Professional Documents
Culture Documents
1
Electrode potentials
determined by the difference between the real species present in any solution [27]; in
chemical potentials. In the literature, the aqueous and certain other media, their
term Galvani potential is also applied to equilibrium concentration is extremely
the separate value of internal potential. low, which, however, does not prevent us
However, it is not significant, because of from formally using the thermodynamic
the relative nature of this value, which is consideration of the electrode potential in
discussed below. the framework of the concept of electronic
The electrochemical equilibrium re- equilibria. However, in real systems the
quires the equality of electrochemical exchange current density for Mz+ /M pair
potentials for all components in the con- appears to be so small in the limit under
tacting phases. From this condition of elec- consideration that equilibrium cannot be
trochemical equilibrium, the dependence maintained, and Eq. (5) no longer applies.
of Galvani potential on the activities of The equilibrium with participation of
potential-determining ions can be derived, solvated electrons can be expressed by a
which represents the Nernst equation [27] scheme
for a separate Galvani potential. That is,
for an interface formed by a metal (M) M (solid) ⇐⇒ M (in solution)
and solution (S) containing the ions of this ⇐⇒ Mz+ + zēs (6)
metal Mz+ ,
The dissolved metal atoms can be con-
RT
M
S ϕ
M S
= ϕ − ϕ = const + ln aMz+ sidered as an electrolyte, which dissociates
zF producing Mz+ cations and the sim-
(5)
where ϕ M and ϕ S are the internal poten- plest anions, ēs . Actually, the condition
tials of metal and solution, respectively, aMz+ = 0 cannot be achieved, because a
and aMz+ is the activity of metal ions in finite value for the ‘‘solubility product’’
solution. of metal exists, and, simultaneously, the
The impossibility to measure a separate condition of electroneutrality is valid. By
Galvani potential rules out the possibil- using the condition of the equality of elec-
ity of establishing a concentration corre- trochemical potentials of electrons in the
solution and in the metal phases, we ob-
sponding to M S ϕ = 0.
A formal consideration of Eq. (5) pre- tain
dicts the infinitely large limiting value of
ē − µē
µM S
RT
M
S ϕ = = const − ln aē
the Galvani potential (M S ϕ → −∞) when F F
the activity of metal ions approaches zero (7)
(a hypothetical solution containing no ions where µM ē and µ S are the chemical
ē
of this sort, or pure solvent). However, potentials of electrons in metal and
this is not the case, because the limit solution, respectively.
of thermodynamic stability of the solvent
will be exceeded. For an ideal thermo-
dynamically stable (hypothetical) solution, 1.3
this uncertainty can be clarified, if the Electromotive Force (emf) and Gibbs
electronic equilibria between the metal Energy of Reaction
and the solvent are taken into account,
which requires the consideration of sol- To study the electrical properties of the M|S
vated electrons (ēs ) [7–11]. The latter are interface presented schematically in Fig. 1,
6 1 Electrode potentials
S M
∆M
Sϕ
it is necessary to construct a correctly M M
E = M11 ϕ = M1 ϕ + SM1 ϕ + MS ϕ
connected circuit (Fig. 2) – a galvanic cell
(9)
satisfying the condition of identical metal
which corresponds to the route of probe-
contacts at both terminals. For the system
charge transfer 1 ⇒ 2 ⇒ 3 and consists
under discussion, it is a circuit of the
of three Galvani potentials. On the other
following type:
hand, when going from M1 to M1 via the
M1 |S|M|M1
(8) route marked by points 1 ⇒ 2 ⇒ 3 ⇒ 4
M1 M1′'
1′ 4′
3
3′
M
2′
these points to the interface should be The interface M1 |M (circuit 1.8) which,
sufficiently large, as compared with the in the simplest case, represents a boundary
characteristic distance of molecular and of two metals, is easy to construct.
image forces (up to 10−4 cm), and, at the However, the M1 |S contact, which includes
same time, not too high to prevent the an additional electrochemical interface, is
weakening of interaction with the charges rarely feasible, being usually unstable and
inside the phases. Volta potential is a dependent on the nature of M1 . Generally,
measurable quantity, because the points in place of M1 , a special electrochemical
between which it is measured are located system, the so-called reference electrode
in the same phase and exclusively in the should be included into cell 1.8. The
field of long-range forces. This means that electrode potential can be determined
the Volta potential does not depend on the as the emf E of a correctly connected
charge of the probe. electrical circuit formed by the electrified
For the metal/metal boundary, these po- interface under discussion and a reference
M
tential differences (M1 ψ) can be defined electrode. According to this definition,
by the difference of work functions [22]. the potential of any reference electrode
The latter can be obtained from the pho- is assumed to be zero.
toelectron emission (or thermoelectron Thus, electrochemists deal with the val-
emission) and also directly by using a cir- ues of potentials, which are, actually, the
cuit differences of potentials. The nature of the
M1 |M|vacuum|M1 emf of a circuit for potential measurement
depends on the type of the electrodes. If
where M1 is the reference metal. The main we ignore a less frequent situation of ide-
experimental problems are the surface pre- ally polarizable electrodes, for the majority
treatment (purification) and the elimina- of systems, two half-reactions take place
tion of potential drop between the samples simultaneously in the cell. For ideally po-
of M1 and M in ultrahigh vacuum. larizable electrodes, their contribution to
Volta potential for the interface metal emf can be considered, in the first approx-
(mercury)/solution can be found by using imation, as the potential drop inside a ca-
the following cell reference electrode|solu-
pacitor formed by the electric double layer.
tion|inert gas|Hg|solution|reference elec-
Hence, this drop depends on the electrode
trode when the solution flows to the system
free charge (see Sects. 1.3 and 1.5):
through the internal walls of a vertical tube,
where mercury flows out via a capillary
Ox1 + n1 ē ⇐⇒ Red1
placed axially in a vertical tube and is bro-
ken into drops. These mercury drops carry (on the electrode under study), (11)
away the free charges, thus eliminating the
Ox2 + n2 ē ⇐⇒ Red2
potential difference in inert gas between
Hg and the solution. If a plate of solid (on the reference electrode). (12)
metal is used instead of liquid metal, it is
necessary to eliminate the potential drop Taken together, Eqs. (11) and (12) corre-
mentioned above. A similar technique can spond to the chemical reaction:
be used for the solution/solution inter-
face. There are also other techniques of ν1 Red1 + ν2 Ox2 ⇐⇒ ν2 Red2 + ν1 Ox1
Volta potential determination [1–6, 28]. (13)
8 1 Electrode potentials
the relationship for stoichiometric coeffi- molecules (salts, acids) in Eq. (17) can be
cients being substituted for partial activities of ions,
which cannot be measured experimentally.
n1 ν1 = n2 ν2 = n (14)
The applications of various types of
where n is the number of electrons electrochemical cells to chemical thermo-
transferred across the electrochemical cell dynamics are considered in Sect. 2.
in a single act of reaction (13). Experimentally, the emf can be mea-
From Eq. (3), the emf of the cell, sured either by compensating the circuit
which corresponds to reaction (13), can be voltage (classical technique which became
expressed as follows: rare nowadays) or by using a voltmeter
1 of very high internal resistance. The accu-
E= (ν1 µRed1 + ν2 µOx2 − ν2 µRed2 racy of emf determination of about 1 µV
nF
can be achieved in precise measurements,
− ν1 µOx1 ) (15)
whereas common devices provide the ac-
The right-hand term in parentheses curacy of about 1 mV. The potential unit
equals the Gibbs energy of reaction (13) named Volt, which is used in the modern
with the opposite sign. Hence, the emf literature (particularly, below), is the so-
of an electrochemical cell (a cell with called absolute (abs) Volt; it differs slightly
eliminated diffusion and thermoelectric from the international (int) Volt value. The
potential drops is implied, see later) corre- ratio abs/int is 1.00033. To determine the
sponds to the reaction Gibbs free energy: sign of emf, a conventional rule is adopted,
G which states that the left electrode should
E=− (16) be considered as the reference one.
nF
For carrying out the experimental mea-
This statement is true also for more surements of electrode potentials, a system
complicated reactions with the participa- chosen as the reference electrode should
tion of N reactants and products, and can be easy to fabricate, and also stable and
be generalized in the form of the Nernst reproducible. This means that any pair of
equation: reference electrodes of the same type fabri-
RT
N cated in any laboratory should demonstrate
E = E0 − νi ln ai (17) stable zero potential difference within the
nF
i=1 limits of experimental error. Additionally,
where the stoichiometric numbers νi < 0 the potential differences between two ref-
for the reactants and νi > 0 for the erence electrodes of different type should
products, ai are their activities, n is the remain constant for a long time.
total number of electrons transferred in Another point is that the transfer of
the coupled electrochemical reactions, and electricity (although of very low quan-
E 0 is the standard potential, which is tity) occurs in the course of emf mea-
expressed by the standard Gibbs free surements. Thus, the reference electrode
energy G0 as should comply with the requirement of
nonpolarizability: when the currents (usu-
G0 ally in the nanoampere range) flow across
E0 = − (18)
nF the system, the potential of the refer-
Under equilibrium conditions, the mean ence electrode should remain constant.
activities or the activities of neutral One of the most important features that
1.4 Diffusion Potential Drop 9
determines this requirement is the ex- In Planck’s model for a sharp boundary,
change current density i0 , which expresses for a 1,1 electrolyte,
the rates of direct and reverse processes
under equilibrium conditions. This quan- RT
ϕdiff = ln ξ (20)
tity determines the rate of establishment F
of equilibrium (the state of the electric where function ξ can be found from a
double layer is established very rapidly as transcendent equation in which ‘‘+’’ and
compared with the total equilibrium). ‘‘−’’ denote the sets, which consist of all
cations and of all anions, respectively, and
λ denotes the limiting conductivity of the
1.4 corresponding ion:
Diffusion Potential Drop
(2) (2) (1) (1)
ξ λ + c+ − λ + c+
The use of reference electrodes frequently + +
poses the problem of an additional poten- (2) (2)
(1) (1)
λ − c− − ξ λ − c−
tial drop between the electrolytes of the
− −
electrode under study and of the reference (2)
one. For liquid electrolytes, this drop arises ci
at the solution/solution interface (liquid i
. ln
(1)
− ln ξ
junction). The symbol .. conventionally de- ci
notes an interface of two solutions with i
= (2)
a diffusion potential drop in between; if ci
this drop is eliminated (see later), then the i
.. ln
+ ln ξ
(1)
symbol .... is used. In this case, the equilib- ci
rium is not exact because of the existence i
of a diffusion potential drop ϕdiff . The (2)
(1)
latter has the meaning of Galvani poten- ξ ci − ci
i i
tial and cannot be measured; however, it × (2)
(1)
(21)
can be estimated by adopting a model ap- ci −ξ ci
proach to the concentration distribution i i
of ions in the interfacial region, the mod-
The Henderson equation that has gained
els of Planck [29] and Henderson [30, 31]
wider acceptance can be written as follows
being the most conventional. The general
for concentrations c having the units of
expression for ϕdiff at the interface of
normality:
liquid solutions (1) and (2) is as follows:
λi (2)
RT (2) ti (1)
(ci − ci )
ϕdiff = − d ln ai (19) RT i zi
F (1) zi ϕdiff = (2) (1)
F λi (c − c ) i i
where ti is the transport number of the
i
ith ion, that is, the portion of current (1)
transferred by this ion through the solu- λ i ci
tion. In the first approximation, ϕdiff can i
× ln (2)
(22)
be estimated by substituting correspond- λ i ci
ing concentrations for the partial activities. i
10 1 Electrode potentials
The solution is given for the case of a electrode classification based on the nature
smeared out boundary and linear spatial of species participating in electrochemical
distributions of concentrations. equilibria.
Generally, Eqs. (20)–(22) yield similar Electrodes of the first kind contain elec-
results; however, for junctions with a tronic conductors as the reduced forms,
pronounced difference in ion mobilities and ions (particularly, complex ions) as
.
(such as HCl .. LiCl), the deviation can the oxidized forms. The equilibrium can
reach about 10 mV. A specific feature of be established with respect to cations and
the Planck equation is the existence of anions; in the absence of ligands, the
two solutions, the first being close to that cations are more typical. The examples
of Henderson, and the second one being are: Cu|Cu2+ . . . or . . . Au|[Au(CN)2 ]− . . . .
independent of the solution concentration This group can be supplemented also
and of no physical meaning [32]. by amalgam electrodes (or other liquid
In practice, in place of model calcula- electrodes) and electrodes fabricated from
tions and corresponding corrections, the nonstoichiometric solids capable of chang-
elimination of the diffusion potential is ing their composition reversibly (interca-
conventionally applied. This is achieved lation compounds based on carbons, ox-
by introducing the so-called salt bridges ides, sulfides, and multicomponent salts,
filled with concentrated solutions of salts, particularly, Li-intercalating electrodes of
which contain anions and cations of close batteries).
transport numbers. A widely known ex- Electrodes of the second kind contain a
ample is saturated KCl (4.2 M); in aqueous layer of a poorly soluble compounds (salt,
solutions, potassium and ammonium ni- oxide, hydroxide), which is in contact with
trates are also suitable. However, the a solution of the same anion. The equi-
requirement of equal transport numbers librium is always established with respect
is less important as compared with that of to anions. The most typical examples are
high concentration of electrolyte solution, based on poorly soluble compounds of
which fills the bridge [33, 34]. A suitable mercury and silver (Table 1).
version of the salt bridge can be chosen The redox polymers, both of organic and
for any type of cells, when taking into ac- inorganic origin (such as polyvinylpyridine
count the kind of studies and the features modified by redox-active complexes of met-
of chosen electrodes. als; Prussian blue and related materials),
In melts, an additional problem of can be considered as a version of electrodes
thermo-emf arises, and the emf correction of the second kind; however, the equilib-
can be calculated from thermoelectric rium is usually established with respect
coefficients of phases in contact [35]. to cations. Electron conducting polymers
(polyanyline, polypyrrol, and so forth) also
pertain, in the first approximation, to the
1.5 electrodes of the second kind, which main-
Classification of Electrodes tain equilibrium with respect to anion.
Ion exchange polymer films on electrode
No universally adopted general classifica- surfaces form a subgroup of membrane
tion of electrodes exists; however, when electrodes.
dealing with thermodynamic aspects of the Rather exotic electrodes of the third
electrode potential notion, we dwell on the kind (with simultaneous equilibria with
1.5 Classification of Electrodes 11
respect to anions and cations) were also Au|[Fe(CN)6 ]3− , [Fe(CN)6 ]4− · · ·
proposed for measuring certain special
characteristics, such as the solubility prod- or Pt|[Co(EDTA)]− , [Co(EDTA)]2− · · ·
uct of poorly soluble salts. The rare When equilibrium is established be-
examples are Ag|AgCl|PbCl2 |Pb(NO3 )2 . . . tween ions in solution and the gas phase,
or M|MOx|CaOx|Ca(NO3 )2 . . . where M = we have gaseous electrodes; the inter-
Hg or Pb, and Ox is the oxalate mediates of this equilibrium are usually
anion. adatoms: Pt, H2 |H+ . . ., or Pt, Cl2 |Cl− . . .
When the metal does not take part in Palladium, hydrogen-sorbing alloys, and
the equilibrium and both components of intermetallic compounds pertain to gase-
the redox pair find themselves in solution, ous electrodes. However, at the same time,
the system is called a redox electrode, these systems are example of intercalation
for example, a quinhydrone electrodes processes operating under equilibrium
(Table 1) or electrodes on the basis of conditions, which brings them close to
transition metal complexes: electrodes of the first kind.
12 1 Electrode potentials
The various kinds of electrodes consid- in which charged groups are bound with
ered above can be unified, if we take into the membrane matrix; solid membranes
account the concept of an electronic equi- reversible with respect to certain ions (as
librium at metal/solution interface, that for glass electrode); biomembranes. The
is, bear in mind that a certain activity of simplest model of a membrane electrode
ēs in solution, which corresponds to any is the liquid/liquid interface, which is
equilibrium potential, exists. prepared by contacting two immiscible
solutions and usually considered together
with liquid electrodes in terms of soft
1.6 electrochemical interface (see Sect. 3.4).
Membrane Electrodes
the zero standard entropy of H+ ions. This In general, electrodes of the second
extrathermodynamic assumption induces kind are more convenient as reference
the impossibility of comparing the values electrodes, because they do not require
referred to the hydrogen electrodes, in dif- a source of gaseous hydrogen. Dynamic
ferent solvents. Of the systems regarded as hydrogen reference electrodes, which rep-
reference electrodes, platinum hydrogen resent a wire of platinum metal (or another
electrodes exhibit i0 values ranked among metal with low hydrogen overvoltage) ca-
the greatest known (10−3 –10−2 A cm−2 , thodically polarized up to the hydrogen
at least, in aqueous acidic solutions). This evolution, give a possibility of avoiding
type of electrodes is used as the ref- the use of gaseous hydrogen. Their sta-
erence for tabulating universal potential ble potential values are determined by the
values [43–47]. existence of the current–potential relation-
The universal definition of the standard ship, which is possible for the electrode
potential E 0 of a redox couple Red/Ox is processes with high degree of stability. A
as follows: the standard potential is the special type of stable hydrogen electrode
value of emf of an electrochemical cell, is based on Pd hydride (β-phase). Another
in which diffusion potential and thermo- advantage of the electrodes of the second
emf are eliminated. This cell consists of an kind is their applicability in a wider range
electrode, on which the Red/Ox equilibria of temperatures, and also suitability for a
establish under standard conditions, and wide range of pressures. Operating at high
a SHE. temperatures presents a real challenge in
finding a suitable reference electrode (al-
though there are few examples stable up
1.8 to 250–300 ◦ C). One should never use any
Specific Features of Certain Reference reference electrode containing mercury at
Electrodes elevated temperatures. For other reference
electrodes (e.g., silver/silver chloride), sta-
An important type of reference electrodes bility and reproducibility should be tested
is presented by the so-called reversible carefully before it is used at temperatures
hydrogen electrodes (RHE) in the same so- above 100 ◦ C.
lution, which makes it possible to avoid For aqueous hydrogen electrodes, the
the liquid junction. These electrodes are potentiometric data are now available in
preferentially used in electrochemical ex- both sub- and supercritical regions, up to
periments when investigating the systems 723 K and 275 bar [48]. However, their use
with H+ and H2 involved in the process as the high-temperature reference systems
under study (adsorption and electrocataly- involves numerous complications.
sis on hydrogen-adsorbing surfaces, such Among reliable systems for high-
as platinum group metals). The RHE can temperature measurements, gaseous (first
be produced for a wide range of pH; certain of all, oxygen) electrodes based on solid
special problems associated with neutral electrolytes of the yttria-stabilized zirconia
solutions can be solved by using buffer- (YSZ) type can be recommended, which
ing agents. In media containing organic are always applied in the electrochemical
components, the possible catalytic hydro- cells with solid electrolytes. Gaseous elec-
genation of the latter poses an additional trodes based on Pt or carbon are also widely
limitation for hydrogen electrodes. used in high-temperature melts, especially
14 1 Electrode potentials
those based on oxides (oxygen reference on Tl− and Zn in liquid ammonia and hy-
electrode) or chlorides (chlorine reference drazine, and also Hg/Hg2 F2 electrode in
electrode). pure HF.
