV photons. rings or through a spherical-PET (S-PET) system to be de-
scribed later. Yet another avenue, one that requires further
c. Evolution of positron camera development. technological development, is the TOF approach. This is
There have been continuous efforts to develop PET imag- considered to be an adjunct to improve system resolution.
ing systems since the early 1950s. Some notable examples ii. Planar system. The planar type consists of two de-
include two NaI detector systems developed by Brownell tector planes (made of either discrete crystal arrays or
and Sweet in the early 1950s, 32 discrete NaI(Tl) detector position-sensitive devices, such as the Auger camera or
systems developed by Rankowitz et al.in 1962, and the multiwire proportional chamber) facing one another. A
PC-I developed by the Massachusetts General Hospital set of projection data can be obtained by rotating the dual
(MGH) with two banks of detectors (127 detectors planes around the patient, from which a series of image
per bank) in 1972. Although these systems ultimately slices covering sufficient axial volume is reconstructed. A
were intended for tomographic imaging, they remained disadvantage of this type is the large number of angular
quasi-tomographic machines until the introduction of the rotations it requires.
X-ray CT scanner by Hounsfield in 1972. MGH systems (PC-I, PC-II) and the commercial ver-
The first tomographic systems developed after the intro- sion of those systems (TCC 4200) fall into the category of
duction of Hounsfield’s model were the PETT (positron discrete crystal arrays.
emission transaxial tomograph) I–III series designed by iii. Polygonal system. Such systems as PETT III and
Ter-Pogossian and Phelps of St. Louis and CRTAPC (cir- PETT IV by Ter-Pogossian and Phelps and ECAT and
cular ring transaxial positron camera) designed by Cho NeuroECAT by Ortec fall into this category. With the ex-
et al. at UCLA. As a hexagonal detector array, the former ception of the octagonal-shaped NeuroECAT, these sys-
employed both translational and rotational motions, while tems are hexagonal. In this type of system, coincidence de-
the latter remained virtually stationary. Since these devel- tection channels are formed between those banks opposing
opments, PET instruments have rapidly improved through one another so that ring efficiency is usually limited, par-
the formulation of a variety of new concepts. For exam- ticularly toward the periphery of the image. An advanta-
ple, system geometry has evolved from planar to hexag- geous feature of the hexagonal system is the relatively sim-
onal type, from hexagonal to circular type, from single ple translational and rotational sampling motions, which
ring to multiring, and so on, Detector material has been fulfill the requirements of uniform linear and angular sam-
changed from NaI(Tl) to BGO, CsF, or BaF2 . The time plings.
of flight (TOF) technique has also been introduced. Also, iv. Circular ring systems. A natural extension of the
several commercial companies have begun to design PET polygonal PET system is circular ring geometry, which
systems. provides uniformity as well as natural symmetry. The first
i. System geometry. It is interesting to observe how circular ring PET system was conceived and developed
system geometry has evolved since the mid-1970s. System by Cho et al. at UCLA in 1975. Various other circular
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
864 Tomography
Tomography 865
of the magnetic system. In this case, it corresponds to the electronics. A main-field strength of 0.5 to 20 kG is used
frequency of gyroscopic precession of the magnetic mo- with a field gradient of 0.01 to 1 G/cm, formed by x, y, and
ment of nuclei in an external static magnetic field. Because z gradient coils. In the case of proton imaging with these
the magnetic resonance frequencies fall typically into the strengths of magnetic field and gradients, the required RF
radio frequency range for nuclear spins, we often use the range is approximately 2 to 85 MHz. The expected reso-
term radio frequency in referring to NMR. In NMR CT, lution in a conventional NMR imaging system depends on
we select a region from samples and obtain spatial distri- the field homogeneity and the available strength of field
butions of nuclear spins through the application of delib- gradients.
erately added spatial field gradients and RF signals. Thus, Two main types of imaging methods—direct Fourier
the cross-sectional images of an object are obtained. imaging (Kumar–Welti–Ernst method) and projection
The advantages of NMR CT are its nonhazardous na- reconstruction—and the possibility of parameter imaging
ture, its high-resolution capability, its potential for chemi- will be discussed in detail here. Hardware configurations
cally specific imaging, its capability of obtaining anatom- and the related statistical aspects of image quality and
ical cross-sectional images in any direction, and its high imaging times will also be discussed briefly.
tissue discrimination capability (high-contrast resolution A typical NMR tomographic imaging system designed
among different tissues). Although it has some minor dis- for human imaging is depicted in Fig. 13a. In this figure, a
advantages, such as its inherently long data acquisition split-solenoidal type of magnet NMR CT system is shown.
time due to spin–lattice relaxation time and low signal- The sample is surrounded by an RF coil (Fig. 13b) and a
to-noise ratio, due to its many advantages the NMR CT gradient coil set (Fig. 13c, d, and e). The configurations
is rapidly becoming a major diagnostic tool. NMR CT is of the magnet and the gradient coils may differ depend-
overcoming the problem of time-consuming data acquisi- ing on the design scheme, but the basic concept will be
tion through the development of new high-speed imaging similar for the majority of NMR imaging systems in the
methods and is surmounting the problem of low signal- future.
to-noise ratio through the use of high-field magnets (su-
perconducting magnets) with a magnetic field as high as
1. Principles of Nuclear Magnetic
2.0 tesla (T) (1 T is equal to 10 kG) or more.
Resonance Tomography
In the early 1970s, both Lauterbur and Damadian
showed that NMR spectroscopic techniques could be a. Nuclear magnetic resonance physics. Since
applied to imaging the human body and demonstrated NMR was discovered in 1946, it has become an indispens-
that these techniques might eventually be applicable able analytical method and tool in chemistry and physics.
to diagnostic imaging techniques similar to those of Although the basic physical properties of NMR are well
X-ray CT. In 1978, Andrew demonstrated the very high known and information on NMR can readily be found
resolution capability of NMR by obtaining a fine and elsewhere, we shall discuss a few topics necessary for un-
detailed image of the submillimeter septum of a small derstanding NMR as an imaging tool.
lemon. Moore and Holland soon followed with images of All materials, whether organic or inorganic, contain nu-
the human head, demonstrating definitively the potential clei, which are protons, neutrons, or a combination of both.
of NMR tomography in diagnostic imaging. Nuclei that contain an odd number of protons, neutrons,
NMR tomographic images can be formed by direct or both in combination possess a nuclear “spin” and a
mapping, projection reconstruction, or Fourier imaging. “magnetic moment.” This situation is equivalent to the
To date, two kinds of potentially useful imaging met- aggregation of many small magnets. In the real world
hods—direct Fourier imaging and projection recon- many materials are composed of several nuclei and the
struction—are most widely used. most common nuclei with magnetic moment are 1 H, 2 H,
7
An interesting aspect of NMR imaging is its diversity Li, 13 C, 23 Na, 31 P, and 127 I. Although some materials
in image formation, data collection, and reconstruction. are composed of nuclei with an even number of protons
Many different imaging and data processing methods are and neutrons that possess no spin or magnetic moment,
presently known and will be discussed from various points they often contain some nuclei with an odd number of
of view, such as imaging time, available field gradients, protons or neutrons. Therefore, they are subjects of NMR
pulse strength, speed, signal-to-noise ratio, and artifacts imaging. For this reason, NMR is practically applicable to
associated with the restored object image. most solid- and liquid-phase materials. Among the many
Great advances have also been made in the area of in- hundreds of known stable nuclei, more than 100 possess
strumentation for whole-body NMR tomography. The for- spin and magnetic moment.
mation of an NMR CT system requires a magnet, gradient When a given material is placed in a magnetic field,
coils, RF coils, computer and peripherals, and associated some of its randomly oriented nuclei experience external
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
866 Tomography
FIGURE 13 Sketch of an NMR tomograph. (a) Physical arrangement of the central part of an NMR CT system. A,
Main magnet of the split-solenoidal type used in many superconducting magnets; B, gradient coil set; C, RF coil in the
saddle type. (b) Saddle-shaped RF coil. Although the optimized shape for the homogeneous RF field is θ = 120◦ and
I /d = 2, the shape can be changed slightly to accommodate the different sample shape or number of coil turns. (c)
z-directional gradient coil. The change of the magnetic field in the z-direction is indicated by the length of the arrows.
This coil is called the Maxwell pair. (d) y-directional gradient coil; often known as the Golay coil.√(e) x-directional
gradient coil; also of the Golay type. In (c), (d), and (e), the optimal coil shapes are θ = 120◦ , I /d = 3/2, I 1 = 0.78r ,
and I 2 = 2.13r .
magnetic torque, which tends to align the nuclei in both scope precessing around the direction of the gravitational
parallel and antiparallel directions to the applied magnetic field. The rotating or precessional frequency of the spins,
field. The fraction of magnetized nuclei in the direction called the Larmor precession frequency, is proportional to
parallel to the applied magnetic field is limited by thermal the magnetic field strength.
agitation. Therefore, it is also limited by the temperature Another important phenomenon of NMR is the cre-
and main magnetic field strength. Because this fraction is ation of an energy “absorption state” (from a statistical
relatively small at room temperature, it has been a limit- point of view) by the applied external magnetic field.
ing factor in the sensitivity of NMR imaging. The spinning The proton has an intrinsic angular momentum or spin of
nucleus responds to the external magnetic field like a gyro- h✏ /2, where h✏ is Planck’s constant divided by 2π . When
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 867
868 Tomography
Tomography 869
transverse relaxation time resulting from field inhomo- In general, T1 , T2 , T2∗ , and T2∗∗ have the following rela-
geneity alone is expressed as tionship:
1 T2∗ = 1/T2 + γ H/2, (44) T2∗∗ ≤ T2∗ ≤ T2 ≤ T1 . (47)
where H is the field inhomogeneity, that is, the maxi- When the two relaxation mechanisms are considered, the
mum deviation of magnetic field over the object region. Bloch equation can be written
When a field gradient is added to resolve the spatial dis-
tribution of spin density, T2∗ is further reduced to T2∗∗ , as d Mz M z − M0
= γ (M0 × H0 )z −
given by dt T1
(48)
1 T2∗∗ = 1 T2∗ + γ G R, (45) d Mx y Mx y
= γ (M0 × H0 )x y − ,
where G (in gauss per centimeter) is the gradient field dt T2
strength and R (in centimeters) is the object diameter. where (·)z and (·)x y represent z and x–y components, re-
The composite sinusoidal signal decaying with an effec- spectively. Equation (48) indicates that the magnetization
tive transverse relaxation time T2∗∗ is then detected with components Mz and Mx y are independently related to the
a phase-sensitive detector. The results are similar to a de- relaxation times T1 and T2 .
caying demodulated AM signal, as shown in Fig. 14a. Several forms of spin-echo techniques play a central
Concurrently, longitudinal or spin–lattice relaxation and essential role in data acquisition for NMR imaging.
forces the spins to realign in the H0 (or z) direction because The two basic forms of the spin-echo technique applicable
it is the lowest energy state or thermal equilibrium state. to NMR imaging are the Hahn spin-echo technique and
Because it involves energy dissipation through the lattice, the Carr–Purcell Meiboom–Gill (CPMG) technique. In the
the longitudinal relaxation time T1 is usually longer than Hahn spin-echo technique, a 90◦ RF pulse is applied to the
T2 and is related to the z component of magnetization, as direction of the x axis, and then the magnetization vec-
stated in the following (see Fig. 14b), tor M rotates to the y axis. The spin magnetizations then
dephase over time, due to field inhomogeneity or added
Mz
M z = M0 1 − 1 − exp(−t/T1 ) , (46) field gradients. A subsequent 180◦ pulse applied along the
M0 x axis rotates the spins around the x axis (Fig. 19b). The
where Mz is the z component of magnetization at the start- spin magnetizations now continue to precess but begin to
ing time of relaxation. rephase. This process is equivalent to a focusing or regrow-
The two relaxation processes work simultaneously and ing process of the FID signal at twice the dephasing time.
vary greatly depending on the characteristics of the mate- At this point, all the spins are completely rephased along
rial. In the case of tissue in field strength of 1 to 20 kG, the −y axis, but the magnetizations have now decayed by
for instance, T1 and T2 are of the order of 0.5 s and T2 relaxation.
50 ms, respectively; T1 is usually larger than T2 . In Fig. 18, In the CPMG method, a 180◦ pulse is applied along the
sequential pictures of the relaxation processes are shown. y axis instead of the x axis, so that the spin flips around
FIGURE 18 Sequential illustrations of the spin relaxation processes. (a) Spins are flipped by the RF pulse H1 ; (b)
spins are dephased due to the spin–spin relaxation and field inhomogeneity; (c) FID signal decays to zero as the spins
lose phase coherence; (d), (e), (f) spins relax to the original equilibrium state by the spin–lattice relaxation process.
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
870 Tomography
FIGURE 19 Hahn spin echo: (a) Spin magnetizations dephase after 90◦ rotating by RF pulse; (b) 180◦ pulse is
applied along x axis; (c) the spins are being refocused; (d) spin echo is generated along −y axis. Carr–Purcell and
Meiboom–Gill spin echo: (e) Spins are dephased after 90◦ rotating by RF pulse; (f) 180◦ pulse is applied along y
axis; (g) the spins are being refocused; (h) spin echo is generated along y axis.
the y axis (Fig. 19f). Both techniques are actively used the effects of T1 and T2 relaxation times are not included;
in all phases of NMR imaging to reduce several adverse they will be discussed in a later section. The generated FID
effects that arise in actual data collection, such as field in- is, in effect, a Fourier transform-domain representation of
homogeneity and the effects of the gradient pulse rise time. the spin density distribution. From this fundamental 3-D
equation, many equations for the imaging algorithms de-
b. Basic theory of nuclear magnetic resonance scribed below can be derived.
tomography. Conventional NMR chemistry requires a
magnetic field of extreme homogeneity, in other words 2. Image Formation Algorithms
as uniform as possible, to reduce the frequency shift ef-
fect caused by the spatially dependent field variations. In One of the interesting aspects of NMR imaging is that the
Fourier NMR imaging, however, a field gradient or set of NMR images can be formed by many different procedures,
gradients is deliberately added to resolve the spatial dis- some of which are described in the following text. The
tribution of spins into Fourier domain components. The discussion is limited to the techniques used most often,
basic form of signal obtained from 3-D Fourier transform however.
NMR, which is known as FID, is expressed as In NMR imaging, data acquisition pulse sequences play
an important role and are intimately related to the image
∞ reconstruction algorithms employed. This description of
t
s(t) = M0 f (x, y, z) exp iγ [x G x (t ) the mathematical formulations is based on the basic pulse
−∞
0 sequence, which uses only 90◦ RF pulses, although in a
real imaging situation the spin-echo techniques that use
additional 180◦ RF pulses are more common.
+ yG y (t ) + zG z (t )] dt d x d y dz (49)
a. Direct Fourier imaging.
where f (x, y, z) is the 3-D spin density distribution and i. Algorithm for direct Fourier imaging technique.
G x (t), G y (t), and G z (t) are the time-dependent field gra- This direct Fourier imaging (DFI) method was first
dients along the x, y, and z axes, respectively. In Eq. (49) proposed by Kumar, Welti, and Ernst (KWE). In this case,
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 871
Hz (x, y, t)
H0 + G z z, for 0 < t < tz
= H0 + G y y, for tz < t < tz + t y (50)
H0 + G x x, for tz + t y < t < tz + t y + tx ,
where each timescale tz and t y is varied according to pre- FIGURE 20 Imaging sequences of 2-D direct Fourier imaging.
assigned sequences, that is, tz ta , t y 0 ∼ ta , where ta The slice in the z direction is selected and spin echo is used.
is the optimal observation time of FID. The FID signal is While G y remains constant, the intensity of Gx is varied for phase
sampled when the x gradient is applied, namely, during encoding. The purpose of the first part of the G y gradient pulse
is to dephase spins in the object after the nonselective 90◦ RF
the tx period. The sampled FID signal reflects the previ-
pulse. Only the spins in a designated slice are rephased by the
ous application of the z and y gradients by retaining the selective 180◦ pulse and form the spin echo at t = 2Ts on the
phase change caused by those gradients. For this reason, second constant G y gradient. The spin-echo signal is collected
this imaging scheme is often called the phase-encoding for image reconstruction.
method. The series of FID signals obtained with the var-
ious tz ’s and t y ’s then form a full 3-D FID signal set suf-
ficient for reconstruction of the spin density image of the z-directional selection gradient. Note that the x-direct-
entire volume. ional phase encoding is achieved by varying the amplitude
The observed FID signal s(tx , t y , tz ) is (neglecting the of the x gradient instead of varying the time interval.
ii. Time-multiplexed multislice imaging method. By
relaxation processes) expressed as
using consecutive excitations of the pulse sequence shown
∞ in Fig. 20 within a suitable repetition time period, it is pos-
s(tx , t y , tz ) = M0 f (x, y, z) exp[iγ (G x xtx sible to obtain several images simply through the use of
−∞ the remaining time, provided that the sum of data acquisi-
tion time of each slice is smaller than the repetition time.