There are lots of systems, especially An exhaustive consideration of the
for electroplating and electrosynthesis, in specific features of various reference
which electrodes of the first kind can be electrodes (fabrication, reproducibility and
used, without any liquid junctions (the stability, modes of applicability, effects
example is liquid Al in AlF3 -containing of impurities, necessary corrections) can
melts). More universal systems for melts be found in Refs. [36, 51–54]. Nowadays,
of various kinds are: a chlorine electrode in certain new findings in this field are
equimolar NaCl + KCl melt and Ag/Ag+ possible because of the novel approaches
electrodes with the range of Ag+ con- to immobilization of redox centers on the
centrations (0.01–10 mM) corresponding electrode surfaces.
to usual solubility values. Reference elec- The attempts to interrelate the poten-
trodes of the second kind can hardly be tial scales in aqueous and nonaqueous
used in melts because of the high solubility solutions have been undertaken and are
of the majority of inorganic solids.
still in progress. Such a relationship
When studying nonaqueous systems by
could have been found if the free en-
means of galvanic cells with aqueous or
ergy of transfer was known at least for
mixed reference electrodes, we cannot
one type of charged species common to
avoid liquid/liquid junctions and estimate
the solvents. It is evident that ways of
the corresponding potential drop from
solving this problem can be based on
any realistic model. In protic nonaqueous
the assumptions beyond the scopes of
media (alcohols, dioxane, acetone, etc.),
thermodynamics. Thus, it was mentioned
a hydrogen electrode can be used; it is
in Ref. [18–20] that the free energy of
also suitable for some aqueous/aprotic
mixtures. However, the i0 values for the solvation of Rb+ ion is low and approx-
hydrogen reaction are much lower as imately the same in different solvents
compared with purely aqueous solutions. because of its great size. However, any
When studies are carried out in nonaque- direct application of rubidium electrodes
ous media, in order to avoid liquid/liquid is hampered by their corrosion activity as
junction preference should be given to the well as by the fact that their potential
reference electrodes in the same solvent as lies in the region of electroreduction of
the electrode of interest. a number of solvents. Another proposition
In aprotic (as well as in protic and concerns the fact that the free energies
mixed) media, the two reference sys- of transfer for certain cations and neu-
tems of choice are ferrocene/ferrocenium tral molecules of the redox pairs A/A+
and bis(biphenil)chromium(I/0). The pen- are the same in different solvents, which
tamethylcyclopentadienyl analog of the is also caused by their great sizes [51].
former was recently shown to yield higher Ferrocen/ferrocenium-like systems were
performance [49, 50]. Among other typical taken as the A/A+ pairs. It is inter-
electrodes, Ag/AgNO3 should be men- esting that these scales – rubidium and
tioned. We can also mention special A/A+ – adequately correlate with one an-
reference systems suitable for certain sol- other. Different extrathermodynamic as-
vents, such as amalgam electrodes based sumptions are compared in Ref. [54].
1.9 Values of Redox Potentials 15
compounds, which can be used for semi- is usually used, which characterizes the
quantitative estimates of potentials of width of the interval between the poten-
novel systems. The most advanced ap- tials of cathodic and anodic background
proaches known from coordination chem- processes. The narrow potential window
istry take into account σ -donating and of water (about 1.3–1.4 V on platinum
π-accepting abilities of the ligands, which group metals, and up to 2 V on mercury-
can be expressed, for example, in terms like metals, which do not catalyze the
of Hammett or Taft, and also other effec- hydrogen evolution reaction) stimulates
tive parameters [59]. For the sequences of the potentiometric studies in nonaqueous
complexes with different central ions and solvents. Aprotic solvents usually demon-
related ligands, the steric factors (namely, strate windows of about 3–3.5 V, if the
chelating ability) are known to significantly optimal supporting electrolyte is chosen.
affect the redox potentials. An extremely wide window is known, for
A separate field deals with correlating example, for liquid SO2 , which is of highest
the redox potentials with spectroscopic interest for measuring extremely positive
parameters, such as the energy of the redox potentials (up to +6.0 V (SCE)) [62].
first allowed d–d transition in a complex,
Low-temperature haloaluminate melts are
the energy of the metal-to-ligand charge-
highly promising systems [63]. Finally, a
transfer band (i.e., the separation between
number of unique mixed solvents with
the HOMO (highest occupied molecular
extremely wide windows were found in
orbital) on the metal and LUMO (lowest
recent studies of lithium batteries [64].
unoccupied molecular orbital) on the
Among electrode materials, the widest
ligand), and nuclear magnetic resonance
windows are known for transition metal
(NMR) chemical shifts. The problem
of extending these correlations over a oxides, borides, nitrides, and some spe-
wider range of reactions, including those cially fabricated carbon-based materials. It
irreversible (for which the kinetics makes should be mentioned that, if the nature
a substantial contribution to the formal of electrode material can affect the formal
potential value, the latter being evidently potential value by changing the mecha-
nonthermodynamic) was considered a nism and kinetic parameters, the solvent
long time ago [60]. Advanced spectroscopic frequently has a pronounced effect on the
techniques are widely used for solving the equilibrium potential, because of the sol-
reverse problem of determining the redox vation contribution to free Gibbs energy.
potentials of irreversible couples [61].
1.11
1.10 Experimental Techniques
Potential Windows
Direct potentiometry (emf measurements)
The feasibility of determining the redox requires the potential of indicator electrode
potential depends strongly on the solvent, to be determined by the potential of the re-
the electrolyte, and the electrode material: dox pair under study. Sometimes this is
all of them should remain inert in po- impossible, because of the low exchange
tential ranges as wide as possible. In this current densities and/or concentrations
connection, the term ‘‘potential window’’ (particularly, at low solubilities or limited
1.12 Effects of Ion Pairing 17
stabilities of the components), when extra- chemistry and the activity coefficients. The
neous redox pairs contribute significantly latter task covers a wide field of pH-metric
to the measured potential. The sensitivity applications, and also the analytical tech-
of direct potentiometry can be enhanced by niques based on ion-selective electrodes.
using preconcentration procedures, partic-
ularly in polymeric films on the electrode
surfaces [65]. However, it is common to 1.12
use voltammetry, polarography, and other Effects of Ion Pairing
related techniques in place of potentiom-
etry. The formal potential (determined as An important factor affecting redox po-
the potential between anodic and cathodic tentials is also ion pairing, which is
current peaks on voltammograms) does conventionally taken into account in terms
not generally coincide with the equilib- of activity coefficients. For redox potentials
rium redox potential, the accuracy depend- of coordination compounds, the following
ing on the reaction mechanism [66–68]. equation is known [70]:
For reversible and simplest quasireversible
electrode reactions, the exact determina- RT Kred
tion of standard potentials from the data of E = E0 + ln (23)
F Kox
stationary voltammetry and polarography
is possible. Only these dynamic techniques
where Kred and Kox denote the equilib-
are suitable for the studies of equilibria
rium constants of a complex (particularly,
with participation of long-living radicals
ion pair) formation. In as much as outer
and other excited species, which can be
sphere ionic association is highly sensitive
introduced into solution only by in situ
to the ionic charge, Kred appears to be
electrochemical generating.
higher than Kox , and the potential shifts
If voltammetric and related techniques
to more positive values with the ionic
are used, the ohmic drop should be either
compensated (now this is usually done by strength. That is, for usual concentrations
the software or hardware of electrochem- of hexacyanoferrates in aqueous solutions,
ical devices [69]), or reduced by using, for the shift induced by the association with
example, a Luggin (Luggin-Gaber) capil- K+ ions reaches 0.1 V and higher at usual
lary (see in Ref. [12]). Another important concentrations.
technical detail is that the components In electroanalysis, Eq. (23) forms the
of reference redox systems (such as fer- basis of highly sensitive techniques
rocene/ferrocenium) are frequently added of potentiometric titration. The practi-
immediately into the working compart- cal applications are presented by numer-
ment when voltammetry-like techniques ous potentiometric sensors, particularly
are applied. biosensors [71–74]. The potential mea-
Applications of potentiometry are rather surements in microheterogeneous media
widespread, and its efficiency is high (microemulsions; anionic, cationic and
enough when operating with relative po- nonionic micelles) were worked out re-
tential values. A mention should be made, cently [75]. A separate field is the potentio-
first of all, of the determination of ba- metric studies of liquid/liquid junctions
sic thermodynamic quantities, such as directed to the determination of ionic
the equilibrium constants for coordination transport numbers.
18 1 Electrode potentials
(chem)
The value of Ghydr cannot be mea- schematically illustrates a ‘‘metal box’’
sured directly; however, it can be estimated corresponding to the jellium model.
According to this figure,
on the basis of information on the liq-
uid structure of water and the structure We = V − εF = e0 χ M + Vex − εF (27)
(chem)
of its molecule. Having Ghydr and
(real) the notations are given in the figure
Ghydr , we can find χ H2 O from Eq. (24).
caption. In the framework of the jellium
The estimates of this value given in the
model, we can calculate χ M by two
literature for water/air interface disagree techniques: directly from the distribution
with one another significantly (from −0.48 of the electronic gas outside the metal, and
up to +0.29 V). A critical comparative con- also by calculating Vex and εF with the
sideration of various studies leads to a subsequent use of Eq. (26). Unfortunately,
conclusion about a low positive value of the jellium model has a serious limitation;
χ H2 O , namely, 0.13 V [1–9]. This value the estimates of χ M will be improved in
corresponds to the infinite dilution and future as the theory of electronic structure
can be changed by changing the solution of metals and their surfaces becomes more
concentration. advanced.
The estimation of χ M is an even As was correctly reasoned in Ref. [10,
more complicated problem. Figure 3 11], the scale of ‘‘absolute potential’’
electron in vacuum
(at infinity)
−e0y M mM
e /NA
electron in vacuum
(near the metal surface)
Fermi level
We
V = Vex + Vel
eF
Fig. 3 A scheme of electron energy levels for a model of ‘‘metal box’’ (according to
C. Kittel, Elementary Solid State Physics, Wiley, New York, 1962): µ̄M
e – electrochemical
potential of electron in metal, NA – Avogadro number, εF – Fermi energy, ψ M – outer
potential of metal, We – electron work function, V – electron potential energy in metal
(V = Vex + Vel , Vex – energy of electron exchange, which corresponds to the electron
interaction with positively charged jellium, Vel = e0 χ M – the surface component of
electron potential energy).
20 1 Electrode potentials
constructed in accordance with Ref. [7] The ‘‘absolute potentials’’ of the hydro-
and recommended in Ref. [9] is, in fact, gen electrode in a number of nonaqueous
equivalent to the EK scale first proposed solvents are reported in Ref. [9].
in Refs. [76–81], which should be named The value of EK can be expressed via
after Kanevskii: metal/vacuum electron work function and
Volta potential [10, 11]:
(real)
[Gsubl + Gion + Ghydr ]
EK = − EK = We + M
S ψ (31)
zi F
(28) The majority of electrochemical prob-
By introducing this scale, we can sep- lems can be solved without separating
arate the total emf of the cell into two the emf into ‘‘absolute potentials’’. How-
quantities, for which the contributions ever, it should be mentioned that the
(real)
Gsubl , Gion , and Ghydr can be ex- problem concerning the structure of the
perimentally determined. For a hydrogen interfacial potential drop becomes the top-
electrode, ical problem for the modern studies of
electrified interfaces on the microscopic
(EK )H+ ,H2 level, particularly, in attempts of test-
ing the electrified interfaces by probe
1 (real)
2 Gdiss + Gion + Ghydr techniques [92–94]. The absolute scales
=− are also of interest for electrochemistry
F
(29) of semiconductors in the context of cal-
which gives the possibility to recommend ibrating the energy levels of materials.
the value −4.44 ± 0.02 V for the ‘‘abso- This problem is related to another general
lute potential’’ of hydrogen electrode (at problem of physical chemistry – the de-
298.15 K). For other temperatures, the cal- termination of activity coefficient of an
culations of this sort are limited by the individual charged species.
lack of information on the temperature
coefficient of work function. It should
be mentioned that the uncertainty of 1.14
absolute potential calculation is substan- The Role of Electric Double Layers
tially higher than the accuracy of direct
potential measurements with respect to Later, a more detailed consideration of the
a reference electrode. The useful com- M|S interface is given, which takes into
ments to Kanevskii’s scale can be found account the formation of electric double
in Refs. [16–20, 84]. In the ‘‘absolute’’ layer. Figure 1 shows the potential drops
scale, the ‘‘absolute potential’’ of any at the metal/solution interface. It follows
aqueous electrode at 298.15 K can be de- from this figure that
termined as
M S M
S ψ = χ + S ϕ − χ
M
(32)
Electrochemical Physical
scale scale
[V] [eV]
(Absolute potential scale) = −(Physical scale)
E(abs) / V = E(SHE) / V + 4.44
− 4.44 0 electrons at rest
in a vacuum
−4
−1
−2 −2.07
E 0(Mg2+/Mg) −2.37
−2
−1
E 0(Zn2+/Zn) − 0.76 −3.68
Potential of zero charge of gallium − 0.69 −3.75
−4
Potential of zero charge of mercury − 0.19 − 4.25
Standard hydrogen electrode (SHE) 0 − 4.44
Saturated calomel electrode (SCE) 0.24 − 4.68
−5
E 0(Ag+/Ag) 0.80 −5.24
1
Standard oxygen electrode (aH+ = 1) 1.23 −5.67
–6
E 0(MnO 4−/MnO2) 1.70 − 6.14
1
2
0.6
In Tl
In
0.4
WeHg − WeMe
Cd Cd
[eV]
Pb
Bi Bi
0.2
Sn Ga Ga
Sn
0
Sb
[V]
Fig. 5 Correlation of work function differences and the differences of
potentials at constant electrode charge σ for σ = 0 (1) and
−18 µC cm−2 (2), as reported in Ref. [5].
By writing such relationships for two Eq. (36) and (37) obtained by Frum-
interfaces, M1 |S and M2 |S and substitut- kin [1–5] can be classified as the solution
ing corresponding M 1 M2
S ψ and S ψ into of the famous Volta problem of the nature
Eq. (10), we obtain of emf of electrochemical circuit. Equa-
tion (37) demonstrates that the difference
E = M
M1 ψ + (ϕ1 − ϕ2 ) of potentials of zero charge (pzc) of two
metals is approximately equal to their Volta
+ (δχ1M − δχ2M + δχ2S − δχ1S ) (35)
potential. In as much as the Volta poten-
If the metals M1 and M2 interact very tial M 2
M1 ψ is equal to the difference of
weakly with the solvent molecules, we can work functions of the metals (WeM2 and
assume that, in the first approximation, WeM1 ), the verification of Eq. (37) can be
the last right-hand term in parentheses is carried out by comparing the differences
zero, and both in pzc and in work functions (Fig. 5).
The data of Fig. 5 confirm Eq. (37) and, si-
E∼
= M
M1 ψ + (ϕ1 − ϕ2 ) (36) multaneously, demonstrate the possibility
of substantial deviations from this rela-
When both metals, M1 and M2 , are tionship for the systems with pronounced
zero charged (their excess surface charge solvent/metal interaction (as, for example,
density σM = 0), and the ionic double for the Ga/water interface).
layers are absent on their surfaces, In the context of concept of zero charge
potential [1–5], a specific scale of electrode
EσM =0 ≈ M2
M1 ψ (37) potentials should be mentioned, a reduced
1.15 Conclusion 23
28. P. P. Edwards in Advances in Inorganic Chem- 51. H. Strehlow, The Chemistry of Non-Aqueous
istry and Radiochemistry (Eds.: H. J. Emeleus, Solvents, Academic Press, 1967, pp. 129–171,
A. G. Sharpe), Academic Press, New York, Vol. 1.
1982, pp. 135–185, Vol. 25. 52. J. N. Butler in Advances in Electrochem-
29. M. Plank, Ann. Physik. 1890, 40, 561–576. istry and Electrochemical Engineering (Ed.:
30. P. Henderson, Z. Phys. Chem. 1907 59, P. Delahay), Intersci. Publi., New York, 1970,
118–127. pp. 77–175, Vol. 7.
31. P. Henderson, 1908, 63, 325–345. 53. G. Gritzner, J. Kuta, Pure Appl. Chem. 1984,
32. B. B. Damaskin, G. A. Tsirlina, M. I. Borz- 56, 461–466.
enko, Russ. J. Electrochem. 1998, 34, 199–203. 54. Y. Marcus, Pure Appl. Chem. 1986, 58,
33. J. B. Chloupek, V. Z. Danes, B. A. Danesova, 1721–1736.
Collect. Czech. Chem. Commun. 1933, 5, 55. A. N. Frumkin, Zh. Fiz. Khim. 1956, 30,
469–478. 2066–2069.
34. J. B. Chloupek, V. Z. Danes, B. A. Danesova, 56. D. M. Stanbury in Advances in Inorganic
Collect. Czech. Chem. Commun. 1933, 5, Chemistry (Ed.: A.G. Sykes), Academic Press,
527–534. INC, New York, 1989, pp. 69–138, Vol. 33.
35. N. Q. Minh, L. Redey, Molten Salt Techniques, 57. D. D. Wayner, V. D. Parker, Acc. Chem. Res.
Plenum Press, New York, 1987, Vol. 3. 1993, 26, 287–294.
36. D. J. G. Ives, G. J. Janz, (Eds.), Reference Elec- 58. G. A. Tsirlina, O. V. Safonova, O. A. Petrii,
trodes. Theory and Practice Academic Press, Electrochim. Acta 1997, 42, 2943–2946.
New York, London, 1961. 59. J. Chatt, Coord. Chem. Rev. 1982, 43,
37. F. G. Donnan, Z. Elektrochem. 1911, 17, 337–347.
572–581. 60. A. A. Vlcek, Prog. Inorg. Chem. 1963, 5,
38. F. G. Donnan, E. A. Guggenheim, Z. Phys. 211–384.
Chem. 1932, 162A, 346–360. 61. F. Sanchez, P. Perez-Tejeda, F. Perez et al.,
39. W. E. Van der Linden in Comprehensive An- J. Chem. Soc., Dalton Trans. 1999, N 17,
alytical Chemistry. (Ed.: G. Svehla), Elsevier, 3035–3039.
Amsterdam, 1981, Vol. 1. 62. E. Garcia, A. Bard, J. Electrochem. Soc. 1990,
40. J. Vesely, D. Weiss, K. Stulik, Analysis with 137, 2752–2759.
Ion-Selective Electrodes, Ellis Horwood, Chich- 63. C. L. Hussey in Chemistry of Nonaqueous
ester, 1978. Solutions (Eds.: G. Mamantov, A. I. Popov),
41. G. Eisenman, Biophys. J. 1962, 2, 259–323. VCH, New York, 1994, pp. 227–275.
42. J. S. Newman, Electrochemical Systems, 2nd 64. K. Xu, N. D. Day, C. A. Angell, J. Electrochem.
ed., Prentice-Hall, Englewood Cliffs, 1991. Soc. 1996, 143, L209–L211.
43. R. Parsons, Redox Potentials in Aqueous 65. J. Huang, M. S. Wrighton, Anal. Chem. 1993,
Solutions: A Selective and Critical Source Book, 65, 2740–2746.
Marcel Dekker, New York, 1985, pp. 1–34. 66. A. J. Bard, L. R. Faulkner, Electrochemical
44. A. J. Bard, R. Parsons, J. Jordan, Standard Methods. Fundamentals and Applications, Wi-
Potentials in Aqueous Solution, Marcel Dekker, ley, New York, 2001.
New York, 1985. 67. D. K. Gosser, Cyclic Voltammetry: Simulation
45. M. S. Antelman, F. J. Harris, (Eds.), The En- and Analysis of Reaction Mechanisms, VCH
cyclopedia of Chemical Electrode Potentials, Publ., Deerfield Beach, 1993.
Plenum Press, New York, London, 1982. 68. J. F. Cassidy, Compr. Anal. Chem. 1992, 27,
46. M. Pourbaix, Atlas d’Equilibres Electrochemi- 1–69.
ques, Gauthier-Villars, Paris, 1963. 69. R. M. Souto, Electroanalysis 1994, 6, 531–542.
47. S. G. Bratsch, J. Phys. Chem. Ref. Data 1989, 70. J. Heyrovsky, J. Kuta, Zaklady Polarografie,
18, 1–21. Nakl. Ceskoslov. Academ. Ved, Praha, 1962.
48. K. Eklund, S. N. Lvov, D. D. Macdonald, J. 71. J. Janata, Chem. Rev. 1990, 90, 691–703.
Electroanal. Chem. 1997, 437, 99–110. 72. J. F. Coetzee, B. K. Deshmukh, C.-C. Liao,
49. J. T. Hupp, Inorg. Chem. 1990, 29, Chem. Rev. 1990, 90, 827–835.
5010–5120. 73. M. Alvarez-Icaza, U. Bilitowski, Anal. Chem.
50. J. K. Bashkin, P. J. Kinlen, Inorg. Chem. 1993, 65, 525A–533A.
1990, 29, 4507–4509. 74. J. Wang, Anal. Chem. 1995, 67, 487R–492R.
1.15 Conclusion 25
75. S. A. Myers, R. A. Mackey, R. A. Brajter- 85. H. Reiss, A. Heller, J. Phys. Chem. 1985, 89,
Toth, Anal. Chem. 1993, 65, 3447–3453. 4207–4213.
76. E. A. Kanevskii, Zh. Fiz. Khim. 1948, 22, 86. Yu. V. Pleskov, J. Phys. Chem. 1987, 91,
1397–1404. 1691–1692.
77. E. A. Kanevskii, Zh. Fiz. Khim. 1950, 24, 87. A. Frumkin, Z. Phys. Chem. 1924, 111,
1511–1514. 190–210.
78. E. A. Kanevskii, Zh. Fiz. Khim. 1951, 25, 88. A. Frumkin, Z. Phys. Chem. 1924, 109, 34–48.
854–862. 89. A. N. Frumkin, Z. A. Iofa, M. A. Gerovich,
79. E. A. Kanevskii, Zh. Fiz. Khim. 1952, 26, Zh. Fiz. Khim. 1956, 30, 1455–1468.
633–641. 90. J. Randles, D. Schiffrin, J. Electroanal. Chem.
80. E. A. Kanevskii, Zh. Fiz. Khim. 1953, 27, 1965, 10, 480–484.
296–309. 91. J. Randles, Trans. Faraday Soc. 1956, 52,
81. E. A. Kanevskii, Dokl. Akad. Nauk SSSR 1573–1581.
1981, 257, 926–929. 92. A. C. Hillier, S. Kim, A. J. Bard, J. Phys.
82. V. A. Pleskov, B. V. Ershler, Zh. Fiz. Khim. Chem. 1996, 100, 18808–18817.
1949, 23, 101–103. 93. R. C. Thomas, P. Tangyunyong, J. E. Hous-
83. B. V. Ershler, V. A. Pleskov, Zh. Fiz. Khim. ton et al., J. Phys. Chem. 1994, 98, 4493,
1951, 25, 1258–1260. 4494.
84. V. M. Novakovskii, Zh. Fiz. Khim. 1956, 30, 94. Z.-X. Xie, D. M. Kolb, J. Electroanal. Chem.
2820–2822. 2000, 481, 177–182.
33
was not available in surface science, and methods, the extensive research area of
thus physicists started to immerse solid semiconductor/electrolyte and metal/SE
surfaces into the electrolyte environment interfaces will not be considered here.
in order to probe the interface response to We will start with phenomenology, and
charging. then move, where possible, to a micro-
Distribution of potential near a charged scopic theory and simulations. Without a
interface is crucial in electrode kinetics, solid phenomenological basis one may risk
since the potential drop plays the role of calculating things from the scratch many
a variable driving force of reaction. The times, whereas the calculation of one pa-
reactions take place at the interface and rameter of the phenomenological theory
with participation of adsorbed reactants. might be all that is actually needed from a
However, not only reactants but other microscopic model! On the other hand, the
species as well can adsorb at the interface. phenomenology often receives its justifica-
Competitive adsorption of these species, tion and validity criteria from microscopic
ions or dipoles can play, depending on the theories.
situation, a catalyzing or inhibiting role on
the reaction rates. 2.1.2
On the whole, understanding the struc- Basic Concepts
ture of the interface and its response to
charging is most essential for electrochem- Electrified interfaces represent the princi-
istry and its implications for colloid science pal object of ‘‘interfacial electrochemistry’’.
and biophysics. Various aspects of this area are described
The electrochemical interface is com- in monographs and textbooks on this
posed of molecules (solvent, adsorbed subject [1–11]. A brief outline of this in-
molecular species) and ions (of electrolyte), formation is given in the following text.
which can be partially discharged when Electrified interfaces represent a spe-
chemisorbed, electrons and skeleton ions cific quasi two-dimensional object in the
in the case of metal electrodes, electrons vicinity of the geometrical boundary of
and holes in the case of semiconductor two phases containing, in general, mo-
electrodes, mobile conducting and im- bile electronic and/or ionic charges, and,
mobile skeleton ions in SEs. Molecules correspondingly, an electric-field distribu-
and ions are classical objects but elec- tion generated by these charges. Practically
trons, holes with small effective mass, at least one of these media in contact is
and protons are quantum objects. Inter- conducting, while the second one may
action between molecules and surfaces be either a conductor of another kind
is quantum-mechanical in nature in the (metal–electrolyte interface, semiconduc-
case of chemisorption. Thus, microscopic tor–electrolyte interface), or of the same
description of the interface requires a kind (interface of two immiscible elec-
combination of quantum and classical trolytes), or an insulator. As important
methods. One can benefit, however, from exceptions, one can point out pure sol-
simple or more involved phenomenologi- vent/air or pure solvent/insulator (e.g.
cal descriptions of the interface. silica-water) boundaries where a signifi-
In this chapter we will focus on charge- cant potential drop may be created by the
able interfaces: metal–LE interface and formation of a dipole layer, without the
ITIES. Although similar in concepts and generation of a space-charge region.
2.1 Electrochemical Interfaces: At the Border Line 35
The simplest example is given by the acquires an extra electronic charge, while
free metal surface. Owing to the Pauli the adjacent layer of another phase con-
principle the metal electrons at the highest tains an excessive positive charge formed
occupation levels possess a significant by the uncompensated ionic skeleton. The
kinetic energy so that their density extends thickness of each charged layer depends
outside the ionic skeleton of the electrode. on the screening properties of the corre-
It results in the formation of a structure sponding medium (their Thomas–Fermi
characteristic for numerous electrified or Debye screening lengths) so that the
interfaces; two oppositely charged spatial whole space-charge region is very thin for
regions (with zero overall charge), this time a contact of two metals (within 0.1 nm),
a positively charged region inside the ion while it may be much more extended in
skeleton due to the depletion of electrons the case of semiconductors, depending on
and a negatively charged region formed their bulk properties.
by ‘‘the electronic tail’’. Because of this In all cases such interfaces are non-
‘‘nanocondenser’’, the bulk metal has got polarizable, since the only way to change
a high positive potential of several Volts the interfacial potential difference is to
with respect to the vacuum. This manifests modify the bulk properties of the phase(s),
itself in the value of the electron work for example, the degree of doping or
function the work needed to withdraw stoichiometry for semiconductors.
an electron from the metal across its Analogous ionic systems are represented
uncharged surface. by a contact of two electrolyte solutions in
A more complicated example is provided immiscible liquids, which contain a com-
by the contact of two electronic conductors, mon ion. Owing to its interfacial exchange
metals or semiconductors. At equilibrium the electrochemical potential of this ion is
the electrochemical potentials of electrons to be constant throughout the system, and
(i.e. their Fermi energies) must be equal it leads again to a relation for the interfa-
in both bulk phases. The Galvani potential cial potential difference similar to Eq. (1)
difference between them is determined by (called Donnan equation this time), that is,
the bulk properties of the media in contact, the interface is nonpolarizable without a
i.e. by the difference between the chemical change of the bulk media properties.
potentials of electrons: The same process of interfacial ion ex-
change determines the potential difference
e − µe
µm1 m2
ϕ m1/m2 = (1) for other ion conductors, like membranes
e or SEs. The existence of this potential dif-
where e is the elementary charge, since ference again leads to the formation of the
the right-hand-side terms are constant, one EDL, but this time it is formed by exces-
cannot change the interfacial Galvani po- sive ionic charges in the surface layers of
tential in equilibrium conditions without both solutions. Since this EDL is created
changing the chemical composition of the by the ion transfer across the interface,
phases. The difference between the bulk without charge transport across the bulk
potentials required by Eq. (1) has to be es- media, the Donnan potential drop corre-
tablished by the formation of an ‘‘electrical sponding to the composition of the phases
double layer’’ (EDL) owing to the electron is established very rapidly.
exchange between the surface layers of the Similar to the contact of two elec-
phases. Finally, one of the surface layers tronic conductors, the imposition of a
36 2 Electrochemical Double Layers
when the electronic and ionic charges of components of the solution. As a result
the EDL are separated by a layer of the the EDL includes this ‘‘fixed’’ charge at the
solvent molecules, as it is in the case of a surface and the counterions in the solution
‘‘surface-inactive electrolyte’’ discussed in part. For a further discussion, see Ref. [14].
the following text. The specific adsorption Such charged interfaces are typical for
of the ionic-solution components, that is, various ionic or polar solids, for ion
their significant accumulation in the im- exchange, lipid or biological membranes,
mediate contact with the metal surface, or for any insulating surface in the
is usually accompanied by the strong re- presence of amphiphilic solute species.
distribution of the electronic charge near Even though one cannot polarize this
each adsorbed species. As a result they interface from the external source, the
may possess a much lower charge, com- potential difference and the separated
pared to the one in the bulk solution charge can change as a function of the
(it is termed as adsorption with a par- solution composition, in particular its pH
tial or complete charge transfer). Under and ionic strength.
these conditions the charge dQ supplied In many cases the solvent molecules
via the external circuit is not equal to the and solutes penetrate into the surface
change of the ‘‘free electronic charge’’ dis- layer of the insulator, thus forming a ‘‘gel
tributed along the surface of the metal. layer’’ composed of the components of
Such ‘‘perfectly polarizable electrodes’’ re- both phases. Its thickness is sometimes
quire a more elaborated description; see significant. Then the charge distribution
Chapter 3.1 of this book [13]. across such interphase is quite different
At approaching the limits of the ideal- from the simple ‘‘condenser’’ idealization,
polarizability interval, the rate of faradaic in particular the overall potential differ-
processes at the interface increases rapidly ence between the bulk phases is composed
(discharge of the solvent or solute compo- of two contributions for the insulator–gel
nents), with a dc current passing through layer and gel layer–solution interfaces,
the system. with a plateau between them, if the gel
The existence of this ‘‘ideal-polariz- layer is sufficiently thick [15].
ability’’ range of potentials is of great This two-step profile of the potential
practical importance, since it allows one across the interphase is also characteris-
to study an individual faradaic process of tic for certain types of modified electrodes in
the added electroactive species in the pres- which the metal surface is coated with a
ence of the background electrolyte, whose film, whose thickness significantly exceeds
presence provides important advantages the atomic scale. Such systems represent
(diminished Ohmic potential drop, rapid a much more complicated type of electri-
charging of the EDL, etc). fied interfaces, since the distribution of
A specific type of the electrified interface charged species depends crucially on the
is given by the contact of an insulating specific properties of the film. In most
phase with electrolyte solutions. Since the cases of sufficiently thick films, the pro-
charged species cannot cross the insulator, file of the Galvani potential across the
the EDL formation originates from the ion- interphase possesses a plateau inside the
ization process of surface groups (most fre- bulk film separating two potential drops
quently, the proton-based dissociation or at its interfaces with the electrode and the
association) or/and the adsorption of ionic solution [16].
38 2 Electrochemical Double Layers
high (less than or about a mole per liter), Interfacial thermodynamics has become
the Gibbs plane is sufficiently close to the one of the bases of the theory of the electro-
physical boundary between the metal and chemical phenomena determined by the
the adjacent solvent layer. Then the above charge transport, in particular in mod-
negative values of ion Gibbs adsorptions at eling the electrochemical impedance [25,
the p.z.c. and the linear slope in Eq. (11) 28]. The analysis of the results obtained
points to the existence of an ion-free layer within the framework of this approach
of the solvent near the metal surface, that has shown that for transport processes
is, a compact layer, the existence of which under usual electrochemical conditions,
has been postulated long before these the characteristic timescales of the interfa-
experiments. The factor in this relation, cial charging and of the diffusion process
zH , corresponds to its thickness, that is, are well separated, and one can use a
to the distance of the closest approach of simplified treatment in which the inter-
the centers of nonadsorbing ionic species facial charging effects are characterized
to the metal surface, compared with the by a single parameter, C. In particular,
similar quantity for the solvent (Gibbs the relaxation after a potential step or the
adsorption plane). impedance in the proper frequency range
Surprisingly this important experimen- may be represented as an in-series com-
tal information on the properties of the bination of the solution resistance and
compact layer has been ignored in mod- the interfacial capacitance. On the other
ern microscopic theories of the interfacial hand, the interfacial contribution into the
structure, while the thickness of the ion- impedance arising from the perturbation
free layer at the electrode surface could be of the electrolyte concentration outside of
calculated on the basis of the distribution the EDL becomes essential for thin-layer
functions for the solvent and ions. systems in which the characteristic fre-
Effects of the interfacial charging are of quencies are overlapping or insufficiently
importance in various kinetic phenomena, separated.
in particular in the ion transport. The
simplified treatment assumes that these 2.1.4
effects can be adequately described by Classical Models of Electrified Interfaces
a single parameter, the capacitance C.
However, the passage of a nonstationary 2.1.4.1 Helmholtz Model
current is accompanied by the time- The thermodynamic treatment does not
variation of the electrolyte concentration allow one to extract more detailed infor-
outside the EDL region, and it must mation on the interfacial structure. The
give an additional contribution to the already discussed experimental results for
change of the interfacial potential, Eq. (8a). the ion-free solvent layer may be consid-
Besides, there is generally an imbalance ered as an illustration of the simplest de-
between the partial ion fluxes inside the scription of the interface in the Helmholtz
diffusion layer and the time derivatives of model. The latter postulates two charged
the corresponding partial ionic charges, planes containing the extra electrons and
σ± , Eq. (10). These factors result in the the counterions, which are separated by
dependence of the nonstationary response a solvent (‘‘compact’’) layer having thick-
of the system on all three interfacial ness zH , that is, the system represents
parameters, C, t+ m/s and Cµ . a nanocondenser. Then the ratio of the
42 2 Electrochemical Double Layers
that the compact (Helmholtz)-layer capac- layer determines the shape of the capac-
itance, CH , is independent of the concen- itance curves in dilute solutions near the
tration of the ‘‘surface-inactive electrolyte’’ p.z.c. where the theory predicts a pro-
(which does not penetrate noticeably into found minimum whose depth increases
the compact layer) but avoids to accept a with diminishing the electrolyte concen-
model hypothesis on its dependence on tration. The minimum is usually well
the electrode charge, CH (σ ), that is, the developed to about 0.01 M concentrations
latter is to be determined from experimen- while in systems with a high hump of
tal data. On the contrary, the properties of CH (σ ) near σ = 0 (such as for Ag or
the diffuse layer are completely defined by Au single-crystal faces) a deep diffuse-
Eqs. (17) and (21a), without any unknown layer minimum should already appear
parameter. at 0.1 M. All these qualitative predictions
The ratio of the two contributions in of the Gouy–Chapman–Stern–Grahame
Eq. (20) depends on the values of the (GCSG) theory are in line with exten-
electrode charge and the electrolyte con- sive experimental information for numer-
centration, as well as on the particular type ous metal–solution interfaces for surface-
of the media in contact. The diffuse-layer inactive electrolytes (i.e. for systems with-
capacitance, Eqs. (21a) and (21b), is grow- out specific adsorption of solute species,
ing with the increase of the absolute value see following text) [22, 33, 34].
of the charge and with the concentration.
Besides, it becomes almost independent of 2.1.5
the concentration far from p.z.c. Near this Diffuse and Compact Layer Properties
point, σ = 0, the diffuse-layer capacitance
shows a minimum, which is becoming 2.1.5.1 Grahame Model in Comparison
progressively deeper at diluting the solu- with Capacitance Data for Liquid Electrodes
tion: CGC (0, c) = 228 µF cm−2 for c = 1 M Let us now discuss how to verify this
but only 7.2 µF cm−2 for 0.001 M aque- theory quantitatively as well as to extract
ous solution of a 1,−1 electrolyte. The the information on the interfacial structure
compact-layer capacitance, CH , is within with the use of the data for the Hg/H2 O-
20–60 µF cm−2 for most ‘‘mercury-like’’ NaF system [35, 36] as an illustration. The
metals (Hg, Pb, In, Cd, Bi, Sb, etc) in measured capacitance curves for several
aqueous and nonaqueous solutions, with electrolyte concentrations are presented in
much greater values for some other sys- Fig. 1. One can notice the expected diffuse-
tems, such as Ag and Au single-crystal layer minimum for dilute solutions but the
faces, up to 100–150 µF cm−2 near p.z.c. whole shape of the curves is sufficiently
As a result, for sufficiently concentrated complicated, in particular within the range
solutions, about 1 M or higher, the diffuse- of negative potentials.
layer contribution in Eq. (20) represents The streaming electrode measurements
in most cases only a minor correction to show a constancy of the p.z.c. value, Eσ =0 ,
the compact-layer term. Then, one can for different concentrations, which is con-
even use the simple Helmholtz model sidered as the necessary condition for the
for qualitative interpretation of the data. absence of a specific ionic adsorption.