+ G y yt y + G z ztz )] dz dy d x. (51)
For example, a data acquisition time of less than 100 ms
Fourier transform of Eq. (51) results in spatial spin density with a repetition time of 300 ms would allow three con-
function as secutive slice imagings without T1 relaxation constraints.
This multislice imaging method is a frequently used NMR
∞
imaging technique. It can also be applied to other 2-D
f˜(ωx , ω y , ωz ) = s(t) exp[−i(ωx tx + ω y t y imaging schemes, such as the line-integral projection re-
−∞ construction (LPR) technique, which is described in the
+ ωz tz )] dtz dt y dtx . (52) following section.
872 Tomography
be summarized as follows. The reconstructed 2-D image the spins in the sample are excited with a 90◦ RF pulse,
f (x, y) is given by and a subsequent 180◦ RF selects the slice. After spins
π are refocused they generate the spin-echo or FID signal
f (x, y) = [ pφ (x ) ∗ h(x )] dφ, (54) (Fig. 21). After 180◦ or 360◦ rotation of projection with an
0 appropriate step through the adjustment of the field gradi-
where pφ (x ) is projection data, h(x ) is the filter kernel ents G x and G y , a complete projection data set sufficient
that corrects 1/r blurring caused by circular symmetric for reconstruction of a slice at a given plane z 0 is obtained.
linear superposition, and (x , y ) is the coordinate system At this point, 2-D image reconstruction can proceed ac-
rotated by an angle φ from the original coordinates (x, y). cording to Eq. (54); that is, each echo or FID signal sφ (t)
In Fourier transform NMR, the nuclear signal can be is Fourier-transformed, convolved with a filter kernel, and
considered to be the inverse Fourier transform of the spa- backprojected.
tial domain spin density function. If a plane at z = z 0 is ii. Slice (plane)-encoded multislice LPR. The single-
selected, then the FID at an angular view φ can be ex- slice line-integral projection technique explained ear-
pressed as lier can be extended to achieve multislice imaging
∞ through several encoding techniques, for example, the
plane-encoding technique explained in the following
sφ (t) = M0 f (x , y ; z 0 ) exp[iγ x G x t] dy d x .
paragraphs.
−∞ Let us assume that the number of planes is n. For the
(55)
data set at a view φi , the same G x y and G z are applied
Although the FID signal sφ (t) appears in the time domain,
n times, each with a different frequency composition of
it represents the Fourier domain projection data. There-
RF pulses. The RF pulses are specially tailored to assign
fore, the projection data pφ (x ) are obtained through the
desired phases to the designated slices. To obtain a com-
Fourier transform of the FID signal as
plete set of view data corresponding to the n planes, the
pφ (x ) = [sφ (t); t → x ]. (56) acquisition of data is repeated n times with differently
composed RF pulses.
The basic form of projection data obtainable in Fourier The key to this method lies in the encoding of signals
transform NMR is similar to the data obtained in X-ray according to the RF pulse sequence. A simple illustration
CT. In Fig. 21, spin-echo signals or FIDs are obtained at of the encoding procedure using a coding matrix is as
different angular views through the application of the field follows. Let the FIDs obtained at each 180◦ composite
gradient and RF excitation sequences. As a first step, all RF pulse sequence be Sφ10 (t), Sφ20 (t), . . . , Sφn0 (t). Each FID
is a composite of the line-integral projection sets, which
include data from several planes at an angular view φ0 ,
that is, sφ0 z0 (t), sφ0 z1 (t), and so on. Therefore, composite
FIDs, Sφ10 (t), Sφ20 (t), . . . , Sφn0 (t) can be given as
Sφ10 (t) sφ0 z0 (t)
2 s (t)
S (t) φ0 z 1
Sφ0 (t) = φ0 = [Hn ] .. . (57)
.. .
.
sφ0 zn−1 (t)
Sφn0 (t)
From Eq. (57), the desired FID signal sφi zi , which cor-
responds to the FID of slice z i , can be obtained through
matrix inversion.
Examples of coding matrices include the Hadamard ma-
trix and the Fourier matrix. The advantage of this method
is the statistical improvement gained as a result of the
increase in total scanning time.
Tomography 873
originate from the entire volume. The FID signal of the a. Imaging modes.
total volume planar-integral projection data in the z i. Saturation recovery imaging. The saturation recov-
direction can be expressed as ery method involves simply repeating the pulse sequence
∞ ∞
at regular intervals T . The equations discussed previously
are unchanged except for the replacement of f (x, y, z)
s(t) = M0 f (x , y , z ) d x dy
−∞ with f (x, y, z), which is expressed as
−∞
f (x, y, z) = f (x, y, z){1 − exp[−T /T1 (x, y, z)]}. (60)
× exp[iγ (H0 + z G z )t] dz , (58)
Note that f is now a function of both T1 and f .
where (x , y , z ) are the rotated coordinates of (x, y, z) ii. Inversion recovery imaging. Inversion recovery is
and the z direction coincides with the projection direc- similar to saturation recovery, except that the 180◦ RF
tion. Through phase-sensitive detection, the FID signal pulse precedes the 90◦ RF pulse with a time interval of
obtained can be written T1 ; f (x, y, z) is related to f (x, y, z) as
∞
sθ,φ (t) = M0 pθ,φ (z ) exp(iγ z G z t) dz , (59) f (x, y, z) = f (x, y, z){1 − 2 exp[−T1 /T1 (x, y, z)]}.
−∞
(61)
where pθ,φ (z ) is the planar-integral projection data with
the angular view (θ, φ). It is easy to see the increased dependency of the image
In Eq. (59), sθ,φ (t) represents the projection data in on T1 over that obtained in saturation recovery. This tech-
the Fourier domain. Therefore, the Fourier transform of nique is often used for measuring T1 values in tissues.
sθ,φ (t) is spatial domain planar-integral projection data Figure 22 shows the pulse sequences for 2-D inversion
with which reconstruction can be performed (see the PPR recovery Fourier imaging.
algorithm in Section II.A). iii. Spin-echo imaging. Through the application of the
180◦ pulse following the first 90◦ pulse at t = Ts , spins are
d. Echo-planar imaging method. One of the inher- refocused at t = 2Ts by the spin echo (Fig. 20). Although
ent disadvantages of NMR CT is the long data acquisition the spins are now refocused and coherent, the amplitude
time caused by constraints on spin–lattice relaxation time. of FID decays exponentially with time constant T2 . The
The echo-planar imaging method is one of the faster imag- decayed spin density f (x, y, z) can be written
ing techniques; imaging time can be reduced to as short
as 50 ms. f (x, y, z) = f (x, y, z) exp[−2Ts /T2 (x, y, z)]. (62)
In this method, the FID data in the spatial frequency do-
As explained earlier, the image is now weighted by T2 as
main is acquired following one simple excitation through
well as by T1 . By setting the appropriate Ts values, images
the application of the oscillating gradient during the signal
reception. The image can be reconstructed through a sim-
ple 1-D Fourier transform of the obtained FID data. This
method is limited by gradient driving power and speed.
Nevertheless, the echo-planar technique itself or varia-
tions of the technique seem to be potential candidates for
future fast imaging techniques in NMR CT.
874 Tomography
weighted mainly by T2 can be obtained, provided that the coding resulting from the time-varying gradient can be di-
repetition time is sufficiently large. vided into two terms: the spatially coded term φs and the
velocity-coded term φv , respectively. The sum appears as
b. Parameter imaging methods. The capability of
extracting many functional parameters is one of the most φ = φ s + φv , (63)
important advantages of NMR CT. Flow velocity, T1 , T2 ,
where
and chemical shift are some of the interesting parameters
in NMR imaging that are discussed in this section. φs = γ G x (t)x dt
i. T1 (Spin–lattice relaxation time) and T2 (Spin–spin
relaxation time). The effects of T1 and T2 are closely
and
related to the NMR imaging modes. In T1 imaging, both
the saturation recovery and inversion recovery modes can φv = γ G x (t)vt dt.
be used. By varying the recovery time and observing the
resulting image intensity variation, one can deduce T1 val-
In Eq. (63), G x (t), x, and v represent the time-
ues. Similarly, by changing the echo time, that is, varying
dependent x gradient, the x coordinate of spins at t = 0,
2Ts in Eq. (62) for the spin-echo method, one can obtain
and the x-directional flow velocity of the moving spins,
several images differently weighted by T2 . From the im-
respectively. In the flow measurement, the flow coding
ages obtained with different echo times, T2 values of each
gradient is applied in addition to the conventional RF and
pixel can be calculated.
gradient pulse sequences, so that the phases on the final
ii. Flow imaging. In NMR CT, one can also measure
image are changed only as a result of flow velocity. Flow
the flow or moving velocity of nuclear spins through ob-
velocity can be determined from the calculated phase,
servation of the FID signal. In the first attempt at flow ve-
which is coded according to the velocity of the spins. The
locity measurement two RF coils were used—one for the
unique advantage of this kind of flow velocity imaging
excitation of spins and the other for reception. In this ex-
method is the capability of multidirection flow imaging
periment, surface RF coils were used to excite and receive
by simply applying the additional flow coding gradient in
the signal at known locations. If the maximum signal is re-
the desired direction. Figure 23 shows a typical gradient
ceived at t seconds after the excitation with the distance
waveform for flow phase coding in the x direction, in
l between two RF coils, the velocity can be estimated
which the spatially coded phase term is canceled so that
by l/t.
φs = 0, while the velocity-coded phase term remains
Several flow imaging methods have been developed.
φv = 0. This technique, therefore, allows us to measure
Among these, two techniques relevant to general flow
velocity by simply measuring the phase, which is now
measurement will be discussed: one using density infor-
purely dependent on velocity.
mation and another using phase information.
iii. Chemical-shift imaging. Another important
The RF and gradient pulse scheme of flow imaging
aspect of NMR CT is its spectroscopic imaging capa-
using the selective saturation method uses intensity in-
bility. Before NMR CT was proposed, NMR had been
formation. In this scheme, the first 90◦ RF pulse and
the homogeneity-spoiling gradients are used to saturate
the spins in the selected slice for flow imaging. The 2-D
Fourier imaging sequence for the same slice follows after
t seconds to measure the signals originating from spins
that flowed in from outside the slice, where spins were not
saturated. From the density change observed for several
different t’s, the flow velocity in the selection gradient
direction can be determined as z/T , where T is the
minimum t with the maximum spin density and z is
the slice thickness.
In another variation of flow imaging, phase information
is used to measure flow velocity. Since the pixel values of
an image are usually extracted by taking the real part or ab-
solute values of the image data in complex form, it is possi-
FIGURE 23 Gradient waveform for phase coding of flow veloc-
ble to use the phase information associated with each pixel ity measurement. Note that the gradient pulsing shown effec-
data. Let us assume that a time-varying gradient G x (t) is tively cancels out the spatial coding. The remaining velocity-coded
applied to moving spins after RF excitation. The phase phase φv can be written φv = γ tp2 Gv, where v is the flow speed.
P1: GSS Final Pages
Encyclopedia of Physical Science and Technology EN016F-965 August 1, 2001 11:1
Tomography 875
876 Tomography
A block diagram of a typical NMR CT system is technique enables us to obtain high-quality images within
illustrated in Fig. 26, which depicts the operation of a few seconds, compared to the conventional spin-echo
each part. The main computer generates RF and gradient technique, which usually requires an average of minutes
waveforms and reconstructs images after data acquisition. or so for the imaging of a slice.
RF and gradient pulse waveforms generated in the main A typical pulse sequence for the fast gradient echo tech-
computer are transferred to a microcomputer and then nique known as SSFP is illustrated in Fig. 27. The unique
to a waveform synthesizer, where data in digital form feature of the method, in addition to the speed advantage, is
are converted to analog form. The gradient is applied the potential of obtaining two characteristically different
to x-, y-, and z-gradient coils after being amplified in image data simultaneously, namely FID and echo. Here
the gradient power amplifier. The RF waveform from the FID image refers to the T1 weighted image while echo
the waveform synthesizer is modulated with the RF image refers to the T2 weighted image. Although the gra-
(reference) signal in the modulator, amplified through the dient echos and SSFP techniques generally suffer from
power amplifier, and transferred to the RF coil via the
coupler. The coupler circuit effectively switches on and
off between the transmitting and receiving operations.
The transmitted RF pulse excites nuclear spins in the
sample. The nuclear signal induced on the RF coil by
precessing spins is transferred to the receiver amplifier
through the coupler. The amplified nuclear signal is
demodulated with the RF reference signal and sent to the
data acquisition part. Acquired nuclear signals (FIDs or
echo signals) are transferred to the main computer via the
microcomputer and are used for the reconstruction of the
image. NMR CT systems often employ array processors
for rapid image reconstruction. After reconstruction, the
images are displayed on a cathode ray tube.
Tomography 877
the susceptibility artifact, the methods nevertheless are Cho, Z. H. (ed.) (1976). “Special Issue on Advances in Picture Re-
becoming more widely used because of the advantages of construction Theory and Applications,” Comput. Biol. Med. Vol. 6,
speed and good image quality. No. 4. Pergamon Press, Oxford.
Cho, Z. H., and Nalcioglu, O. (eds.) (1984). “Special Issue on Physics
and Engineering in Nuclear Magnetic Resonance Imaging,” IEEE
Trans. Nucl. Sci. NS-31, No. 4. Inst. Electr. Electron. Eng., New York.
SEE ALSO THE FOLLOWING ARTICLES Gardner, R. J. (1995). “Geometric Tomography,” Cambridge Univ.
Press, Cambridge, UK.
IMAGE-GUIDED SURGERY • IMAGE PROCESSING • MAG- Herman, G. T. (1980). “Image Reconstruction from Projection,” Aca-
demic Press, New York.
NETIC RESONANCE IN MEDICINE • NONDESTRUCTIVE
Herman, G., and Kuba, A. (1999). “Discrete Tomography: Founda-
TESTING • NUCLEAR MAGNETIC RESONANCE (NMR) tions, Algorithms, and Applications,” Birkhauser Boston, Cambridge,
• RADIONUCLIDE IMAGING TECHNIQUES, CLINICAL • Massachusetts.
X-RAY ANALYSIS Kimmich, R. (1997). “NMR: Tomography, Diffusometry, Relaxometry,”
Springer-Verlag, Berlin/New York.
Macovski, A. (1983). “Medical Imaging Systems,” Prentice-Hall,
Englewood Cliffs, New Jersey.
BIBLIOGRAPHY Mansfield, P., and Morris, P. G. (1982). “NMR Imaging in Biomedicine,”
Academic Press, New York.
Barrett, H. H., and Swindell, W. (1981). “Radiological Imaging: The Nalcioglu, O., and Cho, Z. H. (eds.) (1984). “Selected Topics in Image
Theory of Image Formation, Detection, and Processing,” Academic Science,” Springer-Verlag, Berlin/New York.
Press, New York. Newton, T. H., and Potts, D. G. (eds.) (1981). “Radiology of the Skull and
Bushong, S. (2000). “Essentials of Medical Imaging: Computed Tomog- Brain: Technical Aspects of Computed Tomography,” Vol. 5. Mosby,
raphy,” McGraw-Hill Professional, New York. St. Louis, Missouri.
Carson, R. E. ( ). “Quantitative Functional Brain Imaging with Partain, C. L., James, A. E., Rollo, F. D., and Price, R. R. (1983). “Nuclear
Positron Emission Tomography,” Academic Press, San Diego. Magnetic Resonance (NMR) Imaging,” Saunders, Philadelphia.