In particular, far from the p.z.c., this In this case one can obtain the charge-
approximation is acceptable for all con- potential relations for all concentrations
centrations. On the contrary, the diffuse by the integration of the corresponding
2.1 Electrochemical Interfaces: At the Border Line 45
30
25
[µF cm−2]
C
1
20 2
3
4
15
10
0.3 0 −0.5 −1.0 −1.5
6 E
[V]
Fig. 1 Experimental capacitance curves for the mercury electrode in
aqueous solutions of NaF. Concentrations: 0.916 M (1), 0.1 M (2),
0.01 M (3), 0.001 M (4) [35, 36].
capacitance curve, that is, one plots the on the electrode charge and it may be
capacitance versus the charge for all con- even greater for millimolar solutions). The
centrations as in Fig. 2. These curves retain corresponding ranges of the dispersion are
the diffuse-layer minimum at low concen- shown in Fig. 2 for two dilute solutions.
trations but their behavior at high electrode Even though the initial dispersion in Fig. 1
charges has been strongly simplified, in was independent of the potential, the
particular all curves are practically identi- errors in Fig. 2 turned out to be markedly
cal at high negative charges. varying as a function of the electrode
Any justified comparison of measured charge. They are close to those of the
data with theoretical predictions must take corresponding C(E) curves at the zero
into account the experimental accuracy. charge and near the cathodic minimum.
However, the resulting error bars appear Then the errors are slightly enhanced at
to depend strongly on the choice of the dropping part of the cathodic branch,
the coordinate system, used for this but they are several times smaller at lower
treatment. In order to demonstrate the electrode charges, where the capacitance
latter, let us assume for simplicity that varies very rapidly. The latter effect is due
the measured capacitance data in Fig. 1 to a correlation of errors in the capacitance
have got the precision of ±0.1 µF cm−2 for and the charge within this interval. It
two highest concentrations, ±0.2 µF cm−2 implies that even a considerable deviation
for the 0.01 M solution, and ±0.5 µF cm−2 between the theoretical prediction and
for 0.001 M, within the whole potential experimental data in the initial C(E)
interval (in real measurements at low curves within this potential range may look
frequencies the dispersion also depends rather small in the coordinates of Fig. 2.
46 2 Electrochemical Double Layers
C
[µF cm−2]
30
1
2
3
4
20
10
If one relies on the validity of the GC inside the experimental error dispersion
model for the diffuse-layer properties, one over the whole range of the electrode
can use Eq. (20) to calculate the values of charges. This comparison enables one
the compact-layer capacitance, CH , from to conclude that the deviation of the
experimental data for C and theoretical data for the highest concentration at
formula (21a) for CGC . This operation high positive charges is probably due to
performed for all available values of the the violation of some assumption of the
charge and the electrolyte concentration model, for example, dissolution of mercury
results in Fig. 3, in which the effect of the or specific adsorption of cations. The
experimental dispersion is shown again concentration dependence of the diffuse-
for two dilute solutions. layer capacitance is probably too weak to
According to Grahame’s hypothesis, explain this divergence, see in following
these curves for different concentrations text, even if the GC theory may be
should be coincident. One can notice inapplicable quantitatively.
first of all its brilliant confirmation in The curve for 0.01 M also demonstrates
Fig. 3 for the two highest concentrations; a marked increase of the experimental
a deviation is only seen at high positive dispersion near the zero charge. This
charges, above 10 µC cm−2 . The proximity effect is even more pronounced for the
of the compact-layer capacitance curves 0.001 M solution in which the dispersion
is even more impressive for the 0.1 M does not exceed ±0.8 µF cm−2 , except for
and 0.01 M solutions, whose difference is the range of very small charges where the
2.1 Electrochemical Interfaces: At the Border Line 47
CH
[µF cm−2]
35
25
20
CH
[µF cm−2]
64
35
30
25
20
charge, one can interpolate its values for theory predicts a linear dependence with a
positive and negative charges, to obtain unit slope for each value of the electrode
CH (0) for a dilute solution. Such a charge. Such plots for the same experi-
procedure in Fig. 3 results in the same mental data are shown in Fig. 4, again
value, about 29 µF cm−2 , for this quantity with indication of the error dispersion
as for the other concentrations. of each point. All lines in the figure
The curve for the 0.001 M solution de- have the theoretical slope. Their extrap-
viates downward noticeably for moderate olation to the ordinate axis gives the value
positive and negative charges. The most of CH −1 for the corresponding electrode
probable reason is the increase of exper- charge. The PZ method to check the va-
imental errors for such dilute solution lidity of the Grahame model is more
related to the current passage across the objective (since it suggests verifying the
solution. linearity of the plots and their slope), es-
An alternative and the most commonly pecially if the experimental dispersion is
used way to verify the Grahame model is not taken into consideration. However, in
based on Parsons–Zobel (PZ) plots [37], most publications (especially for solid elec-
in which the inverse values of the exper- trodes) this method has been used in a
imentally measured capacitance, C −1 , are reduced form, where only the data for
shown as a function of the inverse diffuse- the p.z.c. are plotted, thus giving only
layer capacitance given by Eq. (21a). The CH (0). It means that the Grahame theory is
1 2 4
0.07
0.15
1
C −1
C −1
0.10 0.05 5
2
6
0.05 3 0.03
δ C −1
[µF cm−2]
14
12
10
3 2
2
1
10 8 6 4 2 0 −2 −4 −6 −8
σ
[µC cm−2]
Fig. 5 Experimental and theoretical δC −1 versus σ curves for mercury/aqueous NaF
solution [38], see Fig. 2, and the definitions in Eqs. (23a) and (23b). Error bars for the
experimental curves correspond to those in Fig. 2.
immediately calculated from the theo- theories, no attempts have been made to
retically derived ion concentration pro- subject them to the high precision ex-
files. Although drastic deviations from the perimental verification in accordance with
GC model have been predicted in these Eq. (23b).
2.1 Electrochemical Interfaces: At the Border Line 51
One should avoid, however, the mis- range of its applicability to higher ionic
take of interpreting the success of the GC strengths.
passage through this test as its complete
confirmation, in particular the proof of the 2.1.5.2 Compact Layer Properties
validity of Eq. (21a) for the diffuse-layer The exposed variant of the theory has
capacitance at high electrode charges. In been applied to the study of numerous
reality the theoretical lines in Fig. 5 based metal–solution interfaces, in particular for
on this formula demonstrate a weak de- liquid metals, Hg, Ga, and their alloys.
pendence of the diffuse-layer capacitance First of all for a particular metal/solvent
on the concentration (except for the p.z.c.), boundary one has to ensure ‘‘no surface
especially in dilute solutions (it is the ef- activity’’ of the solute components, that
fect, which is correctly reproduced by the is, their absence inside the compact layer.
model). However, the dominant part of the Experimentally the necessary condition of
diffuse-layer capacitance has been cancelled this surface inactivity of an ionic species
(together with the compact-layer contri- in the vicinity of the p.z.c. (which can
bution) by the subtraction in Eq. (23b). be measured directly for liquid electrodes)
Therefore the data in Fig. 5 cannot give is assumed to be the constancy of this
any information on the charge dependence potential, Eσ =0 , for different electrolyte
of the diffuse-layer capacitance far away concentrations. Then the data are com-
from the p.z.c. pared with the predictions of the Grahame
One has to keep in mind this impossi- model, as an additional check of the surface
bility to separate the compact-layer term inactivity.
and the concentration independent part of One should keep in mind that a de-
the diffuse-layer contribution in Eq. (20) tailed analysis has only been carried out for
at significant electrode charges. It results the Hg–H2 O-NaF interface (see previous
text), in which both postulates (concen-
in an uncertainty in the values of the
tration independence of the compact-layer
compact-layer capacitance, which is of pri-
properties and the GC model for the dif-
mary importance in the modeling of the
fuse layer) seem to be confirmed. For some
metal–solvent interfacial structure.
other systems, the Grahame treatment was
Recently the GC model has been mod-
applied, to determine the compact-layer ca-
ified to take approximately into account pacitance curve, CH (σ ), from the data for
the screening effects for ions inside the the highest electrolyte concentration and
diffuse layer [39, 40]. For this goal, the to calculate on its basis the ‘‘theoretical’’
electrolyte concentration in the equations C(E) curves for more dilute solutions, for
of the GC model (see previous text) were their comparison with experimental ones,
replaced by the electrolyte activity in the without taking into account the experimen-
bulk solution. This treatment can be justi- tal dispersion. Moreover, in most cases the
fied from the fundamental principles for a analysis was even reduced to constructing
low charge density inside the diffuse layer the PZ plot for the p.z.c. only. Generally the
(in the vicinity of the p.z.c.). On the other Grahame theory seems to provide at least a
hand, the ion screening in this layer at qualitative description of the results, even
high electrode charges is quite different though certain ‘‘anomalies’’ have been ob-
compared with the bulk solution. Never- served for some systems (e.g. for liquid
theless this modification may extend the gallium [41]), see [38] for review.
52 2 Electrochemical Double Layers
0.15
C −1
0.09
0.06 0.12
CGC−1
Fig. 6 Parsons–Zobel plot at the p.z.c. for mercury electrode in
contact with various aqueous electrolyte solutions [38]. Capaci-
tances are given in µF cm−2 . Experimental capacitance values:
NaF(◦) [35], NaH2 PO4 () [37] and Na2 SO4 (, ∇) [45, 46].
the faradaic current: from p.z.c., Eqs. (16), (13) and (17). Then
◦ the ratio of these potentials
ic = neSkc c(z∗ )
ϕ m/s ∼ CGC (0, c)
ne =1+ (26)
× exp −α(ϕ m/s − ψ1 ) ψ1 CH (0)
kB T
determined by the ratio of the diffuse and
◦ ◦
= neSkc c × exp −α(ϕ m/s − ψ1 ) compact-layer capacitances at the p.z.c. is
about 1 (or even close to 1), see Eq. (21a).
ne zi e The type of the effect on the polarization
− ψ1 (24)
kB T kB T curve depends also on the sign of the co-
efficient, zi − αn. In most cases this sign
Here, n is the number of transferred
is determined by the charge valence of the
electrons (n > 0), S the surface area, kc◦
reactant, zi . In the case of the electrore-
the cathodic rate constant, α the ca-
duction of a cation (e.g. for the hydrogen
thodic charge-transfer coefficient, and zi
evolution or a metal deposition) zi > 0,
the charge valence of the reactant. This
and the diffuse-layer factor Kψ1 changes
expression should be modified to take into
in the same direction as the Tafel term, exp
account the adsorption energies of the re-
(−αϕ m/s ne/kB T ). Therefore the former
actant and the product if the ET takes place leads to a deviation from the straight line
for a species inside the compact layer [11]. with the 2.3 kB T /αne slope in the Tafel
The value of the coordinate, z∗ , which coordinates, but the current-potential de-
determines the ψ1 potential is a priori un- pendence remains monotonous. Besides,
known. In most studies it was postulated zi and αn have opposite signs and for
that the ET took place at the species located single-charged cations, the coefficients of
at the outer Helmholtz plane (the bound- the diffuse-layer effect represents a rather
ary between the compact and diffuse lay- small correction to the Tafel dependence.
ers). Then, ψ1 coincides with the potential Nevertheless it cannot be ignored, both for
drop within the diffuse layer, ϕGC , Eq. (17). the quantitative treatment of the polariza-
Equation (24) contains an additional tion curves and for the interpretation of the
factor, dependence of the current on the concen-
tration of the indifferent electrolyte [53].
e
Kψ1 ≡ exp −(zi − αn)ψ1 (25) Quite a different situation takes place
kB T
for the reduction of anions, or for the
compared to the conventional Tafel for- oxidation of cations. Then the diffuse-
mula. In sufficiently concentrated solu- layer multiplier, Kψ1 , varies in the opposite
tions, for example, in the presence of direction compared to the Tafel term. For
the background electrolyte, the Frumkin sufficiently dilute solutions, the former
correction may be neglected. On the con- factor dominates and the cathodic current
trary, it may radically change the shape diminishes at the negative shift of the
of the polarization curve in dilute solu- electrode potential in the vicinity of
tions near p.z.c., depending on the value the p.z.c. At higher electrode charges
of zi − αn. This effect originates from the (positive or negative) the variation of
linear dependence of the overall (ϕ m/s ) the diffuse-layer potential becomes very
and diffuse-layer (ψ1 ) potential drops on slow (logarithmic function of the electrode
the electrode charge, σ , at small deviations charge, Eq. (17)) compared with the overall
2.1 Electrochemical Interfaces: At the Border Line 55
potential because the compact layer gives then a deep minimum of the current
the principal contribution. Within these within the range of small negative elec-
potential ranges the variation of the trode charges.
diffuse-layer factor only modifies the global Experimentally such nonmonotonous
tendency determined by the Tafel term, curves were observed for numerous re-
that is, the cathodic current increases actions of the anion electroreduction:
for negative potential shift. As a whole, S2 O8 2− , S4 O6 2− , Fe(CN)6 3− , PtCl4 2− , and
Eq. (24) predicts a nonmonotonous current- so on, where all above-mentioned qual-
potential curve in dilute solutions, with itative predictions of the slow-discharge
a dropping branch in the vicinity of the theory were confirmed [54]. For the quan-
p.z.c. The amplitude of this effect depends titative treatment, it was suggested that
strongly on the charge of the reactant, zi , corrected Tafel plots be used in co-
and the electrolyte concentration, and the ordinates: log ic + (zi e/2.3kB T )ψ1 , E −
current of the anion reduction may almost ψ1 [55], where one should observe a
vanish within a wide range of negative straight line with the slope −αne/2.3kB T .
electrode potentials, E − Eσ =0 < 0. Besides, the data for different concentra-
One must keep in mind that all this anal- tions of the background electrolyte should
ysis was performed for systems without a fall on the same line [56]. These predic-
specific adsorption of reacting species or
tions have been confirmed (some cur-
indifferent electrolyte. The violation of this
vature of corrected plots was found in
requirement leads to the necessity to in-
some cases) by experimental studies of
troduce the energies of specific adsorption
the persulfate, tetrathionate, ferricyanide,
of reactants or their interaction with ad-
and some other anions [57].
sorbed species, into Eq. (24). Fortunately
An additional test of the theory is given
this problem is of less importance for the
by the comparison of the data for different
case of anion reduction, since the most
electrode materials. Experimentally the re-
interesting effects take place at the nega-
tively charged electrode surface in which duction of the same anion at different
anion-adsorption effects are minimal. metals leads to quite different patterns
The analysis in Eq. (24) does not take for the polarization curve, in particular, a
into account the transport limitations. drastic variation of the interval in which
Therefore these predictions of the diffuse- the diffuse-layer minimum is observed.
layer effects are only valid for the interval of According to the theory (24), if the react-
the potentials in which the rate of the pro- ing species do not enter the compact layer
cess is determined by the electron-transfer and the possible change of the electron-
step. If this kinetic regime corresponds tunneling factor does not influence the
to high negative electrode charges, the reaction rate (e.g. for adiabatic electro-
Kψ1 factor in Eq. (25) varies slowly and chemical reactions [58]) the corrected Tafel
the deviations from the Tafel behavior are plots must be independent of the electrode
rather weak. On the contrary, the anion material. This prediction has also been
electroreduction wave that starts at posi- confirmed experimentally for the persul-
tive electrode charges may demonstrate a fate reduction at Hg (amalgams), Bi, Sn,
complicated curve: a usual behavior within Pb, and Cd [59]. Similar results have been
this potential range, with approach to the obtained for the persulfate reduction in
limiting current at less positive potentials, some nonaqueous solvents [60, 61].
56 2 Electrochemical Double Layers
The exposed version of the model factors [64, 65] has allowed one to provide
disregards an essential feature of this class a quantitative interpretation of experimen-
of reactions, that is, the formation of ion tal data for anion electroreduction at high
associates between the reacting anion and overvoltages and has confirmed expecta-
background cations, which is especially tions of its proximity to the activationless
pronounced for the multicharged anions. region.
The ways to determine the charge of the
discharging species in the bulk solution 2.1.5.4 Electrokinetic Phenomena
(which is generally different from the one Another area in which the existence of the
inside the EDL) are analyzed in [62]. Some diffuse part of the EDL plays a dominant
deviations from the Frumkin theory may role is represented by various electrokinetic
also originate from the difference of the phenomena [67, 68].
closest approaches of anions and cations Some of them are observed at mem-
to the electrode surface [47]. branes separating two identical electrolyte
The Frumkin correction in Eq. (24) re- solutions. For example, in electroosmosis,
duces the EDL structure effect to its single one can impose a difference between
characteristic, ψ1 potential, while real re- the electric potentials in these solutions,
actants represent a distribution of electric by means of immersed electrodes, which
charges. A recent development has taken leads to the generation of an electric field
into account this distribution in the cal- across the membrane because of the cur-
culation of the work term, so that it rent passage. As a response, one observes
depends on the whole profile of the poten- a solution flow across the membrane. For
tial across the EDL. Besides, the presence its qualitative interpretation, one considers
of the reactant perturbs this potential pro- pores of the membrane as a set of parallel
file in the vicinity of the species (a kind cylindrical capillaries. An EDL is formed at
of discreteness-of-charge effects [48] dis- the internal surface of each capillary, due
cussed in Sect. 2.1.11.3), due to both the to the ionization of surface groups or/and
field created by its charges and the cavity ion adsorption/association, accompanied
effects arising from the replacement of sol- by the accumulation of the counterions
vent molecules in the volume occupied by inside the diffuse layer. As soon as the elec-
the solute species [63, 64]. tric field E is imposed along the capillary
The quantitative theory also has to take surface, the force zi e E acts on each ion,
into account a variation of the reaction including those within the diffuse layer.
volume with the electrode charge [65]. An- Owing to the mobility of the latter and the
other weak point of the model (24) is the viscosity of the solvent, this force induces
hypothesis of the Tafel dependence on the movement of the solution along the
the potential difference across the compact capillary. The velocity of this flow (with
layer. Quantum-mechanical estimates [64] respect to the electric field amplitude) in
testify in favor of a considerable variation the stationary conditions is described by
of the charge-transfer coefficient, α, over the Smoluchowski equation:
the wide potential range explored exper-
v
imentally, or even its approach to zero, = (4π)−1 εη−1 ζ (27)
which corresponds to the activationless E
regime of the charge-transfer process [66]. where ε is the relative dielectric permittiv-
Recent analysis which has included these ity of the solvent, η the dynamic viscosity
2.1 Electrochemical Interfaces: At the Border Line 57
of the solvent, and ζ (’’zeta potential’’) the to the solution, for example, under the
potential in the ‘‘slip plane’’ inside the influence of the gravity force (sedimen-
diffuse layer separating the hydrodynam- tation). The viscosity force results in a
ically immobile (attached to the surface) displacement of the mobile diffuse-layer
and moving volumes of the solution. This charges with respect to the fixed ones at
potential is often identified to the poten- the surface, that is, to the creation of an
tial of the outer Helmholtz plan, ϕGC , electric dipole. The superposition of the
Eq. (16). fields of such dipoles gives an electric field
An inverse phenomenon (streaming po- oriented along the movement axis (sedi-
tential), generation of an electric field mentation potential).
inside the membrane, takes place if the so- It should be noted that the direction of
lution passes through this porous medium the induced effect depends on the sign
due to an imposed hydrostatic pressure. of the diffuse-layer potential, so that one
This time it is the flow of the fluid inside can determine from such experiments the
the pores that induces the displacement of sign of the ‘‘fixed’’ surface charge or even
the mobile part of the EDL at the surface of estimate its value.
capillaries, with respect to the charges at- The variation of the electrolyte concen-
tached to the surface. These dipoles create tration or the solution composition leads to
an electric field, which, under stationary drastic changes in the properties of the col-
conditions, prevents the farther displace- loidal system, in particular to changes of its
ment of the mobile charges. The resulting stability, which is primarily determined by
potential difference across the membrane, the electrostatic repulsion between the par-
ϕ, is proportional to the excessive hydro- ticles. The latter is due to the formation of
static pressure, P : similar EDLs owing to the identical chem-
ical nature of the surfaces. The repulsion
ϕ
= (4π)−1 εη−1 −1 ζ (28) is only efficient at long distances between
P the particles (at shorter distances the in-
where is the electrolyte conductivity, and teraction between the particles becomes
the remaining notation is identical to that attractive due to van der Waals forces). This
in Eq. (27). is why long-range diffuse layers are neces-
The existence of the extended diffuse sary for the colloid’s stability. The increase
layer in dilute solutions is also of crucial of the electrolyte concentration strongly di-
importance in colloidal systems formed by minishes the Debye screening length, that
particles whose size is comparable to the is, the distance of the particle–particle in-
wavelength of light, that is, about 1 µm. For teraction, with the coalescence and the sub-
example, electrophoresis, the movement of sequent sedimentation. Another scenario
colloidal particles with respect to the so- of the kind takes place upon change of the
lution under the influence of an external solution pH. In most cases the fixed charge
electric field, originates from the displace- of the particle’s surface is formed at least
ment of the particle and the mobile part partially by the proton-exchange equilib-
of its EDL in the opposite directions under rium. The change of pH results in the mod-
the action of this field, since these com- ification of this surface charge, and the sys-
ponents of the system possess charges of tems loses its stability at the moment when
opposite signs. A related phenomenon oc- the surface charge passes zero and the re-
curs if the solid particles move with respect pulsion between the particles disappears.
58 2 Electrochemical Double Layers
2.1.6 areas:
S
Solid Electrodes: Effects of Polycrystallinity R≡ , R≥1 (29)
and Surface Roughness Sapp
with the Debye screening length in the is more complicated). Besides, contrary
solution, which is about 10 nm in mM to the theoretical predictions, there was a
solutions. strong variation of the cathodic branch of
In a modified variant of this approach, the curve as a function of the electrolyte
the surface roughness was determined concentration, which was hardly possible
from the same ratio of capacitances (31) to attribute to the specific adsorption of
measured for the identical high negative cations. The anodic branch was even more
charges [69]. Under these conditions the irregular.
diffuse-layer contribution may be practi- As a result, within the initial period of the
cally neglected while the difference be- solid metal studies, the only quantitative
tween the compact-layer capacitances of treatment was made for the data at the
different metals becomes rather small, com- diffuse-layer minimum. In accordance
pared to the vicinity of the p.z.c. or at to Eq. (32), the corresponding PZ plots
positive electrode charges. The principal (dependence of Capp −1 versus CGC −1 ) for
shortcoming of this estimate is a signifi- various solid metals had lower slopes than
cant variation of the cathodic branch of the 1, which were interpreted as R −1 . The plots
capacitance curve with the electrolyte con- (if the concentration range was sufficiently
centration observed in most publications. broad) showed also a significant curvature
All this reasoning is based on the in most cases. The latter was attributed to
hypothesis that the separation (20) of the various experimental errors.
capacitance to unit area of the real surface, After having the RF determined one
C, into the compact- and diffuse-layer could calculate the charge density per unit
contribution remains valid for solid metals ‘‘real’’ surface area, σ , by the integration of
(see [70] and earlier references therein). It the C(E) ≡ R −1 Capp (E) curve. Determin-
gives for experimentally measurable values ing then CGC in Eq. (32) one may find the
of the capacitance: compact-layer capacitance for all electrode
charges. Finally, the ‘‘theoretical’’ capac-
(Capp )−1 = R −1 CH −1 (σ ) itance curves could be found for lower
+ R −1 CGC −1 (σ, c) (32) electrolyte concentrations. However, such
an analysis has almost never been per-
One must keep in mind that the GC expres- formed, that is, only the values of R and the
sion (21a) for the diffuse-layer capacitance, compact-layer capacitance for the p.z.c.,
CGC , contains the charge density per unit CH (0), were extracted (see Ref. [34] for a
real surface area, so that it can only be review).
calculated (except for its value for σ = 0) if An alternative method to determine
the RF is known. the RF and the compact-layer capacitance
The attempts to apply the Grahame curve from experimental data for solid
treatment in the quantitative manner en- surfaces was proposed in [69]. It is based
countered certain problems. First, the on the same model (32) of a uniformly
impedance data interpreted in terms of charged rough surface. On the basis of this
the in-series combination of the elec- equation the compact-layer capacitance
trolyte resistance and the EDL capac- curve, CH (σ ), was calculated for several
itance, C, demonstrated a significant values of the RF R, see Fig. 7, for a fixed
frequency dispersion of these elements electrolyte concentration. Similar to Fig. 3
(which implies that the equivalent circuit for the mercury electrode (in which the
60 2 Electrochemical Double Layers
200
1.1
1.2
1.3
[µF cm−2]
CH
100
10 0 −10
σ
[µC cm−2]
Fig. 7 The CH (σ ) plots for a polycrystalline silver electrode
calculated with the use of Eq. (32) for several values of the RF, R,
indicated at each plot [69].
effect originated from the experimental or (if verified) not satisfied, especially for
errors in the capacitance values), the PC electrodes.
shape of the curve within the interval Equation (32) of this model implies
near the p.z.c. was especially sensitive to that the surface roughness is realized
the choice of R. In particular, a value by the spatial configuration whose scale
of R (R = 1.2 in Fig. 7) was found for is well above the thickness of the EDL,
which the whole curve was passing this in particular the Debye screening length
interval smoothly. It was assumed that of the solution, LD (18), to allow the
this procedure allowed one to determine interfacial structure to follow the curved
both R and the CH (σ ) curve. The value surface without its especial distortion.
of R obtained in such a way was in Sometimes in earlier studies, (see [34] for
accordance with the one found from the review) the treatment of the capacitance
comparison of EDL capacitance values for data for solid metals was based on a
PC samples of this metal (silver) and different relation:
its single-crystal faces at high negative
electrode charges. (Capp )−1 = R −1 CH −1 (σapp )
One should note that one has to verify, + CGC −1 (σapp , c) (33)
whether these characteristics are the same
for different electrolyte concentrations, Physically, it may simulate the situation
like it was done in Fig. 3 for mercury. of a solid electrode with a microroughness
This requirement was mostly not verified whose scale is well below the diffuse-layer
2.1 Electrochemical Interfaces: At the Border Line 61
thickness. Then the compact-layer capac- like Pb, Bi, Sb, Sn, and Cd, for which
itance per unit apparent surface area is the difference of the p.z.c. potentials be-
increased by the RF, while the diffuse layer tween the single-crystal faces of the same
is formed outside this microrough inter- metal is sufficiently small, within 100 mV
face and its structure is identical to the one or less. On the contrary, a great dispersion
for liquid metals. of p.z.c. values for different single-crystal
Model (33) predicts a unit slope for the faces (300 mV and greater) was revealed for
PZ plots. It is in contradiction with ex- Ag, Au, Cu, and so forth. Besides, for all
perimental data, at least for PC electrodes. face-centered cubic metals of this group,
However, it should be noted that recent two basic crystal faces, (111) and (100), pos-
theoretical studies of the EDL structure at sess the most positive among those p.z.c.
rough (but uniform) surfaces have shown values, while even a relatively small change
that these two models may be considered of the crystallographic orientation results
as limiting cases of a more general descrip- in a rapid shift of the p.z.c. to the most
tion, Sect. 2.1.6.4. negative values that are characteristic for
One must keep in mind that the increase most higher-order crystal planes.
of the EDL thickness upon a decrease of
This finding led to the development of
the electrolyte concentration is only valid
a new concept emphasizing the effects of
for a close vicinity of the p.z.c. The GC
crystallographic inhomogeneity of the PC
formulas for the ion charge distribution
electrodes by Frumkin [71]. It postulates
show that outside this narrow interval
(in its simplest version) that the PC sur-
of the electrode charges, the diffuse-layer
face represents a combination of various
thickness becomes independent of the
single-crystal faces – each of them having
ionic strength of the solution, see Eq. (21a)
or (21b) as an illustration. This effect their own EDL, the structure of which is
originates from the nonlinear response of identical to the one at the correspond-
the diffuse layer and makes the area of the ing real single-crystal electrode (model
applicability of Eq. (33) quite narrow. of ‘‘independent electrodes’’ [72] or ‘‘inde-
pendent diffuse layers (IDL)’’ [20]). It was
noted [71] that at a fixed value of the PC
2.1.6.2 Crystallographic Inhomogeneity
Effects for Solid Electrode Surfaces electrode potential, E, the charge densities
The analysis of models (32) and (33) of individual faces were different and that
demonstrates the necessity to be very this factor might change the shape of the
careful in comparing the values of the C(E) curve for the PC electrode near its
RF of the same surface found by different diffuse-layer minimum.
experimental techniques, since they may Later the capacitance properties of a
be crucially dependent on the scale probed PC surface were analyzed on the basis
in the corresponding method. of Eq. (34) [69, 73, 74], where the C PC (E)
Significant new information was pro- curve was represented as the sum of the
vided by capacitance measurements at capacitance curves for individual faces,
single-crystal faces of solid metals. It was C (i) (E), multiplied by their fractions of
found that there existed two groups of the total surface area, θ (i) :
solid metals having qualitatively different
properties. One of them includes numer- C PC (E) = θ (i) C (i) (E) (34)
ous ‘‘mercury-like’’ metals or semimetals, i
62 2 Electrochemical Double Layers
Though this concept looks almost self- in the vicinity of the zero PC charge,
evident, one should emphasize that it σ PC = 0, because of incorrect subtrac-
results in quite different predictions on tion of the diffuse-layer contribution.
the EDL properties, compared to the Moreover, such calculated ‘‘compact-
aforementioned model of a rough but layer curves’’ are strongly dependent on
uniform surface (32) or (33): the electrolyte concentration and have
no relation to the ones obtained by the
• the diffuse-layer minimum of the averaging of the compact-layer curves
C (i) (E) curve in dilute solutions is for individual faces with the same frac-
widened and not so deep, in compar- tions θ (i) , similar to Eq. (34).
ison with the minimums for individual
faces; These strong ‘‘anomalies’’ were predicted
• for metals with a considerable variation for metals whose PC surfaces were com-
of p.z.c. values among the faces the posed by low-index single-crystal faces (e.g.
minimum of the C (i) (E) curve, Emin PC , (111),(100) and (110) for face centered cu-
is close to the most electronegative bic (fcc) lattices) having a strong difference
among p.z.c. of the faces, for example, between their p.z.c. (several hundred mV).
Emin PC ∼ = Eσ =0 (110) for Ag, Au, Cu; For ‘‘mercury-like’’ PC metals, the pre-
• the formally constructed PZ plot, that dicted effects were less pronounced, even
is, the dependence of (Cmin PC )−1 on though the formal PZ plots were to be
CGC −1 , cannot be used to determine markedly curved.
the electrode surface roughness or the It was also recognized [74] that there
compact-layer capacitance; existed an alternative model of the PC sur-
• this plot is expected to be curved (down- face, ‘‘EDL with a common/united diffuse
ward for lower concentrations) as a layer’’. It means that if the size of the in-
result of crystallographic inhomogene- dividual faces at the surface is sufficiently
ity effects, if the concentration interval small, the diffuse layer is smooth along
is sufficiently wide; this heterogeneous surface, i.e., it is anal-
• the overall charge of the PC surface, ogous to the one for a uniform surface but
σ PC , is nonzero at Emin PC , and its value with the charge density in expression (21a)
depends on the electrolyte concentra- replaced by the average charge density of
tion; the heterogeneous surface:
• at the p.z.c. of the PC surface, σ PC = 0,
the charges of regions (faces) at the σ PC = θ (i) σ (i) (35)
surface are nonzero, σ (i) = 0, that is, i
this point corresponds simply to the
compensation of the contributions of Then the interfacial capacitance of such
opposite signs into the overall PC PC surface should be described by the
surface charge; equation:
• if the compact-layer capacitance curve
of the PC surface, CH PC (σ PC ), is cal- (C PC )−1 = (CH PC )−1 + CGC −1 (σ PC , c)
culated with the use of the traditional (36a)
scheme assuming the uniformity of the where ‘‘the compact-layer capacitance of
PC surface, Eq. (32), then this curve dis- the PC surface’’, CH PC , is given by the
plays rapid nonmonotonous variation average of the compact-layer capacitances
2.1 Electrochemical Interfaces: At the Border Line 63
of the individual faces: the individual faces, θ (i) , in Eqs. (34) and
(i) (35) since they have been postulated on
CH PC = θ (i) CH (σ (i) ) (36b) an intuitive basis rather than determined
i from measurements. One should also
The latter depends on the charge density keep in mind the necessity to distinguish
at the corresponding face, σ (i) , rather than between the capacitance values related to
on the average density σ PC . Evidently this the ‘‘real’’ or ‘‘apparent’’ surface areas of
model predicted quite different effects: the PC electrode, Eq. (30). In other words,
one must take into account the RF of the
• the coincidence of the p.z.c. of the PC PC surface, R PC . Then the experimentally
surface with the potential of its C PC (E) measured capacitance value, Capp PC (E),
curve, Emin PC ; is given by Eq. (37) if the ‘‘model of
• the independence of its value on the independent electrodes’’ is used:
electrolyte concentration;
• a straight line with a unit slope for the Capp PC (E) = R PC θ (i) C (i) (E) (37)
i
PZ plots at any fixed value, and so on.
Here, similar to the previous relations, the
In an attempt to choose an adequate sum of the fractions of the individual faces
treatment, the predictions for both mod- at the PC surface, θ (i) , is equal to 1. The
els were compared with experimental data capacitance curves C (i) (E) correspond to
for PC samples of various metals. Qual- perfect faces without any roughness.
itatively the conclusions of the model of To avoid a misunderstanding, one must
‘‘independent electrodes’’ turned out to be keep in mind that all conclusions derived
in agreement with experimental data for on the basis of Eq. (34) remain valid for
such metals as Ag, Au, Cu and so on, see this case, too. In particular, it is impossible
reviews in Ref. [20, 34], contrary to the fail- to obtain the RF of the PC surface, R PC ,
ure of the ‘‘common diffuse layer (CDL)’’ from the slope of the PZ plots. On the
model, Eq. (36a). However, in quantitative other hand, one may determine both the
terms the effects of crystallographic inho- RF and the fractions of the individual
mogeneity were smaller than predicted by faces from the comparison of experimental
the model based on Eq. (34). For example, capacitance curves for the PC surface
the capacitance at the minimum, Cmin PC , and the individual faces (if the latter are
for PC Ag diminishes more rapidly than in known) [75]. These characteristics of the
its theoretical estimates. The experimental PC surface are found by minimizing the
slope of the PZ plots calculated with no difference between the left- and right-
account for the surface heterogeneity ef- hand side terms in Eq. (37) with respect to
fects is close to 1 for Bi, within a very wide the fitting parameters, R PC and θ (i) . This
range of the concentration, in contrast to operation applied to the PC Ag electrode
the model analysis. The value of R found enabled one to reproduce properly its
from the same plot is close to 1 for some experimental capacitance curve on the
PC Ag electrodes in rather concentrated basis of those for the three silver low-index
solutions. faces, (111), (100) and (110). Moreover, it
One of the possible reasons of this was found that the values of R PC and θ (i)
disparity may be an inappropriate choice of obtained from the fitting for each electrolyte
the fractions of the PC surface occupied by concentration turned out to be practically
64 2 Electrochemical Double Layers
the same for all concentrations. The value observed for single-crystal faces of many
of R PC obtained in this way was close to metals. Then the response of the EDL is
the ratio of the Capp PC and C (i) at high always nonlinear, since the electron charge
negative potentials, in accordance with density is sufficiently high at least on some
earlier expectations, see previous text. of the surface domains, since the electrode
potential E is far from their p.z.c. val-
2.1.6.3 EDL Structure for NonUniform ues, Eσ =O (i) . For these surface regions,
Electrode Surfaces the dominant contribution to the local ca-
All this modeling was based on the pacitance is given by the compact layer,
hypothesis that the EDL structure is one- even if the electrolyte concentration is very
dimensional, that is, all profiles (concen- low. For some other regions, the poten-
trations, ion charge density, and electric tial E may be close to their p.z.c., the
potential) depend on a single coordinate EDL structure may be described within
normal to the surface, either within the the linear-response model and their overall
whole diffuse layer in model (36a) or capacitance may be essentially determined
within the EDL at each single-crystal face by the diffuse-layer contribution. However,
in model (34) or (37). However, a rig- the latter leads to a strong reduction of
orous theoretical analysis [76, 77] shows the contributions of such regions into the
that these distributions are essentially per- overall capacitance of the heterogeneous
turbed near all contacts of the surface surface, since the contributions of the re-
domains possessing different p.z.c. or/and gions are additive.
compact-layer capacitance values. It results As a result the criterions of models
in a nonuniform distribution of the surface (37) or (36a) depend mostly on the
charge density. The width of this perturbed difference of the partial p.z.c. values and
zone is determined by the screening prop- on the partial compact-layer capacitances.
erties of the electrolyte inside the diffuse In particular, for the most interesting case
layer, in particular it is close to the De- of a strongly heterogeneous surface (i.e.
bye screening length in the bulk solution, having a significant difference between
LD , in the simplest approximation of the the partial p.z.c. values) the principal
linear response. Then the criterion of the condition relates the characteristic size of
aforementioned models is in the compar- the uniform regions, y∗ , and the lengths
ison between the characteristic size of the (i)
LH derived from the values of the partial
individual uniform areas at the PC sur- (i)
compact-layer capacitances, CH , e.g., for
face (e.g. single-crystal faces), y∗ , and the
the ‘‘model of independent electrodes’’
Debye screening length, LD . For example,
(Eqs. (34) and (37)):
for model (34) and (37) it has the form
y∗ L D .
y∗ max{εLH (1) , εLH (2) , . . .},
This approximation of the linear re-
sponse may be justified for heterogeneous LH (i) ≡ (4πCH (i) )−1 (38a)
surfaces with a small difference of the par-
tial p.z.c. values, in particular for some One should emphasize that this condition
mercury-like solid metals. However, the is sufficient even for very dilute solutions,
analysis is much more complicated [72, in which the Debye screening length
78] in the case of a surface with a great dif- in the bulk, LD , may strongly exceed
ference of these p.z.c. values, for example, the geometrical size, y∗ , contrary to
2.1 Electrochemical Interfaces: At the Border Line 65
the intuitive expectations based on the structure. The corresponding model was
linear-response approximation. first proposed for the imperfect single-
The criterion is slightly more compli- crystal (111) face of Au in Ref [79], which
cated for systems with a moderate (but was considered as a combination of per-
not small) difference of the partial p.z.c. fect (111) regions and intermediate zones.
values for which, the diffuse-layer min- Then in accordance with the approach
ima for the different uniform regions as of ‘‘independent electrodes’’, the over-
a function of the electrode potential over- all capacitance was represented as the
lap partially. Then, condition (38a) must be sum of the perfect C 111 (E) curve and
supplemented by another one; the geomet- the one of the PC Au electrode. Possi-
rical size should also significantly exceed ble influence of various higher-order faces
the shortest among the partial screening and surface defects on the capacitance
lengths inside the diffuse layers at each properties of PC surfaces was discussed
region: in [80]. Later, it was suggested to repre-
sent the overall PC capacitance as the
y∗ max{LD (1) , LD (2) , . . .}, sum of the contributions from all low-
LD (i) ≡ (4πεCGC (i) )−1 (38b) index faces (identical to Eq. (37)) and from
the intermediate regions, the latter be-
determined by the diffuse-layer capac- ing modeled in accordance with Eq. (36).
itances for the corresponding regions, This treatment was extensively used to
CGC (i) . The latter depends on the local analyze the properties of Bi electrodes,
charge density for this region according to see [20].