Cho, Z. H. (ed.) (1974). “Special Issue on Physical and Mathematical Yeon, K. M., Li, G.-Z., and Wol, C. (1995). “Computed Tomogra-
Aspects of 3-D Image Reconstruction,” IEEE Trans. Nucl. Sci. NS-21, phy: State of the Art and Future Applications,” Springer-Verlag,
No. 2. Inst. Electr. Electron. Eng., New York. Berlin/New York.
P1: GPJ/GJK P2: GSS Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN017A-797 August 2, 2001 17:5
Ultrafast Spectroscopy
and its Applications
M. Hayashi S. H. Lin
Y. M. Chang Academia Sinica, Taiwan
T. K. Wang
National Taiwan University Joseph L. Knee
Wesleyan University
217
P1: GPJ/GJK P2: GSS Final Pages
Encyclopedia of Physical Science and Technology EN017A-797 August 2, 2001 17:5
vibrational modes, which is responsible for vibrational extraordinary features in ultrafast spectroscopy is not only
energy relaxation and dephasing. its ability to measure the rate constants of ultrafast pro-
cesses, but also its ability to generate and observe coher-
ence in various materials.
THE PROGRESS of both technology and applications The study of ultrafast elementary dynamics is no longer
in the field of ultrafast processes over the last 20 years limited to model systems but has now been extended to
has broadened the scope of ultrafast sciences. There are systems of great complexity that are of interest to con-
various ways to define the term “ultrafast.” Here “ultra- densed materials including biological systems. One can
fast” means “femtosecond (fsec),” which is a unit of time expect to witness “coherence” of such a wide range of
equal to 10−15 seconds. This timescale becomes accessi- materials. Several promising challenges to even control
ble due to progress in the generation, amplification, and coherence have already been started utilizing controllable
measurement of ultrashort light pulses. With the advent of properties of a femtosecond pulse such as the central fre-
all-solid-state femtosecond laser sources and the extension quency, the width, and the phase together with an itera-
of laser wavelengths via frequency conversion techniques, tive algorithm method. These challenges will lead to ex-
one can now obtain a variety of high-performance sources tensions and new directions in femtosecond light-matter
for extremely short light pulses. control based on the temporal coherence of light and its
Ultrafast laser pulse spectroscopy is a field of study interference with matter waves.
wherein the unique properties of modern pulsed lasers are
used to investigate the ultrafast dynamics of fundamental
I. ULTRAFAST SPECTROSCOPY
processes in physics, chemistry, and biology. The pro-
TECHNIQUES
cesses being investigated by femtosecond spectroscopic
techniques can be categorized into three broad groups:
A. Pump-Probe Techniques
(1) femtochemistry in gas and liquid phases; (2) electron-
hole-phonon interaction in semiconductors and nanostruc- Development of ultrafast lasers has been an area of con-
tures; and (3) ultrafast electron and energy transfer reac- tinued intense activity in last three decades. Several tech-
tions in biological systems. Figure 1 exhibits approximate niques have been developed to generate ultrashort pulses.
timescales that are relevant to physical, chemical, and bio- So far the availability of ultrafast lasers with a wide range
logical changes in ultrafast time region. Clearly, a number of pulse widths, wavelengths, pulse energy, and pulse rep-
of important fundamental mechanisms in chemical, bio- etition rate makes it possible not only to investigate a broad
logical, and semiconductor systems can be investigated range of physical, chemical, and biological phenomena us-
with ultrafast laser spectroscopy, because they take place ing ultrafast optical spectroscopy, but also to investigate
in the femto- and picosecond timescales. One of the most practical applications of ultrafast technology.
FIGURE 1 Typical timescales relevant to physical, chemical, and biological changes. The extension of the presented
time range is not intended to imply upper or lower limits on some of the processes.
P1: GPJ/GJK P2: GSS Final Pages
Encyclopedia of Physical Science and Technology EN017A-797 August 2, 2001 17:5
the kinetics of solution-phase charge transfer reactions. ening caused by the ultrafast dephasing processes in the
Subsequent to a perturbation of the charge distribution of a condensed phases.
solute, solvent molecules reorganize so as to lower the free In two-dimensional Raman spectroscopy, it is the fifth-
energy of solvation. The initial stage of solvation dynamics order nonlinear optical response responsible for the ob-
(<300 fsec) is the inertial motion of the solvent molecules served signal. Between the preparation of the initial vi-
and represents the relaxation of well over half of the solva- brational coherence and the probing step in traditional
tion energy. The friction effects are manifested by the sub- one-dimensional third-order Raman spectroscopy, there
sequent diffusive motion of solvent molecules. The direct is a second interaction with an additional pair of light
experimental observation of these solvation processes is fields that transfers the initial vibrational coherence to a
based on time resolving the fluorescence Stokes shift, that second vibrational coherence. The second coherence is
is, the shift in fluorescence frequency to longer wavelength then probed in the final step. Thus, it has two controllable
as polar solvent molecules reorganize to accommodate the time periods, and the signal S(τ2 , τ4 ) is governed by the
newly created excited-state charge distribution. The time- fifth-order response function. The 2-D signal can be dis-
resolved fluorescence spectrum in femtosecond timescale played in the frequency domain by defining the Fourier
is obtained by frequency up-conversion spectroscopy. In transform of S(τ2 , τ4 ) with respective to τ2 and τ4 , i.e.,
the experiments, the fluorescence induced by the pump S(ω2 , ω4 ). Like in multidimensional NMR spectroscopy,
pulse is collected and is mixing with the probe pulse in a the cross peaks in the 2-D Raman spectrum should provide
nonlinear crystal. The sum-frequency spectrum produced a direct probe for intermode couplings, which can be used
by the fluorescence and probe photons is monitored as as a sensitive probe to local structural information. Future
a function of the pump-probe time delay. With this tech- theoretical investigation will be focused on connecting the
nique, the femtosecond time resolution of the fluorescence detailed shape of this 2-D Raman spectrum to vibrational
spectrum can then be achieved. coupling dynamics.
The fluorescence Stokes shift function S(t), for ex-
ample a coumarin dye in water, obtained by experiment
can then be obtained. Here, S(t) is a normalized func- III. BIOLOGICAL SYSTEMS
tion describing the process toward equilibrium: S(t) =
[v(t) − v(∞)]/[v(0) − v(∞)], where v(t) is a characteris- A. Ultrafast Biological Dynamics
tic frequency of the fluorescence spectrum, for example,
Quite recently, important contributions have been made in
its first moment, at time t. The initial solvent dynamics
the study of the elementary steps in vision, photosynthe-
is responsible more than 50% of relaxation and is com-
sis, protein dynamics, and electron and proton transport in
pleted less than 50 fsec. It results from free, inertial, small-
DMA. An ultrafast event, bond cleavage, or electron trans-
amplitude motions of the solvent molecules within the
fer also occurs in proteins such as those of photosynthetic
potential wells that they initially occupy. Following the
reaction centers and antennas, hemoglobins, cytochromes,
shape break in the response, the solvent begins to restruc-
and rhodopsin. One of the most remarkable achievements
ture itself, and energy flows irreversibly into the solvent
in the studies of ultrafast biological events has been due
and the behavior is now dissipative. This lack of dissipa-
to the fact that the existence of global and coherent nu-
tion during the initial solvation epoch is of significance
clear motions has been revealed in these complex systems.
for chemical processes in which coherent contributions
Quantum beats can be observed in these protein systems
can be important. Examples are electron or energy trans-
using femtosecond spectroscopy. This impact has changed
fer processes mediated by bridge states.
the way of looking at the complexity of these systems and
opened the door that leads to the bold challenge to un-
D. Two-Dimensional Spectroscopy derstand what makes biological systems different from
nonbiological systems.
Multidimensional nuclear magnetic resonance (NMR)
techniques, since their inception in the 1970s, have turned
into a powerful tool that provides structures and dynam- B. Photosynthesis
ics of very complex molecules, crystals, and proteins. By
1. Electron Transfer in Photosynthetic
extending these ideas to the optical regime, it is possi-
Reaction Centers
ble to come up with novel classes of spectroscopies that
can probe complex vibrational motions. Its realization, The study of the primary process of photosynthesis has
however, has to wait for the recent breakthrough in fem- been performed using transient absorption techniques. In
tosecond laser techniques, which has provided necessary this case, the reaction is monitored by measuring the
impulsive excitation to overcome tremendous line broad- absorption of the probe by the reactant, the product, or
P1: GPJ/GJK P2: GSS Final Pages
Encyclopedia of Physical Science and Technology EN017A-797 August 2, 2001 17:5
perhaps an intermediate that exists only transiently as the Cyanobacteria and red algae utilize antenna pigments
reaction proceeds. The most thoroughly investigated of called phycobilins packed into complexes called phyco-
bacterial photosynthetic systems have been those bacterial bilisomes, which are attached to the photosynthetic mem-
photosynthetic systems such as the purple nonsulfur bacte- branes. Phycobilisomes contain several hundred billion
ria Rhodopseudomonas and Rhodobacter. A major break- chromophores, linear tetrapyrroles attached to the protein.
through for the bacterial reaction center occurred about These proteins are organized into disks that are them-
ten years ago when its structure was determined at high selves stacked into rods, with disks containing shorter
resolution. The bacterial reaction center of Rhodobacter wavelength pigments on one end, and longer wavelength
sphaeroides, for example, is a pigment-protein complex pigments at the other end next to a central core. Thus
designed to convert optical excitation into the initial elec- the shorter wavelength absorbers: phycoerythrins (PE),
tron transfer event in photosynthesis. The reaction center 570 nm; are on the outside, phycocyanins (PC), 630 nm;
contains six chlorophyll-like pigments arranged with ap- within them, allophycocyanins (APC), 650 nm; in the
proximate C2 symmetry. The relative arrangement of these core, followed by chlorophyll a inside the photosynthetic
pigments within the protein matrix is well-known as the membrane.
special pair—two accessory bacteriochlorophylls and two The X-ray crystal structure of the trimeric aggregation
bacteriopheophytins. state of APC isolated from the cyanobacterium Spirulina
It has been shown that the initial process is an elec- platensis is well-characterized: the APC trimer can be
tron transfer to form a charge-separated special pair. This described as a C3 -symmetric, ring-like homotrimer of α
charge separation has been measured to occur less than 100 and β polypeptide monomers. The phycocyanobilin chro-
fsec and it has also been shown that subsequent electron mophore (PCB) is arranged as dimers formed across the
transfer reactions from the charge-separated special pair polypeptide interfaces between adjacent α, β polypeptide
to one of the two bacteriopheophytins take place within monomers.
about 1–4 psec depending on temperature. The results of The absorption spectra of the homogenous monomers
femtosecond experiments together with the bacterial re- for APC are almost identical to that of the α-subunit, and
action center structural determination raise the challenge it has an absorption maximum at 614 nm while trimers
of the relating mechanism of the primary electron transfer have a sharp maximum at 650 nm and a prominent shoul-
to the known structures. This fascinating problem is still der at about 610–620 nm. The origin of electronic states
under intensive investigation. involved in steady-state spectroscopy is needed for under-
Coherent nuclear motions of bacterial photosynthetic standing of femtosecond time-resolved spectra, especially
reaction center have been visualized observing pump- when one relates the mechanism of the kinetics appear-
probe-stimulated emission with 80-fsec pulse lasers. ing in the time-resolved spectra to nonradiative transitions
Recent intensive investigations using pump-probe tech- among the electronic states. If electronic Hamiltonian is
niques and femtosecond anisotropy measurements have fully diagonalized, in other words, delocalized electronic
revealed that ultrafast energy transfer (∼80 fsec) from wavefunction is used to describe the system, nonradia-
one of the two accessory bacteriochlorophylls to the spe- tive transitions between the same spin multiplicity should
cial pair and internal conversion (∼120 fsec) within the be due to the break-down of the adiabatic approxima-
special pair occur prior to electron transfer from the spe- tion, i.e., internal conversion (IC) process. In this case,
cial pair to one of the two bacteriopheophytins. to estimate the electronic coupling constant, the Förster
or Marcus type of rate constant cannot be used. Vibra-
tional properties are also needed to estimate the electronic
2. Energy Transfer in Antenna
coupling constant. Vibrational frequencies involved in op-
The role of the antenna pigments is to collect light energy tical transition of APC trimer can be obtained by using a
from the sun and transfer it to reaction centers. In green pump-probe technique with 20-fsec laser pulses. Figure 5
plants and algae, these pigments are found in the thylakoid shows femtosecond pump-probe signals of APC trimer at
membranes of chloroplasts. In photosynthetic bacteria, the room temperature as a function of the probing wavenum-
pigments are found in intracytoplasmic membranes or in ber. Each signal clearly shows oscillatory features called
special vesicles in the cell. The principle antenna pigments “quantum beats” accompanied by with rising or decaying
are chlorophyll a and b in plants, chlorophyll c in some components. Insets of panel A exhibit oscillatory com-
algae, and bacteriochlorophyll a, b, or c in bacteria. Other ponents extracted from the observed signals. Performing
pigments, called accessory pigments, are carotenoids and Fourier transform analysis to these oscillatory components
phycobiliproteins. Following light absorption by a pig- provides possible vibrational frequencies involved in the
ment molecule, the electronic excitation is transferred un- optical transition. This information can be used to estimate
til it is trapped by a reaction center. the electronic coupling constant.
P1: GPJ/GJK P2: GSS Final Pages
Encyclopedia of Physical Science and Technology EN017A-797 August 2, 2001 17:5
FIGURE 5 (a) Quantum beats observed in pump-probe signals of APC trimer at room temperature. Signals were
obtained as a function of delay time between the pump and probe pulses as well as the probe wavelength. Insets
are oscillatory components extracted from the observed signals. (b) Fourier transform analysis of the oscillatory
components.
Femtosecond anisotropy measurements can provide electronically excited state via the 620 nm pump-pulse un-
useful information on the origin of rising and decay com- dergoes nonradiative transition to one of the lower elec-
ponents appearing in the observed signals. Anisotropy can tronic states within 350 fsec. The origin and nature of the
be obtained by monitoring the spectrally resolved probe electronic states of the APC trimer are still under extensive
transmission as a function of temporal delay and polariza- investigation.
tion relative to the pump. The anisotropy can be calculated
from the pump-probe signals as
S (τ ) − S⊥ (τ ) IV. SEMICONDUCTORS
r (τ ) = ,
S (τ ) + 2S⊥ (τ )
A. Ultrafast Dynamics
where S (τ ) and S⊥ (τ ) are pump-probe transients obtained
with parallel and perpendicular pump and probe polariza- The scattering and relaxation processes of photoexcited
tions, respectively. Figure 6 shows obtained femtosecond carriers in semiconductors and their nanostructures, par-
anisotropy signals of APC trimer excited at 620 nm and ticularly the III–V group semiconductor and its quantum
probed at 660 and 620 nm. The obtained anisotropy indi- well structures, have been extensively investigated using
cates that there exist at least two electronic states between the ultrashort spectroscopy techniques described above.