Eq. (21a). For metals with a great difference be-
The characteristic lengths εLH (i) vary tween the p.z.c. values of the faces, it is
widely as a function of the metal and necessary to keep in mind the specific
solvent properties. For aqueous solutions, dependence of the p.z.c. values of the
they are within 5 nm, or even smaller for single-crystal faces as a function of their
metals with high values of the compact- orientation; for example, for various fcc
layer capacitance for single-crystal faces at metals the low-index faces, (111) and (100),
small electrode charges, such as Ag or Au. possess ‘‘singular’’ p.z.c. properties, while
Therefore this analysis testifies in favor this characteristic rapidly approaches a
of the ‘‘model of independent electrodes’’ much more positive value typical for (110)
for such systems, in conformity with and higher-index faces, at a relatively small
the experimental evidences mentioned change of the crystallographic orientation.
above. Therefore, one cannot exclude an alterna-
However, one should keep in mind that tive interpretation of the above-mentioned
all this analysis is based on the assumption experimental observations for such PC
that the principal contribution to the PC ca- electrodes. Namely, one can envisage that
pacitance is given by perfect single-crystal their surfaces do not contain perfect (111)
faces at its surface, which must have a suf- or (100) faces but only perturbed ones.
ficiently large size, Eq. (38a). One cannot Their p.z.c. values are not much differ-
a priori exclude a considerable contribu- ent from those for the (110) or (210)
tion due to intermediate regions between faces.
well-defined uniform surface fragments, The possibility of getting a substantiated
or even a completely amorphous surface answer might lie in the use of modern
66 2 Electrochemical Double Layers
1.5
1
1.4
Roughness function
1.3
[R]
~
2
1.2
3
1.1
1
0 0.2 0.4 0.6 0.8 1
k
[nm−1]
Fig. 8Roughness function dependence on inverse Debye length for the
random Gaussian surface. ε = 80, h = 5 nm, l = (1)10, (2)15, (3)20 nm.
68 2 Electrochemical Double Layers
for nonfractal electrodes. Efforts in this describe the main points of this theory [43,
direction have been made by the Tartu 44, 90].
group [86] that have extracted R̃(κ) =
C/CGC from their data and plotted it 2.1.7.1 Basic Equations
versus κ. It was found that for metals At a given surface charge density, σ ,
with close values of the p.z.c. for different of the electrode, the potential distribu-
crystal faces the trends of the theory are tion along the z-normal to the elec-
basically correct, but even for such metals trode surface, φσ (z), obeys the Pois-
the deviations from the nonequipotential son equation d2 φσ (z)/dz2 = −4π(ρms σ +
character of the outer Helmholtz surface σ σ
ρion ), where ρms is the bound charge
(assumed to be equipotential in Ref. [81]) density of the metal and solvent in con-
are likely to be important for a quantitative tact and ρionσ is the charge density of
comparison. For metals with strong vari- electrolyte ions. Subtracting from this
ance of the p.z.c., the theory in its present equation the same equation at σ = 0, for
form essentially fails. It needs an exten- the so-called rationalized potential, φ(z) =
sion that would combine this theory with φσ (z) − φ0 (z), we get
the models of energetic inhomogeneity de-
scribed in the previous section. Note that d2 φ(z) 0
Lust and coworkers compared the nonlin- = −4π(ρms + ρion
σ
− ρion ) (44)
dz2
ear variant of the theory also [83], that is,
the so-called potential-dependent roughness where ρms = ρms σ − ρ 0 . To close this
ms
function, with the values extracted from equation let us introduce the nonlocal
their data [87], and came essentially to permeability of the metal/solvent system,
the same conclusions. Further system- χ(z, z ), which relates the change in the
atic data would be highly welcome, as bound charge density to the corresponding
well as the mentioned extension of the change of the electric field
theory. ∞
d dφ(z )
ρms = dz χ(z, z ) (45)
dz −∞ dz
2.1.7
Semi-phenomenological Nonlocal Theory Then the equation for φ reads:
of the Double Layer
∞
d dφ(z )
dz ε(z, z )
The dielectric response of the interface dz −∞ dz
can be described in a unified manner in 0
= −4π[ρion
σ
− ρion ] (46)
terms of the nonlocal electrostatic the-
ory [88, 89]. Indeed, it was shown to be where
possible to express the electric proper-
ties of the interface through the dielectric ε(z, z ) = δ(z, z ) + 4πχ(z, z ) (47)
function of the metal/solvent system, not
applying a particular form of this func- is the static nonlocal dielectric function of
tion, for any structure of the interface. the metal/solvent system that keeps all the
Such an approach allows revealing gen- information about its structure [88–91].
eral properties of the double layer and Equation (46) is still not closed, as we
expressing the parameters involved via still need an equation that relates the r.h.s.
the nonlocal dielectric function. We briefly with φ. A particular closure is determined
2.1 Electrochemical Interfaces: At the Border Line 69
but not by the distance of closest approach dipoles in the layer bounding the metal,
of electrolyte ions. ‘‘spin–up’’ (↑) and ‘‘spin–down’’ (↓),
An extension of the nonlocal dielectric with a dipole pointing away or to the
theory to the range of moderate deviations metal, respectively. Some residual energy
from the p.z.c. was reported in Refs. [43, is attributed to each state in the field
90]. The results differ in replacing 1/εκ by of an uncharged electrode, U↑ and U↓ .
1/4πCd , where Cd is the nonlinear differ- Boltzmann statistics relates then the oc-
ential capacitance of the diffuse layer (e.g. cupation numbers of the states, N↑ and
of GC type), and ε(z, z ) by the so-called N↓ : kB T ln(N↑ /N↓ ) = U↓ − U↑ + 2p/d
differential dielectric function ε(z, z , σ ) (for where p is a permanent dipole molecule
details see Ref. [90]). Somehow the Gra- of the molecule, is the potential drop
hame parameterization is again recovered, across the layer of thickness d, and
but the validity criteria now involve the kB is Boltzmann’s constant. Here the
value of σ . Critical is the requirement that dipole moments are assumed to have the
the amount of the electrolyte charge, ac- same magnitude for both orientations,
cumulated within the layer of thickness but the dipole moments are, in general,
z∗ , constitutes a small portion of the total
different from those in the gas phase
charge σ . For aqueous electrolytes, the cri-
or in the bulk of the solvent. The or-
terion limits the results to σ < 10 µC/cm2
der parameter then reads (N↑ − N↓ )/N =
and c < 0.1 M. This is the sufficient crite-
tanh p/kB T d, where = − m ,
rion. In fact the differences between the
m = d/2p(U↑ − U↓ ), and N = N↑ + N↓
values of CH as evaluated from small and
is the total number of molecules in the
high concentrations typically do not exceed
layer per unit surface area. This is not yet
5%. The reason for that phenomenon is
discussed in Ref. [88]. In general Grahame a closed equation; the closure is reached
parameterization may eventually work at by the assumption that is a superposi-
higher charges and concentrations. tion of the potential drop in the absence
of dipoles and the contribution due to
2.1.8
dipoles:
‘‘Dipolar’’ Models of the Compact Layer
1 p
= 4πσ d − 4πpN tanh
Historically, molecular models of the ε∞ kB T d
compact layer were first to rationalize (54)
its nonlinear dielectric response. Starting where σ is the specific charge of the
with Ref. [98] and until the beginning of electrode and ε∞ is the high frequency
the 1980s, this type of models comprised effective dielectric constant, related to
the main direction of research in the the intramolecular polarizability of the
theory of the electrochemical interface (for molecules. The p.z.c., obtained from
detailed review see Refs. [88, 99]). We will Eq. (54) at σ = 0, lies generally between
discuss several of the most important ones 0 and m .
in the following text. Equation (54) implicitly determines the
value of potential for a given charge of
2.1.8.1 Two-State Model the electrode. Its differentiation over σ
This model [98] assumes the existence gives the famous bell-shaped differential
of two admissible orientations of solvent capacitance of the molecular layer with a
2.1 Electrochemical Interfaces: At the Border Line 71
which can be evaluated from independent surfaces [109]. First calculations were per-
data. A better fit was achieved in a more formed in the so-called jellium model [110],
involved analysis of Ref. [106], but as noted in which the skeleton is treated as a ho-
in Ref. [99] the best fit would have been mogeneous semi-infinite background of
achieved for a dipole moment of water positive charge. Later on the calculations
twice as large as its value in the gas phase, were modified to account for the discrete
which may be a bit too much. Furthermore, structure of the skeleton with the use of
the sign of the temperature derivative [105, ionic pseudopotentials [111].
106] is opposite to the experimentally ob- These predictions have been verified by
served one. experiments on low-energy helium atom
The lesson from this four-state model scattering by metal surfaces [112]. The
is that the interplay between the field- helium atom repels the Bloch electrons,
induced destruction of the clusters and because their wave function must become
reorientations of the clusters and sin- distorted to preserve orthogonality to the
wave functions of the closed shells of
gle molecules may be a source of a
helium. The repulsion potential appears
most peculiar CM (σ )-behavior. The rich-
in practise to be proportional to the local
est curve exhibits four maxima and three
free electron density. Thus a helium atom
minima.
scatters like a ping-pong ball from the
2.1.9
electron cloud of the metal, and this allows
The Role of Metal Electrons to probe the distribution of electrons in the
cloud, that is, the profile of the electronic
We now briefly describe the effects asso- tail and its lateral corrugation that usually
ciated with the surface electronic profile follows the periodicity of the surface crystal
plane [112].
of the metal and its influence on the re-
How does the surface electronic pro-
sponse of the electrochemical interface to
file deform with charging? Although there
charging. For a detailed review, the reader
was no direct observation of the charge-
is addressed to Ref. [107].
induced profile modulation, a lot is known
from the theory and indirect manifes-
2.1.9.1 Surface Electronic Profile and its tation of this effect in field emission.
Response to Charging For the jellium model, after the pioneer-
How close do solvent molecules come to ing papers [110, 113, 114], the problem
the metal? Where is the boundary of the was studied in detail in Ref. [115] within
metal? To answer these questions, we must the trial function version of the den-
go back to the structure of the metal sur- sity functional formalism [116] and in
face. The existence of a tail of quasi-free Ref. [117, 118] within the more general
electrons spilled out of the ionic skele- Kohn–Sham scheme [119]. We summa-
ton of the metal was predicted in the rize here the properties necessary for our
early days of quantum mechanics [108]. further discussion.
In the 1970s, this picture was approved If no discreteness of lattice charge is
and detailed by the electron-density func- taken into account or if we speak about the
tional theory of the inhomogeneous elec- laterally averaged electron density profile,
tron gas (later leading to Nobel fame for n(z), neglecting Friedel oscillations [110],
Walter Kohn in 1998) applied to metal it can be roughly approximated by a trial
2.1 Electrochemical Interfaces: At the Border Line 73
0.014
0.012
0.010
0.008
[a.u.]
n(z)
0.006
σ = −0.005 a.u.
0.004 (= −28.5 µC cm−2)
0.002 σ=0
0.000
−4 −2 0 2 4
z
[a.u.]
Fig. 9 Surface electronic profile of a neutral and negatively charged jellium
calculated for the bulk electron density of Hg.
74 2 Electrochemical Double Layers
σ = − 0.005 a.u.
(= −28.5 µC cm−2)
0.002
∆ n(z)
[a.u.]
0.000
−4 −2 0 2 4
z
[a.u.]
Fig. 10 Distribution of excess negative charge corresponding to the profile
shown in Fig. 9.
4.0
3.5 Hg
3.0
[a.u.]
2.5 Ga
zσ
2.0
1.5
1.0
density functional results, because neither and Ga are shown in Figs. 11 and 12,
is the trial function exact nor the den- respectively. For all the metals studied,
sity functional. The dramatic history of the z0 > 0, ã > 0, p < 0. This means that
employment of exact sum rules in electro- for negative charges the electronic profile
chemistry is discussed in Ref. [107]. broadens and the excess charge center-
The zσ − and β(σ )-curves for bulk of-mass moves further outside the metal.
electron densities corresponding to Hg These two features are also essential
2.1 Electrochemical Interfaces: At the Border Line 75
1.5
Hg
[a.u.]
1/β
1.0
Ga
0.5
−0.004 − 0.002 0.000 0.002 0.004
−σ
[a.u.]
Fig. 12 The same graphs as in Fig. 11, but for the characteristic length of
smearing of the surface electronic profiles. Smearing increases with negative
charging.
for understanding the metal response to The profile of an ideally smooth interface
charging in situ, although the bounding is sketched in Fig. 13. The half-space z < 0
medium may affect the values of β(σ ) is occupied by the ionic skeleton of the
and zσ . However, the qualitative behavior metal. This can be described, roughly, in a
of these functions remains the same, jellium model, as a continuum of positive
unless the solvent molecules chemisorb charge n+ and the effective dielectric
on the metal surface. In that case with constant εb due to the polarizability of
negative charging of the metal the extra the bound electrons (this quantity is,
electrons may have to be localized within with rare exceptions (Hg: εb = 2, Ag:
the skeleton half-space, thereby decreasing εb = 3.5), typically close to 1 [125]). The
zσ and β(σ )−1 . gap 0 < z < a accounts for a nonzero
distance of the closest approach of solvent
2.1.9.2 Response to Charging of the molecules to the skeleton. The region of
Electrochemical Interface and the a < z < a + d stands for the first layer
Compact-Layer Capacitance: Main of solvent molecules, while z > a + d is
Qualitative Effects the diffuse-layer region. n(z) denotes the
The reader interested in the history of profile of the density of free electrons. This
the step-by-step comprehension of the is, of course, an extremely crude picture,
role of metal electrons in the double-layer but it eventually helps to rationalize the
theory and various ideas in this field is results of the various theoretical models
referred to the reviews [107, 120–124]. In and simulations.
the following sections, we will quote only The integral KH or differential CH ca-
the references needed for understanding pacitances of the compact layer, related to
the simple but basic qualitative features each other as 1/CH = (d/dσ )[σ/(KH (σ ))],
discussed in this tutorial overview. are extractable from the overall capacitance
76 2 Electrochemical Double Layers
n (z)
n+
z
0 a a+d
Fig. 13 A cartoon of a profile of a smooth electrochemical interface. The
half-space z < 0 is occupied by the metal ionic skeleton that, within the
jellium model, is described as a continuum of positive charge density
(n+) and the dielectric constant due to bound electrons (εb ), the value of
which lies typically between 1 and 2. The gap accounts for a finite distance
of closest approach of solvent molecules to the skeleton; the gap is
determined by the balance of forces that attract the molecules to the
metal and the Pauli repulsion of the closed shells of the molecules from
the free electron cloud of the metal of density n(z). The regions
a < z < a + d and z > a + d correspond, respectively, to the first layer of
solvent molecules (which can be roughly characterized by
charge-dependent effective dielectric constant) and the diffuse-layer part.
of the interface via for example, the PZ Here, we just refer to the behavior that
plots [37] or other methods discussed in explains the systematic trend for all s,p-
this chapter. To the accuracy of exponen- metals [126] – the positive slope of the
tially small terms, they can be written as: capacitance near the p.z.c., shown Fig. 14.
Such a behavior was first found in
1 d
a − zσ + (60) Refs. [127, 128] by density functional
4πKH ε(σ ) simulation. Here, for clarity, the degrees
1 da d of freedom of the molecular reorientations
a+σ − z∗ + in the first layer were frozen, that is,
4πCH dσ ε∗ (σ )
ε(σ ) = const. The peculiar a(σ ) behavior
d(σ zσ ) 1 d σ
z∗ = , = is determined by two main effects. The
dσ ε∗ (σ ) dσ ε(σ ) first effect is asymmetric with respect to
(61) p.z.c. That is, the position of the solvent
Here the d/ε(σ )-contribution is due to molecules that are ‘‘sitting on the tail’’
the potential drop across the first layer of free electrons follows the deformation
of solvent molecules, where the effective of the tail. With negative charging the
dielectric constant ε(σ ) takes into account profile swells out of the metal skeleton
the nonlinear dielectric polarizability of and a increases; with positive charging the
the layer. Thus in order to understand profile moves back towards the metal and a
the σ -dependence of the compact-layer decreases. The second effect is symmetric
capacitance, it is sufficient to understand with respect to the p.z.c. Any capacitor
the σ -dependence of a − zσ and d/ε(σ ). with a flexible distance between the plates
Different microscopic models of the will contract upon charging regardless of
interface may be classified with regards the sign of the charge. The combination
to these dependencies (see Ref. [107]). of these two effects results in the negative
2.1 Electrochemical Interfaces: At the Border Line 77
KH
zσ
σs 0 σs 0
−σ −σ
Fig. 14 A sketch of the charge dependence of the distance of closest approach
of solvent molecules to the skeleton, a, the center of mass of the excess charge
distribution, zσ , and the corresponding plots for the integral capacitance of the
compact layer, KH . In the case of ‘‘soft landing’’ of the molecule onto the metal
skeleton edge (dashed curves), the spike in the capacitance transforms into a
hump.
slope of a(σ ) near the p.z.c. (which is were attributed to the nonlinear response
typically steeper than the slope of zσ ) and of molecular reorientations that were so
the maximum in the cathodic range. At far frozen in order to make clear the
some positive charge, denoted here as σs , mere effect of a(σ ) and zσ . In terms
a will reach zero and will stop changing of Eq. (60), that would mean that ε(σ ) =
with further increase of σ , because the const. For the simplest two-state molecular
molecules will encounter the edge of models, it has a bell-shaped form, the
the metal skeleton. The stop would not maximum of which is expected in the
be abrupt (solid curve), but soft (dashed cathodic range. The hump is, however,
curve) if the skeleton were not a hard observed in the anodic range. Since
wall for solvent molecules. The simplest the (a(σ ) − zσ )-determined ‘‘background’’
approximation for a(σ ) is curve is properly asymmetric, a bell of ε(σ )
a(σ ) = [a(0) − Aσ − Bσ 2 ], centered near the p.z.c. will be seen as a
hump in the anodic range.
A > 0, B > 0 (62) Is the smeared ‘‘stop-spike’’ for KH
(Fig. 14) what we see for a hump on
The difference between a(σ ) and zσ
(Fig. 14) leads to KH (σ ) plots that are the measured capacitance curve, or is it
in line with the data for simple met- screened by molecular reorientations? We
als (Figs. 15 and 16), (for review and don’t know that. Moreover, the specific
references, see Ref. [88]) in contact with adsorption of ions, which was assumed
surface-inactive electrolytes; the slope at absent, can give rise to a hump too.
the p.z.c. is positive, the minimum lies in Somehow this feature is interesting be-
the cathodic range and a kind of a spike, cause of its quantum-mechanical origin
which might be apprehended as a hump, and direct dependence on the nature of the
takes place in the anodic range. metal. The more densely packed the metal
Of course there could be other reasons surface, the more abrupt the stop, the
for the hump. In the molecular models sharper the spike. Open surfaces should
of the compact layer (see Sect. 2.1.8), they favor ‘‘soft landing’’ and the hump would
78 2 Electrochemical Double Layers
Ga
120
100
80
[µF cm−2]
CH
60
InGa
40 Bi
Hg
20
12 8 4 0 −4 −8 −12
σ
[µC cm−2]
Fig. 15 Metal effect on the compact-layer differential capacity: the
series Hg, Bi, InGa alloy, Ga. Data from Refs. [129–132].
130
Ag
100
[µF cm−2]
CH
70
Cd
40 Hg
40 20 0 − 20
σ
[µC cm−2]
Fig. 16 Metal effect on the compact-layer differential capacity: the
series Hg, Cd, polycrystalline Ag. Data from Refs. [35, 133, 134].
be smoother. For single-crystalline sur- metal, the denser the electronic tail, and
faces, the spikes should be sharper than the stronger the repulsion of the molecules
for liquid metals, in which the interface from it. Higher positive charges will be
undergoes fluctuations, and for uneven needed to press the solvent to the skele-
polycrystalline surfaces, in which inhomo- ton and σs will move farther to the anodic
geneous broadening may take place. The range. This is what we see, if we compare
higher the bulk electron density of the Ga and Hg.
2.1 Electrochemical Interfaces: At the Border Line 79
θ
Metal Electrolyte
z
O
a = (aO − Aσ − Bσ 2) θ ( σs − σ)
a
Y ≡ en + dσ
d
σ
10 O σs
da
= (− A − 2Bσ) θ (σs − σ)
dσ
O σ
σs
5
p.z.c.
E In Pb
[V]
−5
−10
plasma contributions may be expected. For compression of the surface electronic pro-
more on this interference, see in Ref. [107]. file under the adsorbed anions. The greater
Simple jellium-model concepts also the field-induced adsorption, the stronger
helped to understand the potential de- the compression, and this affects the signal
pendence of a nonlinear optical signal in line with experimental observations.
from the electrochemical interface – the In this section, we have focused on
second harmonic generation [146, 147], the qualitative features of the response of
and of the frequency of surface plasmon the metal–electrolyte interface to charging
resonance [148, 149]. In both cases, the rather than on first principle calculations
effect of field-induced adsorption of an- of the absolute values of the parameters
ions at positive charges of the electrodes that characterize the interface, about which
was rationalized. An application of the one can learn more in Refs. [107, 124].
‘‘inverted’’ Lang model [107] revealed the Such calculations, however, are often
82 2 Electrochemical Double Layers
insufficiently paid back, due to the lack The simplest model that goes beyond the
of precise knowledge of the solvent effect Debye–Hückel and GC theories of point
on the metal surface electronic properties. ions and takes into account the finite size
Refs. [150, 151] show what a ‘‘resistance’’ of ions is the so-called primitive model.
the steps towards ‘‘improving’’ the picture Here, ions are described as simple hard
of this interaction encounter. spheres and the solvent is approximated
by a dielectric continuum. An even more
2.1.10 special case is the so-called restricted
Effects of the Molecular Nature of primitive model, in which both ions have
Electrolyte Solutions the same size.
Various attempts have been made in go-
The ‘‘real’’ electrochemical double layer ing beyond the primitive models towards
is not a simple arrangement of capacitors civilized models. In the most simple of
with concomitant potentials and fields, but such extensions, called the solvent-primitive
it consists of ‘‘real’’ atomic and molecu- model [152], the solvent is still treated as
lar ions and of a vast number of solvent a dielectric continuum. In addition, in-
molecules, all of which have a finite size dividual solvent molecules are treated as
and whose arrangement depends as much hard spheres. This takes into account the
on the interatomic and intermolecular in- molecular nature of the solvent in but the
teractions as on external electric fields and crudest fashion. Replacing the description
potentials. In addition, molecular ions pos- of the solvent from dielectric continuum
sess a very specific and asymmetric charge by molecular point dipoles at the cen-
distribution. Owing to their (chemical) ter of soft spheres representing solvent
interactions with the metallic phase the molecules (the Stockmayer model) leads
charge of all ions can also depend to some to the ion-dipole models. Various extensions
extent on its distance from the metal sur- of these models incorporate higher-order
face (see previous text). Furthermore, polar electrostatic moments (up to octupole mo-
solvent molecules possess a permanent ments [153]) and molecular polarizability
dipole moment and higher electrostatic for the description of the solvent bound
moments, that lead to lateral interactions charge density.
between them, which in turn, render the In order to understand ionic distribu-
simple picture of the solvent as a dielectric tions in the EDL a realistic description
continuum or n-state dipole models (see of hydration or, more generally, solva-
Sect. 2.1.8.2) inappropriate; protic solvents tion phenomena is necessary, which in
such as water and alcohols have, in addi- protic liquids implies an adequate descrip-
tion, the capability to form directional yet tion of the hydrogen-bond network. Theory
fluctuating hydrogen bonds that give rise and computer simulation of bulk liquids
to specific structural arrangements of the showed that the most efficient way to in-
solvent molecules relative to each other clude these properties into the models
and in the solvation sphere around ions. is via distributed charge models in which
These arrangements, in turn, can either the intramolecular charge distribution is
enhance or counteract the orientational represented by several point charges. The
influence due to the electronic structure point charges are adjusted to reproduce ex-
and surface charge (or potential) of the perimental dipole and/or quadrupole mo-
metal. ments of the molecule, the bulk structure
2.1 Electrochemical Interfaces: At the Border Line 83
of the liquid, or various other properties. interface. Simple models of the metal are
For the ubiquitous solvent water, the sim- hard walls with image interactions that
plest models treat water as a rigid three-site account, in a very simplistic way, for the
molecule in its gas phase geometry with electronic properties of the metal. This
partial charges on oxygen and hydrogen model is nonspecific for the chemical
sites. Ions are described as charged soft identity of the metal. The jellium model
spheres with a point charge at their center. incorporates the chemical identity of the
Molecular and ionic size is maintained metal in the form of the density of free
through Lennard–Jones interactions or electrons and ionic skeleton pseudopoten-
similar potential functions in an attempt to tials; it models the electronic spillover effect
describe short-range exchange correlation into the liquid phase (see Sect. 2.1.9.1).
and long-range dispersion interactions. Several interaction models employed in
Hydrogen bonding and hydration are pro- computer simulations are chemically spe-
duced by these models through the balance cific, that is, they have been developed
between electrostatic and nonelectrostatic for a specific metal and/or surface ge-
interactions. In addition to these simple ometry. These models may or may not
rigid three-site models many more so- be based on quantum-chemical calcula-
phisticated flexible and polarizable water tions of the interaction between water
and ion models have been developed. The molecules or ions and a small cluster
importance of the dipolar nature of the (or an extended periodic slab) of ions. A
solvent and of the interactions between widely used model has been the model of
solvent and electrode were recognized the Platinum(100)–water interface [155].
in the double-layer model by Bockris, It consists of a flexible atomic represen-
Devanathan, and Müller [154]. Water ‘‘hy- tation of the metallic phase, of surface
drates’’ the electrode, which is regarded corrugation through site-dependent water-
as a giant ion, and so contributes to the platinum interactions, and a description
electric fields near the interface. of the orientational anisotropy via specific
Molecular computer simulations are the chemical interactions between the oxygen
method of choice when it comes to in- atom of a water molecule and the plat-
vestigating molecular models like the one inum metal. This model has been extended
discussed earlier, both in the bulk and at to the (111) and other surfaces; a similar
interfaces. Contrary to analytical theories, model has been used for mercury–water
simulation methods can treat a variety of interactions [156, 157].
different models on the same statistical- In spite of the fact that rather sophisti-
mechanical footing, without mathemati- cated models can be treated, several lim-
cally mandated approximations that can- itations of computer simulation methods
not be rigorously justified and/or tested. such as Monte Carlo (MC) or Molecular
In fact, computer simulations have been Dynamics (MD) limit their use in interfa-
used as computer experiments to verify or cial electrochemistry:
falsify theoretical work.
In addition to explicitly accounting for • The spatial and concentration inhomo-
the molecular nature of electrolyte solu- geneities of the EDL are large. This
tions near the EDL, computer simulations requires the simulation of large sys-
can also be used towards realistic model- tems at high electrolyte concentrations.
ing of the metal part of the electrochemical Dynamic processes near the interface
84 2 Electrochemical Double Layers
occur on a much longer time scale date, only few calculations have been
than the inherent fluctuations of the performed [158].
hydrogen-bond network and the hydra- • The use of classical simulations on the
tion shell, or they are very rare, which basis of the Born–Oppenheimer ap-
mandates long simulation times. Even proximation is problematic when deal-
with present-day computer resources, ing with metals, even in the absence
the requirements of large system size of electrochemical reactions involving
and long simulation times are hard to the transfer of electrons, which can cer-
reconcile. tainly not be approximated by classical
• Simulations with molecular solvent de- mechanics.
scription can only be performed in • Ab initio simulation methods (like
the concentration range above about the Car–Parrinello method) are cur-
0.5 mol/liter, much higher than the rently so much limited in system size
concentrations of many electrochemi- (typically substantially less than 200
cal experiments. atoms) and simulation time (a few pi-
• Simulations of the EDL structure coseconds), that they have only been
around surface defects and thus the applied to vacuum systems, for ex-
modeling of real interfaces is almost ample, to the investigation of molec-
impossible, both because of the large ular and dissociative adsorption of
system size involved and because of water on a magnesium oxide sur-
insufficient knowledge of the electronic face [159, 160], and not to an en-
structure around these defects, and thus tire metal–solution interface. However,
the lack of suitable interaction poten- these methods hold substantial promise
tials. for the future.
• MC and MD are methods of statistical
In spite of these shortcomings, simulation
mechanics. The quality of the results de-
methods have nevertheless substantially
pends on the extent to which the phase
contributed to our understanding of the
space of the system can be sampled. At
double layer because of their ability to
best the statistical error of a simulation
describe:
decreases only slowly of the order of the
square root of the computational effort; • Packing effects like the formation of
at worst, the phase space contains bot- distinct layers of solvent molecules
tlenecks, making ergodicity impossible and ions;
to achieve. • The formation of a hydrogen-bond
• Reliable and accurate potential energy network in protic solvents such as
surfaces on the basis of quantum water;
chemistry for the interaction between • Orientational structure and its change
molecules and ions with the metal near charged and uncharged surfaces;
surface are still very scarce. Especially, • Hydration of ions;
cluster calculations fail to reach the • Diffusion and reorientational dynam-
limit in which the cluster of metal atoms ics;
begins to exhibit metallic properties. • Solvent effects on the chemical dynam-
Methods based on periodic slabs appear ics of ion transfer, ET and several other
to be more promising for the future; to electrochemical reactions.
2.1 Electrochemical Interfaces: At the Border Line 85
One of the goals of applying molec- density profile and ρi is the bulk value,
ular simulation to the electrochemical z the distance from the charged surface
double layer is, of course a molecular and i runs over all ion species, shows
interpretation of capacitance data (see drying at low density and exhibits a maxi-
Sects. 5 and 6). Because of the above- mum in the range of the critical density of
mentioned limitations of the method, the bulk electrolyte [161]. With increasing
progress in this field has been rather lim- surface charge density, ionic adsorption
ited. Henderson and coworkers recently passes from negative to positive.
studied the properties of the electrochemi- MC simulations of the solvent primitive
cal double layer both at high ionic concen- model [163] show, not surprisingly, that
trations (a model of the molten salt–metal the double-layer capacity increases with in-
interface) and at low ionic concentrations creasing electrolyte concentration and also
(modeling the double layer of an elec- with increasing solvent concentration. At
trolyte solution) by MC calculations. They the same time, with increasing solvent con-
showed, with the primitive model, that at centration, the ionic density profiles pass
low temperatures the double-layer capac- from nonoscillatory to oscillatory behav-
itance decreases with decreasing temper- ior, a common feature of dense systems.
atures. This behavior was observed both Furthermore, at low surface charge densi-
for the low-concentration electrolyte solu- ties the contact value of the charge density
tion [161] and for the molten salt case [162]. increases and a passage between partial
In their studies, they were only able to cal- drying (at low solvent concentration) and
culate the integral capacitance at the point partial wetting of the surface by the elec-
of zero charge. GC theory without (GC) trolyte is observed with increasing solvent
and with a Stern layer (GCS) as well as density. The temperature dependence is
the mean spherical approximation (MSA), such that the ion adsorption is nega-
on the other hand, predicts a monotonous tive at low temperature and positive at
decrease of double-layer capacitance with high temperature. The negative ion ad-
temperature. The primitive model MC data sorption at low temperature becomes less
showed that the capacitance versus tem- pronounced with increasing particle den-
perature curve goes, at least at lower ion sity. The behavior can be explained by the
densities, through a maximum and that fact that electrostatic ion–ion interactions
at high temperature, analytical theories and dominate at low temperature; whereas the
MC experiment agree well. The primitive excluded volume effect of the solvent is
model simulations thus indicate that ion more important at higher temperature.
size-dependent effects become important, Extending these studies to ion-dipole
thus limiting the range of applicability of mixtures in which a background dielectric
the analytical theories of sizeless ions to constant is replaced by fluctuating solvent
the high-temperature regime. point dipoles, leads to ‘‘practical nonergodic-
At high electrolyte concentrations, the ity’’ due to the formation of clusters and/or
primitive model simulations clearly ex- strings of particles [164].
hibit the importance of ion size in the Computer simulations of ‘‘realistic’’
form of oscillatory density and charge models of water, ions, and metal surfaces
profiles in the vicinity of charged sur- also suffer from nonergodicity problems;
faces[162].
The MC adsorption isotherm however, they are able to paint a very de-
= i [ρi (z) − ρi ]dz, where ρi (z) is the tailed picture of the microscopic liquid
86
10 0.15
0.1
8 0.05
6 0
ρµ (z)
ρ (z)/ρ b
[Cm−2]
4 − 0.05
2 Electrochemical Double Layers
− 0.1
2
− 0.15
(a) (c)
1
0.5
0.0 1
[V]
χ (z)
ρ c(z)
−0.5 0
[1010 Cm−3]
−1.0
−1
−1.5
−2
−15 −10 −5 0 5 10 15 −15 −10 −5 0 5 10 15
z z
(b) [Å] (d) [Å]
Fig. 19 (a) Oxygen (full) and hydrogen (dashed) density profiles ρ(z), normalized to the bulk density ρb . (b) Charge density ρc (z). (c) Dipole density
ρµ (z). (d) Water contribution to the surface potential χ(z) calculated from the charge density ρc (z) by means of Eq. (64). All data are taken from a 150 ps
simulation of 252 water molecules between two mercury phases with (111) surface structure using Ewald summation in two dimensions for the
long-range interactions.
2.1 Electrochemical Interfaces: At the Border Line 87
structure near a metal electrode and near artificial character of these systems should
a liquid–liquid interface. Models of the in- be kept in mind.
terface range from simple image charge The charge density profile and the dipole
models [165] through jellium models [166] density profile can be calculated from the
to complex potential energy functions atomic density profiles and the orienta-
derived from quantum-chemical calcu- tional distributions. Figures 19b and 19c
lations [167, 168] of clusters of metal show the dipole and charge density, respec-
atoms interacting with water molecules. tively, for water near the mercury surface.
These studies have been reviewed re- For both the first and the second water
cently [169–171]. Only some key results layers, a quadrupolar charge profile with al-
are discussed in the following text, and ternating regions of positive, negative, and
the interested reader should consult the again positive charge density is observed.
reviews and the references therein for fur- The form of the charge profiles originates
ther information. from the fact that the centers of mass (or
Figure 19 shows as an example the data the oxygen atom) of the water molecules
from a simulation of TIP4P water that are mostly well localized, while the hydro-
is confined on both sides by a rhombo- gen density distribution is broadened as a
hedral mercury crystal with (111) surface result of librational motions and as a result
structure. Bosio and coworkers [172] de- of the hydrogen bonding between layers.
duced from their x-ray studies that a solid Beyond the second layers the charge den-
α-mercury lattice with a larger lattice con- sity approaches zero within the limits of
stant in z direction may be used as a statistical uncertainty. The dipole density
good structural model for liquid mercury. profile indicates ordered dipoles in the ad-
Thus the mercury phase was modeled as sorbate layer. The orientation is largely due
a rigid crystal, in order to simplify the to the anisotropy of the water–metal inter-
simulations. The surface of such a crystal action potential in this particular model,
shows rather low corrugation. Figure 19a which favors configurations in which the
contains the oxygen and hydrogen den- oxygen atom is closer to the surface. Most
sity profiles. As a result of the significant quantum-chemical calculations of water
adsorption energy of water on transition near metal surfaces to date predict a sig-
metal surfaces (typically of the order of 20 nificant preference of ‘‘oxygen-down’’ con-
to 50 kJ mol−1 ; see, e. g. Ref. [173]), pro- figurations over ‘‘hydrogen-down’’ ones
nounced density oscillations are observed at zero electric field (e. g., Ref. [168,
next to the metal. Between three and four 178–183]). The dipole orientation in the
water layers have also been identified in second layer is only weakly anisotropic (see
most simulations near uncharged metal also Fig. 20). Solving the one-dimensional
surfaces, depending on the model and on Poisson equation with the charge density
statistical accuracy. Beyond about 10–12 Å profile ρc (z) (Fig. 19d), the electrostatic
from the surface the density is typically potential drop near the interface can be
constant and equal to the bulk value. In calculated according to
strong unscreened electric fields, several
z
authors [174–177] report a phase transi-
tion towards a ferroelectric crystalline state χ(z) = −4π ρc (z ) · (z − z ) dz
−∞
in their simulations. However, the rather (64)
88 2 Electrochemical Double Layers
c
σ = 4.4
d
e
σ = 2.2
f
h σ=0
j σ = − 2.2
l
σ = − 4.4
m
σ = − 8.8
o
It has been demonstrated in Ref. [184, potential in the center of the lam-
185] that the use of the Ewald sum- ina.