660 and 620 nm. Fitting the anisotropy decay signal of Intrinsic semiconductors are characterized by their band
probing at 660 nm to a single exponential decay function structure, which contains a bandgap separating the occu-
yields approximately 350 fsec. This result implies that the pied valence band and the empty conduction band. The
P1: GPJ/GJK P2: GSS Final Pages
Encyclopedia of Physical Science and Technology EN017A-797 August 2, 2001 17:5
from the fact that carriers are confined along the growth
direction of the structure, but are free to move in lateral di-
rections. The dimensionality of the semiconductor nanos-
tructures can be further reduced. The one and zero dimen-
sional systems, so-called quantum wire and quantum dot,
have attracted a lot of attention recently and the area of this
research can be expected to broaden and yield fascinating
results that cannot be presently anticipated.
holes reach thermal equilibrium faster than the electrons. phonons can also be important in the thermalization pro-
The long-range Coulomb interaction diverges in the ab- cess of hot carriers. However, if the carrier has sufficient
sence of screening. Screening is generally treated as the energy to emit an optical phonon, then the optical phonon
plasmon-phonon coupling and occurs in the nonthermal scattering rate is generally considerably higher than for the
regime. acoustic phonons. In brief summary, carrier-carrier scat-
Interaction of carriers with phonons plays a major role tering is primarily responsible for redistributing the energy
in the exchange of energy and momentum between car- and leads to a thermalized distribution function of the car-
riers and the lattice, and hence determines the relaxation rier system. The characterized temperature of the carrier
of photoexcited carriers and their transport properties in system can be higher than the lattice temperature, and
semiconductors. Polar optical phonons, particularly, play may be different for electron and hole systems. Typically,
an important role in carrier-phonon scattering processes the electrons and holes thermalize among themselves in
in group III–V semiconductors. The interaction between hundreds of femtoseconds, while the electrons and holes
polar optical phonons and low energy carriers is described achieve a common temperature in a couple of picoseconds.
by the Fröhlich interaction, in which the polar coupling The carrier system finally reaches lattice temperatures in
strength varies inversely with the wavevector q of the hundreds of picoseconds through the interaction with var-
phonon being emitted or absorbed. Since the Fröhlich in- ious phonons in the semiconductor.
teraction rate is expected to vary m 1/2 , the holes might be In the isothermal regime, all the carriers and phonons
expected to interact more strongly with the LO phonons are in thermal equilibrium with each other, i.e., they can
than the electrons, as indicated in Fig. 7. Note that the be described by the same lattice temperature. However,
high-energy electrons and holes in subsidiary valleys can there is still an excess of electrons and holes compared to
also interact with optical phonons through nonpolar op- the thermodynamic equilibrium. These excess electrons
tical deformation potential scattering. This interaction is and holes must return to the thermodynamic equilibrium
also responsible for the intervalley scattering of electrons through either radiative recombination or nonradiative
in the nonthermal regime. Interaction with both LA and TA recombination.
FIGURE 8 (a) Time-resolved second-harmonic generation spectroscopy reveals the generation of coherent phonon
modes in native oxide covered GaAs (100) crystal. (b) Left inset shows the oscillatory part of the time domain data.
(c) Right inset shows the Fourier power spectrum of the oscillatory part.
P1: GPJ/GJK P2: GSS Final Pages
Encyclopedia of Physical Science and Technology EN017A-797 August 2, 2001 17:5
X-Ray Analysis
Ron Jenkins
International Centre for Diffraction Rate
887
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
lines, in which a single diffracting crystal is used to of the diffraction of X-rays by Max von Laue in 1913,
separate the wavelengths from the sample. the use of this method for materials analysis has become
X-Ray powder diffraction Technique for phase identi- very important both in industry and research, to the ex-
fication based on measurement of structure sensitive tent that, today, it is one of the most useful techniques
diffracted X-ray lines. available for the study of structure-dependent properties of
materials.
X-Ray fluorescence spectrometry also has its begin-
X-RAY ANALYSIS is the identification and characteri- nings back in the early part of the 20th century with the
zation of materials through the use of X rays and includes work of Moseley and others. The technique uses either
the two techniques of spectrometry and diffractometry. the diffracting power of a single crystal to isolate nar-
X-Ray fluorescence spectrometry is a means of qualita- row wavelength bands, or a proportional detector to iso-
tively and quantitatively determining elements by mea- late narrow energy bands, from the polychromatic beam
surement of the wavelengths and intensities of character- characteristic radiation excited in the sample. The first
istic emission wavelengths. The technique is applicable of these methods is called wavelength-dispersive spectro-
to all but the elements of very low atomic number, with metry, and the second, energy-dispersive spectrometry.
sensitivities down to the low part per million level. X-Ray Because of the known relationship between emission
powder diffractometry uses diffraction data obtained by wavelength and atomic number, isolation of individual
study of structure-dependent scattering phenomena. The characteristic lines allows the unique identification of an
shapes and positions of diffraction peaks can be used to element to be made, and elemental concentrations can be
establish a wide range of physical properties, including estimated from characteristic line intensities. Thus this
stress, texture, crystallinity, and particle size, but the main technique is a means of materials characterization in terms
application is for phase identification. A large data base of chemical composition.
of pure single-phase materials is available so that rapid X-Ray powder diffractometry involves characterization
empirical file-searching techniques can be employed for of materials by use of data that are dependent on the atomic
measuring phases (compounds). arrangement in the crystal lattice. The technique uses sin-
gle or multiphase (i.e., multicomponent) specimens com-
prising a random orientation of small crystallites, each of
I. USE OF X RAYS FOR MATERIALS the order of 1–50 µm in diameter. Each crystallite in turn
CHARACTERIZATION is made up of a regular, ordered array of atoms. An or-
dered arrangement of atoms (the crystal lattice) contains
X rays are a short-wavelength form of electromagnetic planes of high atomic density, which in turn means planes
radiation discovered by Wilhelm Roentgen in 1895. of high electron density. A monochromatic beam of X-ray
X-Ray-based techniques provided important tools for the photons will be scattered by these atomic electrons, and
theoretical physicist in the first half of this century, and if the scattered photons interfere with each other, diffrac-
since the early 1950s they have found an increasing use tion maxima may occur. In general, one diffracted line
in the fields of materials characterization. Today, the an- will occur for each unique set of planes in the lattice. A
alytical techniques based on X-ray diffraction and X-ray diffraction pattern is typically in the form of a graph of
spectrometry, both of which were first conceived almost diffraction angle (or interplanar spacing) versus diffracted
70 years ago, play a vital role in the analysis and study of line intensity. The pattern is thus made up of a series of
inorganic and organic solids. superimposed diffractograms, one for each unique phase
X-Ray photons are produced following the ejection of in the specimen. Each of these unique patterns can act as
an inner orbital electron from an irradiated atom and sub- an empirical “fingerprint” for the identification of the var-
sequent transition of atomic orbital electrons from states ious phases, using pattern-recognition techniques based
of high to low energy. When a monochromatic beam on a file of standard single-phase patterns. Quantitative
of X-ray photons falls onto a given specimen, three ba- phase analysis is also possible, albeit with some difficulty
sic phenomena may result: absorption, scatter, or fluo- because of various experimental and other problems, not
rescence. The coherently scattered photons may undergo the least of which is of the large number of diffraction
subsequent interference, leading in turn to the generation lines occurring from multiphase materials.
of diffraction maxima. The angles at which the diffrac- Although the major thrust of this review is to cover
tion maxima occur can be related to the spacings be- X-ray diffraction and X-ray spectroscopic techniques,
tween planes of atoms in the crystal lattice, and hence these are by no means the only X-ray-based methods that
X-ray-generated diffraction patterns can be used to study are used for materials analysis and characterization. In ad-
the structure of solid materials. Following the discovery dition to the many industrial and medical applications of
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
diagnostic X-ray absorption methods, X rays are also used As long as the vacancy in the K shell exists, the atom is
in areas such as structure determination based on single- in an unstable state and there are two processes by which
crystal techniques, space exploration and research, in the atom can revert back to its original state. The first of
lithography for the production of microelectronic circuits, these involves a rearrangement of the atomic electrons,
and so on. One of the major limitations with the further as illustrated in Fig. 1d. This rearrangement does not in-
application of X-ray methods is the inability to “focus” volve the emission of X-ray wavelengths but does result in
X-rays, as can be done with visible light rays. Although the emission of other photoelectrons from the atom. The
it is possible to partially reflect X-rays at low glancing effect is known as the Auger effect, and the emitted pho-
angles, or to diffract an X-ray beam with a single crys- toelectrons are called Auger electrons. These Auger elec-
tal, these methods cause significant intensity loss and fall trons have little further consequence in this specific review
far short of providing the high-intensity, monochromatic on X-ray fluorescence and diffraction methods, but they
beam that would be ideal for, for example, an X-ray micro- do provide the basis of another important spectroscopic
scope. Use of synchrotron radiation offers the potential of method called electron spectroscopy, which is especially
an intense, highly focused, coherent X-ray beam but has important for surface and bonding studies.
practical limitations due to size and cost. The second process by which the excited atom can re-
The X-ray laser could, in principle, provide an attrac- gain stability is by transference of an electron from one
tive alternative, and since the discovery of the laser in of the outer orbitals to fill the vacancy. This is illustrated
1960, the possibilities of such an X-ray laser have been in Fig. 1c. The energy difference between the initial and
discussed. Although major research efforts have been and final states of the transferred electron may be given off
are still being made to produce laser action in the far ultra- in the form of an X-ray photon. The wavelength of the
violet and soft X-ray regions, production of conditions to X-ray photon is inversely related to the energy of the pho-
stimulate laser action in the X-ray region with a net posi- ton. There are several different combinations of quantum
tive gain is difficult, mainly because of the rapid decay numbers held by the electron in the initial and final states;
rates and high-absorption cross sections that are experi- hence several different X-ray wavelengths will be emit-
enced in practice. Many different experiments have been ted from a given atom. For those vacancies giving rise to
designed to circumvent these problems, and one of the characteristic X-ray photons, a series of very simple se-
more promising new approaches in soft X-ray laser de- lection rules can be used to define which electrons can be
velopment is based on rapid plasma cooling by radiation transferred: The principal quantum number n must change
losses after the laser pulse. This leads to a fast recombi- by 1, the angular quantum number l must change by 1, and
nation and collisional cascade into upper excited levels, the vector sum of l + s must be a positive number chang-
while the lower excited levels are rapidly depopulated by ing by 1 or 0. In effect, this means that for the K series
radiative transitions. This causes a population inversion only p → s transitions are allowed, yielding two lines
leading to an overall gain. A CO2 laser is focused onto for each principal level change. Vacancies in the L level
a target of Teffon and the resulting plasma confined to a follow similar rules and give rise to L series lines. There
field of 50–90 kG. are more of the L lines since p → s, s → p, and d → p
transitions are all allowed within the selection rules.
In general, electron transitions to the K shell give be-
II. ORIGIN AND PROPERTIES OF X RAYS tween two and six K lines, and transitions to the L shell
give about 12 strong to moderately strong L lines. As
If a high-energy particle, such as an electron, strikes a an example, Fig. 2 shows the L emission spectrum of
bound atomic electron, and the energy E of the particle is gold (Z = 79). Since all emitted X-ray photons have en-
greater than the binding energy O of the atomic electron, ergies proportional to the differences in the energy states
it is possible that the atomic electron will absorb all of the of atomic electrons, the lines from a given element will
energy from the particle. The primary particle disappears be characteristic of that element. As will be shown in the
in the process, and its energy is transferred to the atomic next section, the ability to use a measured wavelength as
electron, which is then ejected from its atomic position a unique property of an excited element forms the basis of
and leaves the atom with a kinetic energy E pe equivalent X-ray emission spectrometry.
to the difference between that of the initial particle and the All X-ray photons with wavelengths in the range gener-
binding energy of the atomic electron. The ejected electron ally employed for analytical work are absorbed to a lesser
is called a photoelectron and the interaction is referred to or greater extent by solid materials. If a monochromatic
as the photoelectric effect. Figure 1 illustrates the process, beam of X-ray photons of wavelength λ and intensity I0
and here an interaction with an electron in the K shell is falls onto a material of thickness x and density ρ, the frac-
indicated. tion of the beam transmitted I is given by
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
FIGURE 1 Production of characteristic X rays. (a) Excitation of a K-orbital electron producing a K vacancy.
(b) Emission of a photoelectron producing a K vacancy. Deexcitation leads to (c) the emission of a characteristic
K photon, or (d) an Auger electron.
FIGURE 2 The L-series X rays from gold. (a) The energy-level diagram of gold shows all possible diagram lines and
(b) the actual emission spectrum.
as that of the incident beam. It can also happen that the beam of radiation falls onto these atomic layers scatter-
scattered photon gives up a small part of its energy dur- ing will occur. In order to satisfy the requirement for in-
ing the collision, especially where the electron with which terference, it is necessary that the scattered waves orig-
the photon collides is only loosely bound. In this instance inating from the individual atoms, that is, the scattering
the scatter is said to be incoherent (Compton scatter), points, be in phase with one another. The geometric con-
and the wavelength of the incoherently scattered photon ditions for this condition to occur are illustrated in Fig. 3.
will be greater than that of the incident wavelength. Two parallel rays strike a set of crystal planes at an angle
Under certain geometric conditions, wavelengths that θ and are scattered as previously described. Reinforce-
are exactly in phase add to one another, or if exactly out ment will occur when the difference in the path lengths
of phase, cancel each other out. Under such conditions, of the two interfering waves is equal to a whole num-
coherently scattered photons may constructively interfere ber of wavelengths. This path-length difference is equal
with each other, giving diffraction maxima. Since a crystal to CB + BD, and since CB = BD = x, nλ must equal 2x
lattice consists of a regular arrangement of atoms form- for reinforcement to occur, where n is an integer. It will
ing layers of high atomic density, when a monochromatic also be seen that x = d sin θ , where d is the interplanar
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
spectrometers became commercially available in the early pable of providing up to around 3 kW of power at a poten-
1970s, and today there are around 2500 units in use. tial of typically 60–80 kV, plus a sealed X-ray tube. The
The analytical chemist has available a wide range of in- sealed X-ray tube has an anode of chromium, rhodium,
struments for the qualitative and quantitative analysis of tungsten, silver, gold, or molybdenum and delivers an in-
multielement samples and in the choice of technique will tense source of continuous radiation, which then impinges
generally consider such factors as sensitivity, speed, ac- on the analyzed specimen, where characteristic radiation
curacy, cost, and range of applicability. is generated. The actual spectrometer itself consists of a
All conventional X-ray spectrometers comprise three specimen holder support, a primary collimator, an analyz-
basic parts: the primary source unit, the spectrometer it- ing crystal, and a detector. The geometric arrangement of
self, and the measuring electronics. The primary source these components is shown in Fig. 4a, and a line drawing
unit consists of a very stable high-voltage generator, ca- of an actual spectrometer is shown in Fig. 4b. A portion of
FIGURE 4 The wavelength-dispersive spectrometer. (a) The component parts of the wavelength-dispersive spec-
trometer. (b) The same components as they would appear in an actual commercial spectrometer.
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
the characteristic “fluorescence” radiation from the speci- Most modern wavelength-dispersive spectrometers are
men is passed via a collimator or slit onto the surface also controlled in some way by a minicomputer or micro-
of an analyzing crystal, where individual wavelengths are processor and by use of specimen changers are capable
diffracted to the detector in accordance with the Bragg of very high specimen throughput. Once they are set up,
law. A goniometer is used to maintain the required θ –2θ the spectrometers will run virtually unattended for several
relationship between crystal and detector. Typically, six or hours.
so different analyzing crystals are provided in this type of Like the wavelength-dispersive spectrometer, the
spectrometer, giving the operator a wide range of choice of energy-dispersive spectrometer also consist of the three
dispersion conditions. In general, the smaller the d spac- basic units—excitation source, spectrometer, and detec-
ing of the crystal, the better the separation of the lines but tion system. In this case, however, the detector itself acts
the smaller the wavelength range that can be covered. as the dispersion agent. The detector is typically a lithium-
A tandem detector system is typically employed, com- drifted silicon, “Si(Li),” detector, which is a proportional
posed of a gas flow counter and a scintillation counter. detector of high intrinsic resolution. The Si(Li) detector
This is used to convert the diffracted characteristic photons diode serves as a solid-state version of the gas flow detector
into voltage pulses, which are integrated and displayed as in the wavelength-dispersive system, but with a detector
a measure of the characteristic line intensity. The gas flow gain of unity. When an X-ray photon is stopped by the
counter is ideal for the measurement of the longer wave- detector, a cloud of ionization is generated in the form
lengths, and the scintillation counter is best for the short of electron–hole pairs. The number of electron–hole pairs
wavelengths. created—that is, the total electric charge released—is pro-
The output from a wavelength-dispersive spectrometer portional to the energy of the incident X-ray photon. The
may be either analog or digital. For qualitative work an charge is swept from the diode by a high voltage applied
analog output is traditionally used, and in this instance across it. A preamplifier is responsible for collecting this
a rate meter is used to integrate the pulses over short charge on a feedback capacitor to produce a voltage pulse
time intervals, typically of the order of a second or so. proportional to the energy of original X-ray photon. Thus
The output from the rate meter is fed to an x /t recorder, when a range of photon energies is incident upon the de-
which scans at a speed that is synchronously coupled with tector, an equivalent range of voltage pulses is produced as
the goniometer scan speed. The recorder thus displays an a detector output. A multichannel analyzer is used to sort
intensity–time diagram, which becomes an intensity–2θ the arriving pulses to produce a histogram representation
diagram. Tables are then used to interpret the wavelengths. of the X-ray energy spectrum.
For quantitative work it is more convenient to employ dig- The output from an energy-dispersive spectrometer is
ital counting, and a timer–scaler combination is provided generally displayed on a CRT, and Fig. 5 shows such
that will allow pulses to be integrated over a period of sev- an output obtained from a geological sample using a
eral tens of seconds and then displayed as count or count molybdenum-anode X-ray tube. The X-ray tube is some-
rate. times fitted with a thin transmission filter, the function of
The intensities of the characteristic lines can be used to which is to lower the background. In the figure, two spectra
estimate the concentrations of the appropriate elements. were acquired, the first with the molybdenum tube alone
Characteristic line intensities are related to a number of in- and second with the tube fitted with a molybdenum filter.
strumental parameters, such as source intensity, diffracting The counting time in both instances was 400 s and the ef-
efficiency of the analyzing crystal, and detector efficiency. fect of the filter on the background is clearly seen. Fifteen
These factors can be calibrated under a given set of exper- elements have been identified. The operator is able to dis-
imental parameters to provide a measure of the sensitivity play the contents of the various channels dynamically as
of the spectrometer to the element in question. The preci- an energy spectrum. Provision is generally made to al-
sion of an intensity measured with a modern spectrometer low zooming in on portions of the spectrum of special
is of the order of one-tenth of a percent. Before this in- interest, to overlay spectra, to subtract background, and
tensity can be used to calculate the element concentration, so on, in a rather interactive manner. As in the case of the
certain systematic errors from interelement effects must modern wavelength-dispersive systems, nearly all energy-
be corrected for. There are many standard algorithms that dispersive spectrometers will incorporate some form of
can be used for this process, and most automated spec- minicomputer that is available for spectral stripping, peak
trometers will allow the automatic application of these identification, quantitative analysis, and a host of other
correction procedures along with the final calculation of useful functions.
the elemental concentrations. Accuracies of the order of As has been previously mentioned, within the two major
a few tenths of a percent can be obtained where these categories of X-ray spectrometers specified, there is a wide
procedures are correctly applied. diversity of instruments available. The major differences
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
FIGURE 5 Geological sample, 400-s analysis. A multichannel analyzer output from a rock sample, using a 400-s
counting time. Two spectra are compared: dashed line is direct excitation using an Mo-anode tube; dotted line is the
same but using an Mo filter between tube and sample.
generally lie in the type of source used for excitation, tion compared to the wavelength-dispersive systems but
the number of elements that are measurable at one time, find great application in quality control, trouble-shooting
the speed at which they collect data, and finally the price problems, and so on. They have been particularly effec-
range. All of the instruments are, in principle at least, capa- tive in scrap alloy sorting, in forensic science, and in the
ble of measuring all elements in the periodic classification provision of elemental data to supplement X-ray powder
from Z = 9 (F) and upward. Most can be fitted with mul- diffraction data.
tisample handling facilities, and all can be automated by Like all instrumental methods of analysis, the high pre-
use of minicomputers. All are capable of precisions of the cision can be translated into high accuracy only if the
order of a few tenths of a percent, and all have sensitivities various systematic errors in the analysis process are taken
down to the low parts per million (ppm) level. As far as the care of. The precision of a well-designed X-ray spectrom-
analyst is concerned, they differ only in their speed, cost, eter is typically of the order of ∼0.01%; the major source
and number of elements measurable at the same time. of this random error is the X-ray source, that is, the high-
voltage generator plus the X-ray tube. In addition, there is
an error arising from the statistics of the actual counting
V. APPLICATION OF X-RAY
process. The production of X rays is a random process.
SPECTROMETRIC METHODS
Thus, in any experiment a number of counts N are taken
by counting a certain count rate R for a period of time
The great flexibility and range of the various types of
t, and there will be a random error σ associated
√ with a
X-ray fluorescence spectrometer, coupled with their high
measured value of N , this being equal to N :
sensitivity and good inherent precision, make them ideal
for quantitative analysis. Single-channel wavelength- √ √
σ = N = R×t (4)
dispersive spectrometers are typically employed for both
routine and nonroutine analysis of a wide range of prod- Since the measured value of N is time-dependent, the
ucts, including ferrous and nonferrous alloys, oils, slags counting error can be reduced to the level of other random
and sinters, ores and minerals, thin films, and so on. These errors (e.g., the source error) simply by taking a suitable
systems are very flexible but relative to the multichan- value of t. Equation (4) can be extended for the case where
nel spectrometers are somewhat slow. The multichan- background is significant or where multiple measurements
nel wavelength-dispersive instruments are used solely for are employed.
routine, high-throughput analyses where the great need Systematic errors in quantitative X-ray spectrometry
is for fast, accurate analysis, but where flexibility is of arise mainly from absorption- and specimen-related phe-
no importance. Energy-dispersive spectrometers have the nomena, much as in X-ray powder diffraction, except that
great advantage of being able to display information on all in the spectrometry case they are much more complicated.
elements at the same time. They lack somewhat in resolu- This is because a single characteristic wavelength is used
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
in the diffraction experiment, whereas many wavelengths the sensitivity varies by about three orders of magnitude
are involved in spectroscopy. This means, for example, over the normal range of 0.2 to 20 Å. The lower limit of
that the equivalent effect expression given in Eq. (1) car- detection for a given element is related to the sensitivity
ries an integral sign. Although these so-called matrix ef- for the element in question, relative to the background
fects are somewhat complicated, many excellent methods intensity at the measured wavelength. The lower limit of
have been developed for handling them. The advent of the detection (LLD) is expressed as the concentration of the
minicomputer-controlled spectrometer has done much to element equivalent to two standard deviations of the back-
enhance the application of these correction procedures, ground counting rate Rb . It is generally assumed that in the
and today in most cases one is able to quantify most ele- low concentration region the counting error is the major
ments in the periodic table of atomic number 9 (F) and source of random error, and the equation for LLD is re-
upward to an accuracy of a few tenths of a percent. The lated to the background counting time Tb in the following
areas of application of the X-ray fluorescence technique manner:
now cover almost all areas of inorganic analysis.
LLD = (3/m) Rb /Tb (5)
The background counting time Tb is equal to half of the
VI. TRACE ANALYSIS BY X-RAY total counting time T . The factor 3 in the equation arises
FLUORESCENCE SPECTROMETRY from the fact that two measurements must be made, that
is, peak and underground,
√ which increases the two stan-
The X-ray fluorescence (XRF) method is particularly ap- dard deviations by 2. As an example, the magnesium
plicable to the qualitative and quantitative analysis of low Ka line measured in an aluminum alloy gives a sensitiv-
concentrations of elements in a wide range of samples, as ity m of about 1000 counts/s percent, over a background
well as allowing the analysis of elements at higher con- Rb of about 30 counts/s. In a total counting time of 100 s
centrations in limited quantities of materials. The sensi- (Tb = 50 s) the LLD would equal (3/1000) × (30/50)1/2
tivity m of the XRF method is expressed in terms of the or 0.0023%.
intensity of the measured wavelength per unit concentra- It will also be noted from the figure that not only does the
tion, expressed in counts per second per percent. Figure 6 sensitivity of the spectrometer vary significantly over the
shows the approximate counting rate that would be ob- wavelength range of the spectrometer, but so too does the
tained from 1% of the indicated element and shows that background counting rate. The background curve shown
was obtained using a chromium-anode X-ray tube oper-
ated at 50 kV, using a sample of very low atomic number,
that is, of high scattering power. It will be seen that the
background varies by about two orders of magnitude over
the range of the spectrometer. By inspection of Eq. (5), it
will be seen that the detection limit will be best when the
sensitivity is high and the background is low. A combina-
tion of these two factors leads to the detection-limit curve
shown in the figure. The curve is approximate, since both
the spectrometer sensitivity and the measured background
vary with the average atomic number of the sample. Nev-
ertheless, detection limits over most of the atomic number
range lie in the low part per million range. The sensitiv-
ity of the X-ray spectrometer falls off quite dramatically
toward the long-wavelength limit of the spectrometer, due
mainly to low fluorescence yields and the increased influ-
ence of absorption. As a result, poorer detection limits are
found at the long-wavelength extreme of the spectrometer,
which corresponds to the lower atomic numbers. Thus the
detection limits for elements such as fluorine and sodium
are at the levels of hundredths of a percent rather than parts
FIGURE 6 Sensitivity, background, and lower limit of detection
per million.
curves. Plots of spectrometer sensitivity in counts per second
per percent, background counting rates in counts per second, Even though the long-wavelength limit of the spectrom-
and lower limits of detection in parts per million, as a function eter is generally taken as about 20 Å, it is possible to ex-
of wavelength. tend this limit to about 50 Å by use of special techniques.
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
This allows the measurement of two more elements, car- men does not actually have to be a powder. The diffrac-
bon (Z = 6) and oxygen (Z = 7). The detection limits for tion method is equally applicable to bulk metals, thin films,
these elements are, however, very poor and are typically polymer sheets, and a variety of other forms. In addition to
of the order of 3–5%. the use of the method for qualitative and quantitative phase
The irradiation area in a typical X-ray spectrometer is identification, it can also be used to determine the degree
of the order of 5 cm2 and the penetration depth of an av- of crystallinity, that is, the dimensions of the crystalline
erage wavelength is about 20 µm. This means that even regions in otherwise amorphous substances. Where the
though 20 g of sample may be placed in the spectrometer, crystal lattice is distorted by strain, this will also manifest
the analyzed volume is still only of the order of 50 mg. itself in modifications to the diffraction pattern. Where the
The smallest sample that will give a measurable signal strain is isotropic, the lines in the diffraction pattern will be
above background is at leas three orders of magnitude displaced; where the strain is anisotropic, the lines will be
less than this mass: ∼0.05 mg, provided that the sample broadened. Study of selected line positions and line shapes
is spread over the full irradiation area of the spectrome- can reveal much information about stress/strain conditions
ter sample cup. Where this is impracticable, the smallest within a given polycrystalline material such as a quenched
value that can be analyzed is increased by a factor corre- steel. Study of diffraction intensity distributions occuring
sponding to the ratio of the area of the beam (∼5 cm2 ) over the whole surface of a specimen can reveal infor-
to the actual area of the sample. In practical terms, this mation about the “texture” of a material. This technique
sensitivity is quite sufficient for the sample masses typi- is referred to as a “pole figure” study. Similar techniques
cally encountered in the analysis of trace metals in air and can be used to determine topographic information from the
water samples, and XRF methods find increasing appli- surface of single crystals. Diffraction patterns can also be
cation in this area. A second important area of materials recorded over a wide range of temperatures from liquid
analysis involving small samples involves the investiga- helium up to 2000◦ C. These low- and high-temperature
tion of thin films, and, although the technique is limited to experiments are invaluable for the study of phase trans-
rather large areas (typically a few square millimeters), it formations. However, by far the greatest single use of the
does provide useful information about bulk composition powder technique is for phase identification of inorganic,
of surface films. mineral, and organic solids.
All substances are built up of individual atoms, and nearly The instrumentation that is used for powder diffraction
all substances have some degree of order of periodicity in measurements has not changed much from the instruments
the arrangement of these atoms. A crystal can be defined developed in the late 1940s. The major difference found in
as a homogeneous, anisotropic body having the natural modern instrumentation is the use of the minicomputer for
shape of a polyhedron. In practical terms, whether a sub- control, data acquisition, and data processing. Figure 7a
stance is homogeneous or not can only be defined by the shows a photograph of a typical vertical powder diffrac-
means that are available for measuring the crystallinity. In tometer system, and Fig. 7b illustrates the geometry of
general, the shorter the wavelength, the smaller the crys- the system. This geometric arrangement is known as the
talline region that can be recognized. Even noncrystalline Bragg–Brentano parafocusing system and is typified by a
materials have a degree of order, and each will give some diverging beam from a line source F, falling onto the spec-
sort of a diffraction pattern. For example, glassy materials imen S, being diffracted and passing through a receiving
and liquids will give diffraction patterns of sorts, gener- slit R to the detector. Distances FA and AR are equal. The
ally in the form of one or more broad diffuse peaks or amount of divergence is determined by the effective focal
halos. A crystalline substance has a definite form, which width of the source and the aperture of the divergence slit
is retained no matter what the physical size of the crystal. D. Axial divergence is controlled by two sets of parallel-
A certain type of crystal can thus be defined in terms of plate collimators (Soller slits) P and RP placed between
specific physical characteristics that determine its shape. focus and specimen and between specimen and scatter slit,
Since every ordered material is made up of a unique ar- respectively.
rangement and number of atoms, every ordered material Use of the narrower divergence slit will give a smaller
will give a diffraction pattern that is, to all intents and specimen coverage at a given diffraction angle, thus al-
purposes, also unique. lowing the attainment of lower diffraction angles where
In X-ray powder diffractometry one is generally dealing the specimen has a larger apparent surface (thus larger val-
exclusively with polycrystalline materials, and the speci- ues of d are attainable). This is achieved, however, only
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
FIGURE 7 (a) The vertical diffractometer. A Philips PW1050 vertical goniometer system. (b) Geometry of the
Bragg–Brentano diffractometer. The geometrical layout of a typical diffractometer system showing source F, Soller
slits P and RP, sample S, divergence slit D, and receiving slit R. The axis of the goniometer is at A.
at the expense of intensity loss. Choice of the divergence intensity as a function of diffraction angle, which may be
slit, plus its matched scatter slit, is thus governed by the in the form of a strip chart or a hard copy from a computer
angular range to be covered. The decision as to whether or graphics terminal.
not the slit size should be increased at a given angle will The powder method derives its name from the fact that
be determined by the available intensity. A photon de- the specimen is typically in the form of a microcrystalline
tector, typically a scintillation detector, is placed behind powder, although, as has been indicated, any material that
the scatter slit and converts the diffracted X-ray photons is made up of an ordered array of atoms will give a diffrac-
into voltage pulses. These pulses may be integrated in a tion pattern. The possibility of using a diffraction pattern
rate meter to give an analog signal on an x/t recorder. By as a means of phase identification was recognized by 1935,
synchronizing the scanning speed of the goniometer with but it was not until the late 1930s that a systematic means
the recorder, a plot of degrees 2θ versus intensity, called of unscrambling the superimposed diffraction patterns was
the diffractogram, is obtained. A timer/scaler is also pro- proposed by Hanawalt, Frevel, and Rinn. Their technique
vided for quantitative work and is used to obtain a mea- was based on the use of a file of single-phase patterns,
sure of the integrated peak intensity of a selected line(s) characterized in the first stage by their three strongest re-
from each analyte phase in the specimen. A diffracted flections, and a search technique based on matching strong
beam monochromator may also be used in order to im- lines in the unknown pattern with these standard pattern
prove signal-to-noise characteristics. The output from the lines. A potential match was then confirmed by a check
diffractometer is a “powder diagram,” essentially a plot of using the full pattern in question. The identified pattern
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
was then subtracted from the experimental pattern, and found, a match is made with the complete pattern. If all
the procedure was repeated on the residue pattern until all lines agree, a phase confirmation is assumed and the lines
lines were identified. for the match are subtracted from the original pattern. This
Manual techniques for this “search/matching” process process is repeated until all significant lines in the pattern
have changed little over the years. In the hands of experts are identified.
manual search/matching is an extremely powerful tool, The two basic parameters being used in this search/
but for the less experienced user it can be rather time- match process are the d values calculated from the mea-
consuming. Typically 2–4 h may be required for the com- sured 2θ values in the diffractogram and the relative inten-
plete identification of a four-phase mixture. A growing sities of the lines in the pattern. Whereas the d value can be
complication is that the file of standard patterns increases accurately measured (with an accuracy of better than 0.5%
by about 2000 each year and currently stands at about in routine analysis), the intensities are rather unreliable by
46,000 entries. Most manual methods of search/matching comparison and can be subject to error, sometimes run-
that are used to identify phases in an unknown mixture ning into tens of percent. Because of the unreliable nature
are similar to that shown in Fig. 8. The three strongest of the measured intensities, any search procedure based
lines in the pattern are used to locate potential matches in on the selection of lines by their intensities must be used
the index of standards. Each time a potential candidate is with caution.
The responsibility for the maintenance of the Powder
Data File lies with the International Centre for Diffraction
Data (JCPDS), which is a nonprofit organization located
in Swarthmore, Pennsylvania. This group is made up of
a staff of permanent officers along with a number of aca-
demic and industrial scientists who are active in the field
of X-ray powder diffractometry. The Powder Data File
is a unique assembly of good-quality single-phase pat-
terns and is used by thousands of chemists, geologists, and
materials scientists. The automation of the search/match
process promises to make the use of the file even more
widespread, since this should do much to relieve much of
the tedium associated with manual search/matching.
limited number of experts in the field. In more recent years In practice it is rather unusual to analyze specimens in
the minicomputer has played an enormous role in increas- which nothing at all is known; generally one has access
ing the effectiveness of searching, so that now relatively to information about, for example, sample history and en-
inexperienced users are also able to analyze even rather vironment. In addition, other analytical information can
complex mixtures. be employed where available. As an example, in the rou-
The whole of the JCPDS Powder Data File can now tine analytical laboratory it is common to find the X-ray
be stored onto a 10-megabyte disk, and suitable search diffractometer and the X-ray spectrometer being used to
match programs can be run on an average-size minicom- give complementary information. This type of informa-
puter. As an example, Fig. 9 shows a flow chart of one tion can be used to impose what are called “hard” and
such commercially available program, typical of what is “soft” restrictions to the search. The restictions are used
currently available from most instrument manufacturers. to generate a subset of the total data file, and that subset
The diffractogram is reduced to a list of d spacings and is searched pattern by pattern, matching the d/I list of
intensity values by means of a peak searching routine, and the unknown sample to those known phases in the sub-
this d /I list is then submitted for file search/matching. In set. In the example given in Fig. 9, a probability matching
order to simplify the search procedure further, the JCPDS scheme is used for this latter process. The final output from
has subdivided the file of standard patterns into mineral, the program is a list of possible matches, each given with
inorganic, organic, metals, and alloys. Where these sub- a score representing the certainty of the match. In addi-
file restrictions are known, the total search procedure can tion, some method of graphically representing the original
be restricted to the appropriate subfile(s), giving a signi- pattern and possible matched candidates is also provided.
ficant reduction in the time required for the search/match Although the X-ray powder diffraction technique is in-
process. A file of common phases, sometimes called fre- valuable for qualitative phase identification, it is neither
quently encountered phases, is also available, and a pre- a sensitive nor a rapid means of analysis. The minimum
search is made of this file, plus a prematch of selected detectable limit found by routine qualitative procedures
preconceived possibilities. As an example, if one were would be of the order of several percent, compared, for
analyzing a specimen obtained by aspirating air through a example, with a few parts per million in X-ray fluores-
filter paper, and the X-ray spectrometer indicated signifi- cence. A complete analysis would take several hours to
cant concentrations of silicon and iron, likely candidates to complete, although this will of course depend to a very
search for in the diffraction pattern would be SiO2 , Fe2 O3 , large extent on the experience of the analyst and the com-
and FeSiO3 . plexity of the problem in hand. The method is applicable
to almost any crystalline material, whether inorganic or
organic.
Once the presence of a phase has been established in a
given specimen, one can, at least in principle, determine
how much of that phase is present by use of the intensities
of one or more diffraction lines from the phase. However,
as has been pointed out previously, it may be difficult to
obtain an accurate value for these intensities. The inten-
sities of the diffraction peaks are subject to a variety of
random and systematic errors, and the errors fall roughly
into three categories:
influence on the quantitative procedure. If one calibrates include grinding, sieving, spray-drying, and a host of other
the diffractometer with a sample of the pure phase of in- techniques.
terest and then uses the same conditions for the analysis The areas of application of quantiative X-ray powder
of the unknown mixture, the systematic errors that are diffraction are many and varied. Many hundreds of ana-
instrument-dependent are constant and only the random lysts are using this technique on a daily basis. Some of the
errors associated with counting statistics need be consid- more common applications would include ore and min-
ered. However, the specimen-dependent factors remain. eral analysis, quality control of rutile/anatase mixtures,
The most important of these are the effects of size, dis- retained austenite in steels, determination of phases in air-
tribution, and orientation of particles and the effects of borne particulates, various thin-film applications, study of
absorption. catalysts, and analysis of cements. The current state of the
The absorption effect has already been referred to in art in the quantitative analysis of multiphase materials is
Section II. Clearly, in a multiphase mixture, different that accuracies of the order of a percent or so can be ob-
phases will absorb the diffracted photons by different tained in those cases where the particle orientation effect
amounts. As an example, the mass absorption coefficient either is nonexistent or has been adequately compensated
for CuKα radiation is 308 cm/g for iron, but only 61 cm/g for.
for silicon. Thus iron atoms are five times more efficient
than silicon atoms in absorbing CuKα photons. There is
a variety of standard procedures for correcting for the ab- X. COMBINED DIFFRACTION AND
sorption problem, of which by far the most common is SPECTROSCOPIC TECHNIQUES
the use of the internal standard. In this method a standard
phase is chosen that has about the same mass absorption Like all analytical techniques, both X-ray powder diffrac-
coefficient as the analyte phase, and a weighed amount of tion and X-ray fluorescence have their advantages and dis-
this material is added to the unknown sample. The relative advantages. Table I indicates the more important features
intensities of lines from the analyte phase and the inter- of the two techniques. In terms of the range of application,
nal standard phase are then used to estimate the relative X-ray fluorescence analysis allows the quantitation of all
concentrations of internal standard and analyte phases. elements in the periodic table from fluorine (atomic num-
The relative sensitivity of the diffractometer for these two ber 9) upward. Accuracies of a few tenths of a percent
phases is determined by a separate experiment. Other pro- are possible, and elements are detectable in most cases
cedures are available for the analysis of complex mixtures, to the low parts per million level. X-ray powder diffrac-
but these are beyond the scope of this particular work. For tometry is applicable to any ordered (crystalline) material,
further information the reader is referred to specific texts and although it is much less accurate or sensitive than the
dealing with the X-ray powder method. fluorescence method, it is almost unique in its ability to
The handling of particle problems is unfortunately differentiate phases.
much more complex. As has been previously stated, the The techniques differ widely in terms of their speed of
powder method requires a specimen that is randomly ori- analysis. The modern multichannel wavelength-dispersive
ented since the geometry of the system requires that an spectrometer is able to produce data from 20–30 elements
equal number of crystallites be in the correct position in less than 1 min. An energy-dispersive spectrometer,
(i.e., orientation) to diffract at any diffraction angle where or even a relatively simple wavelength-dispersive spec-
the goniometer happens to be. Where particles lie in a trometer, is able to perform a full qualitative analysis on
preferred orientation there will be more particles avail- an unknown sample in less than 30 min. The diffraction
able to diffract at the angle corresponding to this orien- technique, on the other hand, can take as much as an
tation, and, equally important, fewer particles available
to diffract at other diffraction angles. The overall effect
TABLE I Features of X-Ray Spectrometry and
is to enhance some intensities and to diminish others.
Diffraction
Thus, the intensities are dependent on particle distribu-
tion and orientation. Some materials, just by virtue of Feature Spectrometry Diffractometry
their crystal habit, may become preferentially oriented
Range Z >5 All ordered materials
during sample preparation. As an example, mica, being
Speed Seconds to minutes Minutes to hours
a rather “platy” material, will prefer to stack one plate on
Precision 0.1% 0.25%
top of another rather than take up a random orientation.
Accuracy 0.1–1.0% 0.5–5%
The overall effect of preferred orientation can introduce
Sensitivity Low ppm 0.1–2%
errors that vary from insignificant tens of percent. Care-
Cost $50,000–250,000 $25,000–125,000
ful specimen preparation is always called for, and this may
P1: GPQ/GPA P2: GPB Final Pages
Encyclopedia of Physical Science and Technology EN017B-828 August 2, 2001 22:4
order of magnitude longer than this. Even with the most business. There are after all only 100 or so elements, and
sophisticated computer-controlled powder diffractome- within the range of the conventional spectrometer each
ters available today, diffraction experiments are invariably element gives, on an average, only half a dozen lines. In
very time-consuming. A similar difference also exists in diffraction, on the other hand, there are as many as sev-
the sensitivities of the two methods. Whereas the fluores- eral million possible compounds, each of which can give
cence method is able to measure signals from as little as on an average 50 or so lines. However, a combination of
1 ppm of a given element, the diffraction technique is often spectrometric and diffraction data can often be used to
hard put to measure 1%. greatly simplify what would otherwise be a complex and
The spectrometry method also outperforms the diffrac- time-consuming analysis.
tion technique both in precision and accuracy, perhaps by
a factor of two to six. The diffraction technique is at best a
rather insensitive, slow technique, giving somewhat poor SEE ALSO THE FOLLOWING ARTICLES
quantitative accuracy. On the other hand, it will be realized
that the information given by the X-ray diffraction method CRYSTALLOGRAPHY • LASERS, X-RAY • X-RAY PHOTO-
is unique, and no other technique is able to provide such ELECTRON SPECTROSCOPY
data. This is not true of the X-ray fluorescence method,
since there are many other techniques available to the an-
alytical chemist today for the quantitation of elements. BIBLIOGRAPHY
Atomic absorption, inductively coupled plasma emission,
and solid-source mass spectrometry are all examples of Automated powder diffractometry, new dimensions in instrumentation
competing methods. and analytical software. (1983). Norelco Reporter (special issue) 30.
The fluorescence and diffraction techniques are to a Barrett, C. S., and Gilfrich, J. V. (1990). “Advances in X-Ray Analysis,”
Vol. 33. Plenum, New York.
large extent complementary, since one allows accurate Dzubay, T. G., ed. (1977). “X-Ray Fluorescence Analysis of Environ-
quantitation of elements to be made and the other allows mental Samples,” Ann Arbor Science, Ann Arbor, Mich.
qualitative and semi-quantitative estimations to be made Holy, V., Pietsch, U., and Baumbach, T. (1998). “High-Resolution X-Ray
of the way in which the matrix elements are combined to Scattering from the Films and Multilayers,” Springer Tracts in Modern
make up the phases in the specimen. Physics, Vol. 149, Springer-Verlag, Berlin.
Jenkins, R. (1984). X-Ray technology. In “Kirk-Othmer Encyclopedia of
Both the simultaneous wavelength-dispersive spec- Chemical Technology,” Vol. 24, 3rd ed., pp. 678–708. Wiley, London.
trometer and the energy-dispersive spectrometers lend Jenkins, R., Gould, R. W., and Gedcke, D. A. (1981). “Quantitative X-
themselves admirably to the qualitative analysis of ma- Ray Spectrometry,” Marcel Dekker, New York.
terials. As was shown in Eq. (3), there is a simple relation- Kikkert, J. N. (1984). X-ray spectrometry in the eighties. Norelco Re-
ship between the wavelength or energy of a characteristic porter 31, 20–25.
Klug, H. P., and Alexander, L. E. (1974). “X-Ray Diffraction Proce-
X-ray photon and the atomic number of the element from dures,” 2nd ed. Wiley, New York.
which the characteristic emission line occurs. It was also Michelte, A., and Pfauntsch, S., eds. (1996). “One Hundred Years of
stated in Section II that each element will emit a num- X-Rays,” John Wiley & Sons, New York.
ber of characteristic lines within a given series (K, L, M, Nagel, D. J. (1982). Potential characteristics and applications of X-ray
etc.). Thus by measuring the wavelengths, or energies, of a lasers. In “Advances in X-Ray Spectroscopy” (M. Bonnelle and C.
Mande, eds.), pp. 371–410. Pergamon, Oxford.
given series of lines from an unknown material, the atomic Siegbahn, K., ed. (1967). “Atomic Molecular and Solid State Structure
numbers of the excited elements can be established. Be- Studied by Means of Electron Spectroscopy,” Almquist, Uppsala.
cause the characteristic X-ray spectra are so simple, the Turcu, I. C. E., and Dance, J. B. (1998). “X-Rays from Laser Plasmas:
actual process of allocating atomic numbers to the emis- Generation and Applications,” John Wiley & Sons, New York.
sion lines is a relatively simple process and the chance of Van Griener, R. E., section ed. (2000). “X-ray spectrometry.” In “Ency-
clopedia of Analytical Chemistry” (R. A. Meyers, ed.), John Wiley &
making a gross error is rather small. The procedures for Sons, Chichester, U.K., pp. 13269–13443.
the qualitative analysis of multiphase materials with the Zevin, L. S., Kimmel, G., and Mureinik, I. (1995). “Quantitative X-Ray
X-ray powder diffractometer are a much more complex Diffractometry,” Springer-Verlag, Berlin.
P1: LEF/GRI P2: GTV Final Pages Qu: 00, 00, 00, 00
Encyclopedia of Physical Science and Technology EN017H-830 August 2, 2001 19:31
X-Ray Photoelectron
Spectroscopy
Charles C. Chusuei
D. Wayne Goodman
Texas A&M University
I. Fundamentals of XPS
II. Instrumentation
III. Analysis of Electronic and Chemical Structure
of Atoms, Molecules
IV. Examples
921
P1: LEF/GRI P2: GTV Final Pages
Encyclopedia of Physical Science and Technology EN017H-830 August 2, 2001 19:31
state appears as a peak along with the main core-level in the nomenclature of a measured XP spectral peak (e.g.,
photoionization. Ca 2p3/2 where l + s = 1 + 1/2 = 3/2).
Surface sensitivity The ability to probe the topmost (on The technique is sensitive to all elements except H and
the order of angstroms) atomic layers of a solid. He and has a detection limit of about 0.1% of a monolayer.
Work function The energy required to remove an elec- There is little systematic overlap of the spectral lines be-
tron from the solid. This energy level above the Fermi tween elements albeit some does occur, such as Ca 2p/Au
reference is known as the vacuum level. 3d, Pt 4f/Al 2p, O 1s, Sb 3d and Al 2s, 2p/Cu 3s, 3p. Dif-
X-ray-excited Auger emission A secondary electron ficulties arising from spectral overlap are easily overcome
emission process that follows the photoionization and via examination of additional transitions (excitations) that
appears as a peak in the X-ray photoelectron spectrum. can arise from the same element. In well-calibrated XPS
After the initial photoemission, an upper level valence systems, the precision of the quantitation measurements
electron relaxes into the vacant core-level state, fol- is typically within ±5%; thus, XPS is well-suited for both
lowed by an ejection of another electron in the valence quantitative and qualitative elemental analysis.
level. In the photoemission process, an atom absorbs a pho-
ton of a known energy (hν) resulting in an ejection of a
core level electron, which is detected and its energy mea-
X-RAY photoelectron spectroscopy (XPS) is an important sured. Figure 1 shows an energy level diagram relevant for
and widely used surface analysis method in a many fields binding energy measurements of a photoelectron escap-
of study in physics and chemistry (e.g., microelectron- ing from the solid (sample specimen) and into the elec-
ics, heterogeneous catalysis, environmental geochemistry, tron spectrometer. The conducting specimen is in electri-
etc.). The technique probes the energy distribution of elec- cal contact with the spectrometer housing so that both the
trons ejected from solids via irradiation by X-rays and the sample and spectrometer have a common reference for
photoelectric effect; the electrons contain information re- measuring electron energy, called the Fermi level, E F . In-
garding chemical oxidation state, electronic structure, and coming photons (with energy h ν) create a photoelectron
atomic composition of the analyte being studied. Thus, with kinetic energy, E k1 relative to the vacuum level, E v ,
surface composition as well as the electronic environment of the sample. The kinetic electron energy at the sample
can nondestructively (in many cases) be determined. X-ray surface, E k1 , is determined from the kinetic energy of the
photoelectron spectroscopy is also useful for quantitative electron, E k , measured inside the spectrometer, from the
analysis, capable of probing ultrathin layers of material relation:
(0.1% of a monolayer). Since the photoelectrons analyzed
emanate only from the topmost atomic layers of the solid E k = E k1 − (φspec − φs ) (1)
surface being studied (≤100 Å), the technique is an in- where φspec and φs are work functions (energies required
valuable tool for studying interfacial phenomena at the to remove electrons from the Fermi to vacuum levels) of
solid-solid and solid-gas boundaries. the spectrometer and sample, respectively. From Fig. 1,
it is evident that the binding energy of analyte from an
electrically conducting sample may be obtained from the
I. FUNDAMENTALS OF XPS
relation:
A. The Photoelectric Effect and h ν = E b + E k + φspec (2)
Secondary Processes
where h ν = photon energy from and an X-ray source and
X-ray photoelectron spectroscopy (XPS) operates on the E b = binding energy. Notably, φs is not involved in the
principle of the photoelectric effect, which occurs via measurement of the E k that is “seen” by the spectrometer.
a primary excitation process brought about by X-ray- In practice, the E b of the incoming electrons is computed
irradiation producing electrons ( photoelectrons) of dis- from the measured E k from Eq. 2. Photoelectrons are gen-
crete energy, containing chemical information regarding erated via ejection from the solid surface upon excitation
the surface analyte. It should be noted that X-rays are only of core level orbitals by a photon source:
one of many types of excitation sources that can be used
A + h ν → A+∗ + e− (photoelectron) (3)
to induce emission of electrons for analysis. X-ray photo-
electron (XP) spectral peaks (generated by the photoelec- where A is a neutral atom or molecule, A+∗ is the excited
trons) are named according to the orbital (l = 0, 1, 2, 3 . . . ion and e− is the ejected photoelectron. The photoion-
denoted as s, p, d, f. . .) and spin (s = ±1/2) quantum num- ization along with competitive secondary processes are
bers of the core levels from which they emanate. The total illustrated in Fig. 2. XPS owes its relatively nondestruc-
momentum of the photoelectrons (J = l ± s) is included tive nature to the fact that only the ejection of electrons is
P1: LEF/GRI P2: GTV Final Pages
Encyclopedia of Physical Science and Technology EN017H-830 August 2, 2001 19:31
FIGURE 3 XP survey scan of Ni/H-ZSM-5 catalyst (DeGussa) B. Properties of Solids and Surface Sensitivity
containing O KLL and Ni LMM Auger transitions.
In order for electron spectroscopy to be effective, the
ejected photoelectron has to “escape” from the solid, into
spectrum is a plot of the counting rate versus the binding the vacuum level, and finally into the electron energy an-
energy of the detected photoelectrons. Other XPS core alyzer and detector. Only electrons located near the top
levels that are readily observed in the survey scan include surface can escape without loss in kinetic energy because
those of the Ni 2p, C 1s, O 1s, and Si 2p orbitals. Different of the high probability of inelastic scattering inside the
XPS transitions arising from various orbitals from a sin- solid. Thus, XPS has severe limitations for studying bulk
gle element can appear at multiple peak positions in the material. Surface compositions of materials can substan-
survey scan. For example, Si has photoelectron transitions tially differ from that inside the bulk due to segregation
from the 2s and 2p levels that appear at about 150 and 100 or surface contamination. On the other hand, the surface
eV (Fig. 3), respectively. Note that the high binding en- sensitivity of the technique (i.e., ability to probe the top-
ergy side of the XPS peak (e.g. O 1s, C 1s) has a higher most atomic layers) is advantageous for studying inter-
background; this increase in count is due to partial loss of faces and analyzing surface coverages that are only frac-
kinetic energy of the ejected electrons from inelastic col- tions of a monolayer thick. Thus XPS is valuable for the
lisions as it travels through the solid. Counts emanating investigation of molecular-level adsorbates on solid sur-
from these partial losses have a reduced kinetic energy and faces. Figure 4 shows a “universal curve” plot of the in-
would thus show up at higher binding energy in the XP elastic mean free path (IMFP) versus the kinetic energy
spectrum (Eq. 2). The high background observed (Fig. 3) of the electron. The IMFP is defined as average distance
at about 900 eV extending up to the bombardment pho- that a particle (photoelectrons) can travel along a trajec-
ton excitation energy is due to bremmstralung (German tory between inelastic collisions within a solid. The larger
for “braking radiation”) that is generated from the X-ray the kinetic energy, the longer the IMFP. In this particu-
source. lar plot, the IMFP of various elements (Ag, Au, Mo, Be,
The Auger effect is characterized by an upper level P, C, W) is given in angstrom units. A broad minimum
valence electron relaxation into the vacant core-level IMFP is evident at 4–10 Å corresponding to a 10–500 eV
state (after the initial photoionization), followed by an electron kinetic energy. While photoionization occurs up
ejection of another electron in the valence level. In the to a few microns below the surface, only the electrons at
nomenclature of the Auger effect, for example, for the the first few tens of angstroms exit the solid without en-
KL1 L2,3 transition in Fig. 2, the first shell corresponds ergy loss and provide most of the intensity for XPS. As
to the core level in which the initial vacancy was cre- a rule of thumb, approximately 95% of the XPS signal
ated (K) either via photoemission from XPS or electron arises from depths less than or equal to three times the
impact bombardment from an electron beam, the second IMFP.
P1: LEF/GRI P2: GTV Final Pages
Encyclopedia of Physical Science and Technology EN017H-830 August 2, 2001 19:31
FIGURE 4 Universal curve plot of electron IMFPs of various elements. [Reprinted with permission from Somorjai,
G. A. (1981). “Chemistry in Two Dimensions: Surfaces,” p. 41, Cornell University Press, Ithaca, NY.]
B. Radiation Sources
must at least span the distance between the sample and an-
alyzer, typically <1 m. Hence, in order to avoid dissipation In selecting an elemental source suitable for producing
via inelastic collisions of the escaping photoelectrons (or useful XP spectra, several factors need to be consid-
Auger electrons in the case of XAES), pressures of 10−5 ered: (1) the energy resolution of the X-rays; (2) energy
Torr or lower are required. Second, even lower pressures of the photons that are produced; and (3) ease of ap-
(about 10−7 Torr; in the high vacuum pressure regime) are plication of the material to an anode surface. Y Mζ
required to operate the X-ray anode and filaments with- (hν = 132.3 eV; fwhm = 0.47 eV) and Zr Mζ (hν = 152.4
out damage. Finally, in surface science experiments, low eV; fwhm = 0.77 eV) lines produce photon energies that
pressures are needed to minimize surface contamination are too low in energy to be effective since they excite a
from interfering with analysis. From the kinetic theory narrow range of photoelectrons, which limits the number
of gases, higher pressures accompany increased surface of elements that can be analyzed. Cu Kα (hν = 8048 eV;
bombardment with gas molecules. Assuming a unity stick- fwhm = 2.5 eV) has a sufficiently large excitation line to
ing coefficient, a 10−6 Torr pressure for 1 sec (defined cover the full range of elements; however, its natural line
as the unit, Langmuir) will produce a 1.0 monolayer of width is too large (ideal fwhm <1.0 eV); this would limit
P1: LEF/GRI P2: GTV Final Pages
Encyclopedia of Physical Science and Technology EN017H-830 August 2, 2001 19:31
chemical oxidation state analysis due to poor spectral res- coolant fluid are removed chiefly for two reasons: (1) to
olution. Si Kα produces a sufficiently high photon en- avoid current generation that may be exhibited as artifacts
ergy and narrow resolution to be a suitable X-ray source in the XP spectrum and (2) as an electrical safety precau-
(hν = 1739.5 eV; fwhm = 1.00 eV). However, its insu- tion due to the fact that the anode is typically operated at
lating properties and difficulties associated with applying high voltage.
the material onto an anode surface make its use imprac- The drawback of using discrete line sources (as men-
tical. The Al Kα (hν = 1486.6 eV; fwhm = 0.85 eV) and tioned above) is that photoionization cross sections for
Mg Kα (hν = 1253.6 eV; fwhm = 0.70 eV) lines suffer different core levels excited by a fixed photon energy dif-
none of these limitations and are almost universally used fer markedly; some cross sections will be at a maximum
in laboratory XPS instruments. Both have sufficient pho- while others are at a minimum for the same line energy.
ton energy to excite characteristic lines of a wide range Moreover, as discussed above there are severe limitations
of elements and their natural line widths are sufficiently in the number of choices for other fixed line sources. The
narrow to allow differentiation between multiple oxida- use of a continuously tunable source to circumvent this ob-
tion states. Both anodes produce XP spectra where over- stacle is available in the form of synchrotron radiation. In
lap of photoelectron and X-ray-excited Auger lines can synchrotron, electrons are accelerated to near-relativistic
occur in certain photoelectron binding energy regions, speeds via pulsed magnetic fields around a circular ring,
complicating analysis. For this reason, most commercial known as a torus, giving rise to X-ray irradiation. These
systems produce X-ray sources with a dual anode sys- electrons emit light in a continuous spectrum, having a
tem (e.g., Mg Kα/Al Kα). The XAES/XPS signal overlap shape dependent upon the radius of curvature and elec-
of one anode is oftentimes not present when switching tron energy. Since the radiation is concentrated into a nar-
to the other anode for use. Both anodes can be used, re- row cone tangential to the electron orbit, it can be easily
spectively, during an experiment to eliminate these “blind passed into a monochromator for energy selection. Pho-
spots.” ton energies can be monochromatized to very high reso-
It should be noted that while the Mg or Al radiation lution in this manner. Another distinct advantage of us-
source is typically labeled with “Kα” for the sake of ing synchrotron radiation is that surface sensitivity can
brevity, this unmonochromatized, “natural” line (spec- be tuned via adjusting the photon energy close to that of
trum) radiation is more complex. The X-ray lines (of each the photoionization threshold. Since the IMFP is propor-
source) are an unresolved doublet, Kα1,2 , that have satel- tional to the kinetic energy of the photoelectron, kinetic
lite features, Kα3,4 and Kβ. Excitations from the Kα3,4 energies can be tuned to about 25–100 eV. The photo-
and Kβ appear as satellites along side the photoelectron electron signals thus obtained would have high surface
peak(s) produced by the Kα1,2 and lead to spectral mis- sensitivity.
interpretation. Kα3,4 excitations are ∼10 eV on the low Synchrotron XPS systems are immense in size. The
binding energy side of the main core-level peak and is torus is typically tens of meters in diameter, providing
approximately one-tenth of the Kα1,2 intensity; Kβ exci- energy and current outputs of several GeV and 1A, re-
tations appear ∼70 eV on the lower binding energy side. spectively. Stronger S/N is readily seen (as compared with
Removal of the satellites can be achieved via monochrom- single photon line XPS instruments) throughout the entire
atization of the X-ray source. These satellite intensities to equivalent conventional XPS binding energy regime. In
some degree in nonmonochromatized systems are elimi- addition to the tunability of the X-ray source, the syn-
nated by using an aluminum foil X-ray window (approxi- chrotron radiation is 100% plane-polarized, pulsed, and
mately 8 µm thick) to reduce the bremsstrahlung radiation has the advantage of allowing for time-resolved experi-
from the X-ray source. ments to be performed. Due to its considerable expense in
Mg Kα and Al Kα radiation are typically excited with construction, repair, maintenance, and operation as well
electron energy that is about 10 times greater than their re- as the large space needed to house the apparatus, syn-
spective photon energies (e.g., 15 kV, 300 W) for efficient chrotron XPS systems tend to be nationally owned and
production of photons. To insure optimum sensitivity and shared among numerous outside researchers. Thus, there
detection, the analyte surface is brought as close to the X- are severely restricted time allowances available for exper-
ray source as possible. Since a relatively large amount of iments. Synchrotron instruments, due to their complexity,
energy is used to generate the X-ray source within a com- also tend to be out of operation for maintenance and repairs
pact area (in which the X-ray anode is held), the anode much more frequently than laboratory XPS instruments.
must be kept sufficiently cool to avoid damage (to both Nevertheless, to many scientists the gains offered by a syn-
the sample and anode). Deionized cooling water is typi- chrotron source (not available to conventional laboratory
cally used to transfer heat away from the anode (in thermal XP spectrometers) warrant the inconvenience associated
contact with a copper block). Conducting ions from the with its use.
P1: LEF/GRI P2: GTV Final Pages
Encyclopedia of Physical Science and Technology EN017H-830 August 2, 2001 19:31
C. Binding Energy Calibration the Au 4f7/2 and Cu 2p3/2 orbitals are thus readily acces-
sible for instrumental calibration. Accompanying XAES
Since the XPS binding energy scaling may be compressed
Cu LMM features are also seen in the spectrum.
or expanded, points both at high and low binding energy
bounds are needed in order to calibrate the entire binding
energy span. The calibration procedure needs to be un- D. Sample Charging
dertaken periodically to insure accurate binding energy
Many sample specimens have electrically insulating prop-
measurements. This is typically accomplished with
erties that would affect the XPS binding energy measure-
sputter-cleaned metal foils that are free of contaminants,
ments and impede spectral interpretation. Sample charg-
such as carbon and oxygen. Energy positions of well-
ing occurs as a result of an accumulation of positive charge
documented lines include Au 4f7/2 (84.0 eV; low) and
(as electrons are ejected away from the solid during pho-
Cu 2p3/2 (932.7 eV; high) are typically used. Binding en-
toemission) onto the surface when grounding of the sam-
ergy peak centers of their respective photoelectron transi-
ple (that would otherwise restore electrical neutrality) is
tions are obtained via high-resolution narrow scans. The
impaired. For the nonconducting surface, electrons can-
distance (i.e., scaling) between the high and low binding
not return to the surface easily and are lost faster than they
energy points is a parameter known as the scale factor,
return. After a steady state is reached, a positive charge
which can be controlled using the instrumental software.
develops due to deficiency in electron density. Photoelec-
In practice, the scale factor is first adjusted and then fol-
trons from the analyte would thus be ejected with a de-
lowed by changing the work function [φ in Eq. (2)] to
creased kinetic energy (higher binding energy) along with
shift the entire binding energy scale so that the measured
broadening of the peak fwhm.
peaks match those of the standards. The work function ad-
Corrections for charging can be accomplished by ei-
justment oftentimes will change the scale factor and vice
ther (1) the use of reference standards or (2) an electron
versa; thus, both variables should be changed taking into
flood gun to neutralize the positive surface charge. One
account the results of their interplay. This procedure thus
common practice is to evaporate Au dots onto the sur-
requires multiple sets of XPS scans of high and low bind-
face (after detailed high-resolution scans) and use the Au
ing energy core levels. Figure 5 shows an XP survey scan
4f7/2 level as the reference. In the case where the sam-
of Au and Cu foils juxtaposed alongside each other; both
ple specimen has been exposed to atmospheric pressure,
the C 1s line at 284.7 ± 0.2 eV denoting adventitious car-
bon (from hydrocarbons prevalent in the air) can be used.
It should be noted that this type of correction may not
adjust for changes in the scale factor due to differential
(inhomogeneous) charging, i.e., some areas of substrate
have more charge than others. In this instance, the charge
correction for a high binding energy region of the XP
spectrum may differ from that of a low one. A flood gun
may be used instead to irradiate electrons to the sample
specimen in order to neutralize charging. However, care
should be taken in its use since many materials, especially
organic substances, are sensitive to electron beam induced
decomposition.
E. Energy Analyzers
X-ray photoelectron spectroscopy analysis allows for
identification of various elements as well as differentiation
of various chemical oxidation states. In order to accom-
plish these tasks, the energy resolution must be sufficiently
high so that peaks from different chemical oxidation states
can be deconvoluted. Furthermore, the resolution should
remain relatively constant for all elemental transitions, i.e.,
FIGURE 5 Sputter-cleaned Cu and Au foils (juxtaposed together
on sample specimen holder). Binding energy peak positions of the absolute resolution, E, (defined by the fwhm) must
the Cu 2p3/2 and Au 4f7/2 levels serve as calibration standards. remain the same. If the linewidths are to be matched by
The inset shows an expanded view of the Au 4f orbital intensities. the absolute resolution at the maximum photon energies
P1: LEF/GRI P2: GTV Final Pages
Encyclopedia of Physical Science and Technology EN017H-830 August 2, 2001 19:31
FIGURE 6 Double-pass cylindrical mirror analyzer for XPS and AES analysis. [Reproduced with permission from
P. W. Palmberg, Journal of Vacuum Science and Technology A, Vol. 12, No. 1, 1975, p. 380; Copyright 1975,
American Institute of Physics.]
(e.g., fwhm = 0.70 eV for Mg Kα), a relative resolution constant pass energy (by applying voltages to the inner
(defined as R = E /E 0 where E 0 is the kinetic energy of and outer cylinders). Inner and outer magnetic shields
the peak position) of (0.70 eV ÷ 1253.6 eV =) 6 × 10−4 depicted prevent external magnetic fields from altering
would be required. Attaining this resolution is feasible but the photoelectron flight path. The entrance angle into the
the data obtained herein would suffer unacceptable losses CMA is fixed at 42.3 ± 6◦ . When the analyzer is used
in sensitivity. In practice, the kinetic energies of the incom- in the AES mode (best performed without retardation),
ing photoelectrons (toward the detector) are retarded to a these grids are grounded via a rotary motion feedthrough
preselected analyzer energy and kept fixed during data ac- adjustment.
quisition. Thus, for a pass energy of 50 eV, the relative res- Unlike the CMA which was developed for use for
olution needed would be (0.70 eV ÷ 50 eV = )1 × 10−2 , both AES and XPS, the concentric hemispherical ana-
which can be easily achieved with minimal loss in sen- lyzer (CHA) was fabricated specifically for XPS since its
sitivity. Decreasing the pass energy effectively increases inception. The CHA (Fig. 7) consists of two concentri-
resolution. In the commonly used double-pass cylindri- cally positioned hemispheres with inner (R1 ) and outer
cal mirror analyzer (CMA) setup (Fig. 6), two spheri- (R2 ) radii; φ and r are the angular and radial coordinates
cal grids at the front of the first stage are used to retard of the photoelectrons (with kinetic energy E 0 ) entering
electrons (from the sample) entering the analyzer (and fi- the analyzer at an entrance angle α. These spheres have
nally into the electron multiplier) when used in the XPS negative and positive potentials, respectively, when volt-
mode. The X-ray source is usually positioned as close as ages (V0 and V) are applied across them. The median
possible to the sample in this configuration. An electron equipotentials between the hemispheres is found at radius,
gun is housed inside the CMA to excite Auger transi- R0 = (R1 + R2 )/2. Similar to the CMA, it is customary to
tions. In some applications, the electron gun can also be preretard the photoelectrons as they enter the analyzer.
modified for use as a flood gun to correct for charging. The angle of acceptance into the analyzer is adjustable.
As a precaution, however, electron beam currents should In the case where the angle (φ) between the entrance and
not exceed 1 × 10−8 A since high currents can damage exit aperture is 180◦ , as the incoming photoelectron travel
the electron multiplier. In an Auger electron spectrom- across R0 with kinetic energy, E 0 , the deflecting potential,
eter, only a single CMA is used to optimize the lumi- eV, varies with E 0 by
nosity of the ejected Auger electrons. In this design for
R2 R1
XPS, two CMAs are connected together in series to in- eV = E 0 − (7)
crease resolution; the exit aperture of the first stage is R1 R2
the entrance aperture to the second. Thus, the double- In order to optimize energy resolution, α can be adjusted so
w
pass CMA evolved (i.e., from AES analysis) for use in that α ≈ 2R 0
where w = the slit width. Similar to the CMA
XPS systems. In front of the first stage are spherical grids system, photoelectrons are retarded to preselected pass en-
used to retard incoming photoelectrons at the preselected ergies via planar grids across the entrance slit as they enter
P1: LEF/GRI P2: GTV Final Pages
Encyclopedia of Physical Science and Technology EN017H-830 August 2, 2001 19:31
FIGURE 7 Concentric hemispherical analyzer (CHA) diagram. [Reproduced with permission from D. Roy and J.-D.
Carette, Canadian Journal of Physics, Vol. 49, p. 2141, 1971; Copyright 1971, NRC Research Press.]
the CHA. The CHA has the added option of the use of of magnitude as the IMFP, provided that the surface is
the lens system for improvement of overall efficiency of flat. Angle-resolved analysis is not available with the CMA
detection. The entrance slit to the CHA is narrow and system due to its large solid angle, which impairs its ability
one-dimensional, which gives the added advantage to detect photoelectrons emanating from a single direction.
of angle-resolved analysis capabilities used for depth
profiling the specimen.
d = λ sin α (8)
The single-dimensionality of the entrance slit also allows E of a single oxidation state observed in an XP spectrum
the added benefit for the CHA to be fitted with an input is a convolution of three components:
lens system.
E = E n2 + E 2p + E a2 (9)
III. ANALYSIS OF ELECTRONIC where E n is the natural width of the core level, E p
AND CHEMICAL STRUCTURE is the width of the photon source, and E a is the width
OF ATOMS, MOLECULES from the analyzer resolution. Both E p and E a can be
controlled experimentally by choosing an optimal photon
A. Chemical Oxidation State Determination source and varying instrumental settings of the analyzer,
respectively. The photon line widths, E p , for standard
Chemical shift data are typically obtained from high- Mg Kα and Al Kα sources are 0.70 and 0.85 eV, respec-
resolution scans of smaller binding energy regions to de- tively, and are fixed values. E n , however, depends on
termine precise peak center positions (also referred to as the uncertainty of the core hole lifetime of the ionized
narrow scans). Small differences in the binding energy state in the photoemission process. From the uncertainty
peak position (typically 0–3 eV) are attributed to chem- principle, this natural line width can be expressed as:
ical shifts arising from different electronic environments
of the ejected photoelectrons. Variations in the number of h 4.1 × 10−15
E n = = eV (10)
valence electrons and the type of bonds they form (be- τ τ
tween neighboring atoms) influence the binding energy where h = Planck’s constant (eV · sec) and τ (sec) is the
of the ejected photoelectrons, thus giving rise to chemical lifetime of the photoemission process. The fastest and
shifts exhibited in differences in the peak centers. Table I slowest lifetimes vary between 10−15 to 10−13 sec, result-
shows the chemical shifts (in eV) measured with respect ing in line widths with a lower and upper limits of 0.04 and
to the zero oxidation state. Note that for each element de- 4 eV, respectively. Thus, based on the natural line widths,
picted, as the oxidation number increases (becoming more these fwhm values represent the upper and lower bounds of
positive), binding energy increases due to greater attrac- a single chemical oxidation state. In practice, the absolute
tion of the nucleus to a core electron by the relative ab- resolution (E) is obtained from by measuring the fwhm
sence of outer valence electrons. When one of these outer of an “ideal” photoelectron peak (from the analyte of in-
shell electrons is removed, the effective charge sensed by terest) that is free from broadening effects or experimental
the core electron increases (arising from a deficiency of artifacts (having a symmetrical Gaussian shape) and has
electron density). Conversely, in the case where the at- the narrowest peak width in the XPS survey scan. The E
traction of the nucleus for a core electron is diminished obtained can serve as a guide for deconvolution of multi-
by the presence of outer electrons, photoelectrons ejected ple peaks when curvefitting spectra. For example, Fig. 9
from an electron-rich environment will exhibit a lower shows an O 1s XPS peak separated into contributions from
binding energy. Multiple oxidation states can appear in hydroxyl oxygen and the TiO2 metal oxide oxygen. The
a given narrow scan. In order to make meaningful peak curvefitting was not allowed to deviate from a 2.0–2.5 eV
assignments to these different states, issues regarding in- fwhm. The XP spectrum “A” shows the O 1s signal af-
strumental resolution need to be addressed. In theory, the ter a TiO2 substrate was exposed to aqueous solution for
−2 −1 0 +1 +2 +3 +4 +5 +6 +7
Nitrogen 1s — ∗ 0b — +4.5c — +5.1 — +8.0 — —
Sulfur 1s −2.0 — ∗0 — — — +4.5 — +5.8 —
Chlorine 2p — ∗0 — — — +3.8 — +7.1 — +9.5
Copper 1s — — ∗0 +0.7 +4.4 — — — — —
Iodine 4s — ∗0 — — — — +5.3 — — +6.5
Europium 3d — — — — ∗0 +9.6 — — — —
a All shifts are measured in eV relative to the oxidation states denoted by the asterisks (∗ ).
b Arbitrary zero measurement, end nitrogen in NaN3 .
c Middle nitrogen in NaN .
3
Reprinted with permission from Hercules, D. M. (1970). Anal. Chem. 42, 28A; Copyright 1970, American Chemical Society.
P1: LEF/GRI P2: GTV Final Pages
Encyclopedia of Physical Science and Technology EN017H-830 August 2, 2001 19:31
where C is the concentration of analyte, x, of interest be- FIGURE 11 The XP spectrum of Ca 2p level showing spin-orbit-
ing scanned. The element, x, is a subset of all elements, i. coupled l ± s = 1/2 and 3/2 peaks.
Thus, given the measured absolute intensities of different
core levels along with their respective sensitivity factors, to spin orbit coupling are generally not observed (resolved
the atomic percent composition of elements on the surface into 2 peaks) for low atomic number elements (Z ≤ 20).
can be computed. It should be noted that this formula as- A Ca 2p core level (Fig. 11), for instance, gives two pho-
sumes a homogeneous distribution of elements throughout toemission peaks: 2p1/2 (l = 1 and J = 1 − 1/2) and 2p3/2
the sample. The structural model of the analyte needs to (l = 1 and l = 1 + 1/2). Similarly, d and f orbitals can split
be known to insure accurate determinations of concentra- upon photoionization. The relative intensity ratio of the 2
tion. For example, an oxide thin film covering the topmost peaks of a spin-orbit-coupled doublet is determined by the
layers of a refractory metal substrate will attenuate under- 2J + 1 multiplicity of the levels. For example, the relative
lying photoelectron signal while accentuating the relative intensity of J = 3/2 and J = 1/2 components of a 2p is
oxygen intensity. Thus, the atom percent determination of 4:2 = 2:1; for that of 5/2 and 3/2 peaks of a 3d level, it is
the oxygen in this system would be an overestimation if 6:4 and for the 7/2 and 5/2 peaks of the 4f level, it is 8:6.
the oxide overlayer structure is not taken into account. Binding energy shifts from initial state effects can be
interpreted using the charge potential model:
C. Initial and Final State Effects qi ref
E bi = kqi + + Eb (17)
j ri j
Koopmans’ theorem predicts that the XP spectra observed
represents the electronic states of electrons in the analyte where E bi = the binding energy of an electron from an
atom before the photoemission process, i.e., the initial atom i, qi is the charge on the atom, k = constant, q j = the
state. Spin orbit coupling is an example of an initial state charge on a neighboring atom j, ri j = the distance between
ref
effect, in which spectral features arise from the inherent atom i and atom j, and E b = the energy reference. The
unpaired electron make-up of the atoms prior to the pho- kqi term indicates that binding energy increases with in-
toemission event. In this scenario, peak splittings due to creasing positive charge on the atom from which the photo-
q
energy differences between singlet and triplet states via electron emanates. The ri jj term, known as the Madelung
j
interaction of the spin and orbital magnetic moments oc- sum, in ionic solids negates the contribution from the
cur whenever there are unpaired electrons in the valence kqi term, since the neighboring atom has an opposite
shells. If the electron ejected is parallel (triplet state) to that charge.
of the valence electrons, it can undergo exchange interac- During the photoionization process, changes in the elec-
tion and result in a lower kinetic energy (higher binding en- tronic environment due to the creation of the core-level
ergy) than the case for an anti-parallel spin (singlet state). vacancy leading to final state effects play a large role
The result is a doublet in the XP spectrum. Splittings due in influencing the binding energy. In order to properly
P1: LEF/GRI P2: GTV Final Pages
Encyclopedia of Physical Science and Technology EN017H-830 August 2, 2001 19:31
bulk value) supported on various transition metals simi- resulting spectrum is indicative of the admetal binding
larly follows the trend: Ta > W > Re > Ru. The core-level strength to the surface.] Desorption temperatures of
shift decreases in the left-to-right direction of the periodic Pd monolayers from several surfaces are presented in
table. The magnitude of the perturbations induced by the Fig. 14b. These data show a general trend of stronger
loss of electron density increases as the fraction of empty Pd-substrate bond strengths for elements with the least
levels in the valence band of the metal substrate increases: occupied valence band. The Pd/Ta(110) system has a
Ta > W > Re > Ru (Fig. 14a). Cu, on the other hand, has a desorption temperature of 1540 K, the highest observed,
4s valence band that is half empty and can act as either an and the Pd/Ru(0001) has the lowest desorption temper-
electron donor or acceptor, depending upon the fraction ature (1440 K). The larger the electronic perturbations
of empty states in the valence band of the metal substrate. on the Pd atoms, the stronger the bimetallic bond. An
Interestingly, there is an excellent correlation between interrelationship between the magnitude of the Pd 3d5/2
the changes in the admetal binding energies and the core level shift and the Pd-substrate binding strength
relative abilities of these transition-metal substrates to is thus readily observed. Similarly, a strong correlation
adsorb Pd as measured by temperature programmed has been found between changes in the XPS core-level
desorption (TPD), which can also be explained using the binding energies and variations in the ability for the
model described above. [The TPD technique involves supported films to adsorb CO. Temperature programmed
adsorbing the analyte metal onto a substrate followed by desorption data indicate CO desorption temperatures
linearly heating the sample to desorb it and subsequent from Pd1.0 ML /Ta(110) (∼250 K), Pd1.0 ML /W(110)
detection via mass spectrometer. A plot of the ion current (∼300 K), Pd1.0 ML /Ru(0001) (∼330 K) that are much
(mass spectral intensity) is obtained as a function of lower than that from Pd(100) (∼480 K). According
temperature. The peak temperature maximum of the to the above-described model, the electron density of
supported monolayers of Pd is smaller than that of the
surface atoms of Pd(100). A partial positive charge
on the Pd adatoms is consistent with a reduction in
their ability to coordinate CO via π –back–bonding,
producing a weaker Pd CO bond on the supported
monolayers compared to Pd(100). The model described
above predicts that the Pd surface atom electron density
and ability to π –back–donate will follow the sequence
Pd1.0 ML /Ta(110) < Pd1.0 ML /W(110) < Pd1.0 ML /Ru(0001)
< Pd(100). This exact trend is exhibited in the TPD
data. For monolayers of Cu (having a relatively electron-
deficient density) deposited on electron-rich metals
(Ru, Rh, and Pt), the model predicts an enhancement
in electron density and π –back–bonding capacity of
the Cu adatoms with respect to the surface atoms of
Cu(100). Adsorption of CO on Cu films induces a large
increase in the Cu 2p3/2 XPS binding energy, in part,
due to π –back–donation. Figure 15 illustrates how the
CO-induced shift in the Cu 2p3/2 peak position (measured
after saturating the Cu surfaces with CO at 100 K)
changes with the metal substrate. The CO induced shift
increases as the metal substrate “moves” from left to right
in the periodic table. This trend is attributed to an increase
in the ability of the Cu adatoms to π –back–donate.
Briggs, D. (1998). “Surface Analysis of Polymers by XPS and Static D. R. (1999). Surface sensitivity of Auger-electron spectroscopy
SIMS,” Cambridge Univ. Press, Cambridge, U.K. and X-ray photoelectron spectroscopy. J. Electron Spectrosc. Related
Henrich, V. E., and Cox, P. A. (1996). “The Surface Science of Metal Phenomena 98–99, 1–15.
Oxides,” Cambridge Univ. Press, Cambridge, U.K. Somorjai, G. A. (1994). “Introduction to Surface Chemistry and Catal-
Hüfner, S. (1996). “Photoelectron Spectroscopy: Principles and Appli- ysis,” Wiley, New York.
cations, 2nd edition,” Springer-Verlag, Berlin. Turner, N. H., and Schreifels, J. A. (2000). Surface analysis: X-ray photo-
Oleford, I. (1997). X-ray photoelectron spectroscopy. In “Surface Char- electron spectroscopy and Auger electron spectroscopy. Anal. Chem.
acterization: A User’s Sourcebook” (D. Brune, R. Hellborg, H. J. Whit- 72, 99R–110R.
low, and O. Hunderi, eds.), pp. 291–319. Wiley-VCH Verlag, Wein- Woodruff, D. P., and Delchar, T. A. (1994). “Modern Techniques of
heim, Germany. Surface Science, 2nd edition,” Cambridge Univ. Press, Cambridge,
Powell, C. J., Jablonski, A., Tilinin, I. S., Tanuma, S., and Penn, U.K.