mation is crucial to obtain the field- Lateral density fluctuations are mostly
free bulk region of constant electrostatic confined to the adsorbed water layer. In the
2.1 Electrochemical Interfaces: At the Border Line 89
first layer the oxygen distribution shows the basis of the water–metal interaction
the structure of the substrate lattice. In the potential. Within the adsorbate layer, there
second layer the distribution is more or is a transition from the preference for ori-
less isotropic. As a consequence, oxygen entations in which the dipoles point more
motion is predominantly oscillatory rather or less into the solution (a and b) to one
than diffusive in the first layer. Con- where a substantial fraction of the dipoles
sequently, the self-diffusion coefficient point more or less towards the surface (c
in the adsorbate layer is strongly re- and d). The orientational anisotropy ranges
duced compared to the second or third as far into the liquid phase as the density
layer [186–189]. It was concluded that the inhomogeneities do (roughly up to panel
motion in the first layer is characteristic of m), with increasingly less pronounced
a solid phase, while the motion in the sec- features. Slightly beyond the second max-
ond layer has more liquid-like character. imum in the density profile the orienta-
The liquid–liquid water–mercury inter- tional distribution is isotropic, as it has to
face has also been studied [190]. The major be the case for a bulk-like liquid. Thus,
difference of the water structure between like in the case of nonpolar surfaces, the
the liquid–solid and the liquid–liquid in- orientational distribution is governed by
terface is due to the capillary width of water–water interactions. All orientational
the liquid mercury surface, which smears distributions are rather wide and liquid-
out the water-density profiles consider- like, although the fact that the orientational
ably [191, 192]. preference changes within the adsorbate
The orientational structure of water near layer is sometimes attributed to ice-like
a metal surface has obvious consequences structural elements near the interface (see
for the electrostatic potential across an in- Ref. [193] for a more detailed discussion).
terface, since any orientational anisotropy An external electric field changes the ori-
creates an electric field that interacts with entational distribution and consequently
the metal electrons. Hydrogen bonds are the orientational polarization of the water
formed mainly within the adsorbate layer molecules in the interfacial region. The ef-
but also between the adsorbate and the fect of homogeneous [174, 175, 194–197]
second layer. The left side of Fig. 20 shows and inhomogeneous [198] electric fields on
the orientational distribution of the molec- the orientational distribution near smooth
ular dipole moment, relative to the surface, model surfaces has been investigated us-
normal in various distance ranges from the ing lattice models [197] and distributed
Hg(111) surface. Additionally, the oxygen point charge models for water. Also the
density profile is plotted. The baselines be- field-induced changes on the more realis-
tween distribution functions cut through tic Pt(100) surface were studied [176, 199,
the density profile. The distribution func- 200]. Because of the absence of free ions
tion in each panel on the left side is for in these simulations of pure water the
the subset of molecules that are located in electric field is only screened by the wa-
the distance range between these lines on ter dipoles themselves; consequently a net
the right side. Over the first peak in the electric field and a concomitant polariza-
density profiles (panels a to d), there are al- tion persist through the lamina. At large
most no molecules whose dipole moment field strengths, this has been found to
is perpendicular to the surface, as would lead to a field-induced phase transition
be expected for an isolated molecule on to a crystalline water phase [174–176]. The
90 2 Electrochemical Double Layers
right part of Fig. 20 shows the orientational are not contact adsorbed, Cl− being a
distribution of the dipole moment vectors borderline case. Calculations on the ba-
in the adsorbed water layer for various sur- sis of quantum-chemical potentials [167,
face charge densities, σ . In the field-free 191] show contact adsorption for almost
case a wide bimodal-orientational distribu- all ions, although a detailed analysis of
tion is observed. The overall preference of the local structures around these ions re-
‘‘oxygen down’’ bonding is manifest in the vealed characteristic differences between
larger probability to find positive values large and small ions.
of cos(θ) (corresponding to angles smaller The first simulation studies of full
than 90◦ between dipole vector and the double layers with molecular models of
surface normal that points into the liquid ions and solvent were performed by
phase) than to find negative ones. For pos- Philpott and coworkers [212, 213] for the
itive surface charges, the average dipole NaCl solution. The authors studied the
moment (the first moment of the distribu- screening of a negative surface charge by
tion) shifts with increasing surface charge free ions in several highly concentrated
density towards larger absolute values and NaCl solutions. A combination of (9-3) LJ
the distribution becomes increasingly nar-
potential and image charges was used to
rower. For low negative surface charge
describe the metal surface. More recently,
densities, the dipole-orientational distri-
Spohr [214, 215] investigated the 2.2 molal
bution becomes more symmetric around
NaCl and CsF solutions in the vicinity of a
the parallel orientation. At larger negative
corrugated surface as a function of surface
surface charge densities, the orientational
charge density. In a similar manner the
distribution changes in such a way that the
interface between a 1 M KCl solution and a
hydrogen atoms point preferentially to the
mercury electrode was studied by Dimitrov
surface. The effect of field-induced crys-
and Raev [216].
tallization was observed at much higher
surface charges [176]. The surface X-ray In the studies of Ref. [214, 215], which
experiments by Toney and coworkers [201] will be discussed in some more detail, wa-
give experimental evidence for voltage de- ter films consisting of 400 water molecules
pendent ordering of water on a silver solvate 32 ions in the vicinity of the metal
electrode. They observed a shift of the surface, with image charges and a full 2D
silver-oxygen distance with applied poten- Ewald treatment for the Coulomb interac-
tial. However, they also made a debated tions. Like in the studies by Philpott and
conclusion a strong increase in local water coworkers, the solution is not necessarily
density near the surface, while MD data electroneutral; the total charge in the so-
suggest only an increase in correlation but lution is balanced by the image charges,
not in overall packing density. which give rise to a surface charge density
Several simulation studies of the poten- σ equal to the excess image charge divided
tial of mean force of a single ion near by the area of the interface.
a metal surface have been performed, Figure 21 shows the ion density profiles
with image charge models and quantum- near the metal surface for three surface
chemical models of the metal [202–211]. charges. Beyond z = 15 Å all ion density
With the image charge models, only the profiles are identical within the limits
larger halide ions are contact adsorbed, of statistical errors. The oxygen density
whereas small ions such as Li+ and F− profile does not change much with surface
NaCl CsF
ρO(z) ρO(z)
4 4
3 3
2 2
1 1
10 σ = +9.9 10 σ = + 9.9
8 8
6 6
4 4
2 2
10 σ=0 10 σ=0
8 8
6 6
4 4
2 2
10 σ = −9.9 10 σ = − 9.9
8 8
6 6
4 4
2 2
5 10 5 10
z z
[Å] [Å]
Fig. 21 Density profiles ρ(z) and running integrals Nion (z) of the ion densities for cations (full lines) and anions (dashed lines) at three different surface
2.1 Electrochemical Interfaces: At the Border Line
charge densities in units of µC cm−2 as indicated. Left: NaCl solutions; right: CsF solutions. The top graphs show, for reference, the corresponding
oxygen density profile near the uncharged surface. Density profiles are normalized to the bulk densities corresponding to 2.2 molal solutions in each case.
91
92 2 Electrochemical Double Layers
charge and is repeated here only to surface charge densities. Obviously, with
provide a geometric reference for the ion the models used in this study, Na+ does
positions. Together with the ion densities not contact-adsorb. The Cl− density profile
the running integrals of the densities, at σ = −10 µC cm−2 is similar to the one
defined as at σ = 0.
z The monotonically increasing curves of
Nion (z) = Bx By ρ(z ) dz (65) the running integrals give the number of
0
ions, whose distance from the electrode
are plotted for cations (full lines) and is smaller than z. The position at which
anions (dashed lines). Bx = By = 18 Å are the difference between the anion and
the box dimensions parallel to the interface cation running integral becomes 0 (for
and ρ is the particle number density. the uncharged surface) or ±2 (for the
Figure 21 (left) shows the ion density charged surfaces) corresponds roughly to
distributions of NaCl in the vicinity of the thickness of the diffuse part of the
the metal electrode. Near the uncharged double layer. This thickness is less than
electrode, there are no pronounced adsorp- 10 Å in all cases.
tion maxima. The density of Na+ (full line) Figure 21 (right) shows the correspond-
is slightly increased in the range around ing density profiles for the CsF solutions.
z = 4.3 Å between the first and second At zero surface charge, there is very little
density maximum of water. The Cl− den- contact adsorption of Cs+ and no contact
sity (dashed line) is significantly reduced adsorption of F− . The preferred position of
up to about z = 5 Å. There is no contact ad- the Cs+ ions is in the second water layer,
sorption of Na+ , since no cations are found while F− ions prefer the region between
for z < 2.8 Å, while water molecules can be two water layers, similar to Na+ . At posi-
found up to z = 2 Å. Cl− ions, on the other tive surface charge densities, the amount
hand, can be found (with low probability) of F− in the interlayer region increases,
at distances below 2.3 Å. but no contact adsorption is observed. No
At positive surface charge density, the Cs+ cations are found in the first layer. At
Cl− density exhibits a large maximum at the negative surface charge density, Cs+
very short distance from the electrode. forms a contact-adsorbed layer, similar to
The position of this maximum is closer Cl− at the positive surface charge. The
to the electrode than that of the first thickness of the diffuse layer is also in
water layer, thus giving a clear indica- the range of 10 Å, judging from the be-
tion of contact adsorption of this anion. havior of the running integral of the ion
The Na+ density near the electrode is density.
slightly reduced as a result of the re- No contact adsorption occurs in the
pulsion between the positively charged simulations on uncharged electrodes. The
cations and the positively charged sur- small ions Na+ and F− do not adsorb
face. The position of the first maximum directly on the electrode surface at mod-
of the Na+ density profile is not shifted erate negative and positive surface charge
very much. densities, respectively. These ions form
At negative surface charge density, the rather rigid hydration shells consisting of
Na+ density exhibits a large maximum six or seven water molecules and favor
at around z = 4 Å. This position is very the interlayer region between the first and
similar to the one at vanishing and positive second water layer, in which they can
2.1 Electrochemical Interfaces: At the Border Line 93
∞
form stable hydration shells. However, ◦
i = [ci (z) − ci ] dz (66)
the large ions Cs+ and Cl− exhibit con- zG
tact adsorption when their interaction
with the surface charge density becomes Here, coordinate z is normal to the inter-
attractive. face, and the integration is performed over
From the ion density profiles, it is the whole perturbed region. An important
obvious that the surface charge is screened point in this formula is the lower integra-
within less than 10 Å. Thus the thickness tion limit, zG , corresponding to the choice
of the diffuse layer is of the same order of the ‘‘phase boundary’’, whose attribu-
tion is not self-evident. This ambiguity of
of magnitude as the one derived from the
the definition (66) is removed by the ther-
Debye length (rD = 2.1 Å at 2.2 mol/liter),
modynamics of charged interfaces based
in spite of the invalidity of the GC theory
on the Gibbs electrocapillary equation (67):
at these high concentrations. The results
clearly show the correlation between the
dγ = − j dµj (67)
strength of the hydration forces and j
the preferred positions of the various
ions relative to charged and uncharged where γ is the interfacial energy, µj
electrode surfaces. the electrochemical potential of the com-
Solvent effects on chemical reactions ponent j (the value of µj is identical
are presently being studied intensively by in the bulk medium and within the
molecular computer simulation methods. interphase, in view of the thermody-
Outer sphere ET (e. g. Ref. [217–228]), ion namic equilibrium); summation includes
transfer [229–232], proton transfer [233], all components of the phases in contact.
and bondbreaking [234, 235] are among The variations of different electrochemical
the reactions studied near electrodes and potentials in Eq. (67) are not independent;
liquid–liquid interfaces. Much of this work in the bulk phase they are coupled by
has been reviewed recently [170, 236, 237] the Gibbs–Duhem identity. If the elec-
and will therefore not be discussed here. trochemical potentials of all components
All studies involve the calculation of a free j , except for component i and solvent s,
energy profile as a function of a spatial or are kept constant, Eq. (67) leads to the
a collective solvent coordinate. relation:
(s) Ni ∂γ
2.1.11 i ≡ i − s =− ,
Ns ∂µi
Ionic Adsorption
keeping µj = constant, j = i, s
2.1.11.1 Qualitative Aspects (68)
In the general meaning of this term, The multiplier Ni /Ns in this equation rep-
adsorption of a solution component means resents the ratio of moles of components i
any perturbation of its concentration and s in the bulk phase, or the ratio of their
within the interphase, ci (z), compared molar fractions. The resulting quantity,
with its value in the bulk phase, ci ◦ . Its (s)
i , which is called the (Gibbs) adsorption
intensity is characterized by the integral of component i with respect to component
value of this perturbation, which is called s (or the relative interfacial excess of com-
the adsorption of this component: ponent i), is already independent of the
94 2 Electrochemical Double Layers
choice of the ‘‘phase boundary’’, zG . The The estimates show that the charge-
latter is often chosen in the way to make the independent term in Eq. (69) is only
solvent adsorption zero, s = 0; then, both noticeable for not too dilute solutions,
(s)
quantities become identical, i ≡ i . In about 1 M or higher. For lower elec-
dilute solutions the so-defined position of trolyte concentrations, it can be mostly
the separation plane is close to the physi- disregarded. The other terms in Eq. (69)
cal boundary between the electrode and the represent monotonous functions of the
adjacent solvent layer, being practically in- electrode charge, with the anion/cation
dependent of the electrolyte concentration. contribution being dominant for suffi-
Let us consider first a surface-inactive ciently positive/negative values.
solution, whose properties have already The behavior, that the electrolyte is sur-
been outlined in Sects. 2.1.3 to 2.1.6. By face–inactive with respect to an interface
definition, this term means that the com- over a wide range of potentials includ-
pact layer is solely composed of solvent ing the p.z.c., represents a rare exception,
molecules. The distribution of compo- while the majority of solute ion and neu-
nents within the diffuse layer is deter- tral species demonstrate specific adsorption,
mined mostly by the electrostatic forces. If that is, they significantly penetrate the
the image forces (see Sect. 2.1.11.3) can be compact layer. However, the existence of
disregarded the concentrations of solute at least one surface-inactive electrolyte MA
species, ionic or neutral, at the p.z.c. are for a particular metal–solvent interface
identical to their bulk values, and the in- provides very important advantages for
tegral (66) over the diffuse layer vanishes. the further study of the surface activity
Inside the compact layer the concentra- of other species. First, the measurements
tion, ci (z), is zero, and the integration in pure MA solutions allow one to de-
in (66) yields Eq. (12) for the Gibbs ad- termine the background characteristics of
sorption of surface-inactive components, the metal–solvent interface, including the
± = −zH c, which allows one to measure value of the p.z.c. and the compact-layer
capacitance curve (see Sects. 2.1.2–2.1.4).
the thickness of the ion-free layer of the
Then, further studies (e.g. of the adsorp-
solvent, zH .
tion of anion A ) are performed in mixed
For nonzero charges of the electrode,
electrolyte solutions, MA + MA , possess-
σ = 0, the distribution of ions across
ing a constant ionic strength but a variable
the diffuse layer is governed by the self-
fraction of anions A [238, 239]. If the prop-
consistent electric potential, for example,
erties of these two anions inside the diffuse
given by the formulas of the GC theory
layer are sufficiently close, the variation of
given in Sect. 2.1.4.2. In particular, for a
the interfacial energy, γ , with the fraction
binary 1 : −1 electrolyte, this model yields
of anions A enables one to determine the
for the adsorption of surface-inactive ions:
specific adsorption of these anions, that is,
(1)
its amount inside the compact layer, A ,
± = −zH c−(1/2e)
rather than the overall interfacial excess of
× [σ∗ ± σ − (σ 2 + σ∗2 )1/2 ] (69) these species, A , that appears in Eqs. (67)
and (68).
F is the Faraday constant, c the electrolyte Alternatively a study can be realized
concentration, and σ∗ is the characteristic for binary electrolyte solutions, MA , at
charge defined by Eq. (18). varying concentrations [24] that provide,
2.1 Electrochemical Interfaces: At the Border Line 95
in favorable cases, the same information. cations. In particular, the alkali cations
However, the specific adsorption of A are almost surface-inactive, except for
must vanish at sufficiently high (but still the interval of sufficiently high negative
experimentally available) negative elec- electrode charges, where the adsorption
trode potentials, so that the capacitance of Cs+ is most noticeable (see also
curves for the MA and MA solutions Sect. 2.1.10). However, even for this case
at identical concentrations coincide. This the adsorption is weak, that is, the charge
condition is necessary for the back- of their specific adsorption, σ1 , is less than
integration operation, thereby allowing the the absolute value of the electrode charge,
calculation of the charge-potential depen- σ , that is, the charge of the diffuse layer
dence for MA solutions. Besides, one has has the opposite sign compared to the
to subtract the diffuse-layer adsorption of electrode charge, and the distribution of
A , A , from its overall adsorption, which
(2)
the average potential across the interphase
is based on the formula (69) in which is given by a monotonous function.
the electrode charge, σ , is replaced by Anion adsorption possesses quite differ-
(1)
the sum, σ + σ1 , σ1 = zA eA being the ent features. For most anions, the specific
charge of A inside the compact layer, and adsorption is only absent for sufficiently
zA the charge valence of anions A . As high negative electrode charges. The
one can conclude from the analysis in diminution of this charge results in a rapid
Sect. 2.1.5.1, the charge dependence pre- increase of the amount of anions inside the
dicted by this formula cannot be verified compact layer, even before the p.z.c. of the
from existing experimental data for high system is reached. For the zero charge of
electrode charges, so that the separation of the electrode, the EDL is formed by the an-
the overall adsorption value into the diffuse ions inside the compact layer and the coun-
and compact-layer contributions becomes tercharge in the diffuse layer. It means a
ambiguous. shift of the corresponding electrode poten-
These measurements enable one to de- tial, that is, the p.z.c., in the negative direc-
termine the (specific) adsorption isotherm, tion, the value being correlated with the in-
that is, the dependence of the interfacial tensity of the anion specific adsorption. For
excess of the species on their concentra- their adsorption at the positively charged
(1)
tion in the bulk phase, A (cA ). For a metal surfaces, one observes the recharg-
constant value of the bulk concentration, ing effect; the absolute value of the anion
its adsorption also depends on the electri- charge, σ1 , exceeds the electrode charge so
cal variable, for example, on the electrode that the diffuse layer should be charged pos-
potential E. In mixed electrolytes, this pro- itively. It leads to a nonmonotonous profile
cess is also affected by the overall ionic of the electric potential across the inter-
strength, in particular via the diffuse-layer phase, with important consequences for
structure. In more complicated cases a the phenomena determined by the diffuse-
combined specific adsorption is observed, layer effects, Sect. 2.1.5.
that is, the compact layer contains (besides Data on ionic adsorption are available for
the solvent) species of different kinds, for numerous systems, see for example, [34,
example, both cations and anions. 240–245]. One can see that the surface
Generally the specific adsorption at the coverage by adsorbed ions (adsorption
surface of s,p-metals from protic solvents values with respect to their maximum
is more pronounced for anions than for level corresponding to the completely
96 2 Electrochemical Double Layers
covered surface) is mostly much lower adsorption obeys the Henri isotherm, that
than 1 for mercury and ‘‘mercury-like’’ is, the amount of ions inside the compact
metals, even at high electrode charges. layer is proportional to the bulk con-
This limitation arises from a strong centration. At higher concentrations the
repulsive interaction of the electrostatic adsorption increases much more slowly,
origin between adsorbed ions. On the approximately like the logarithm of the
other hand, the adsorption may approach a concentration. The behavior in both re-
complete coverage in some other systems, gions is influenced by the electrode poten-
for example, for halides adsorption at the tial and the ionic strength of the solution.
silver surface [246], which implies that the In the further discussion of the modeling
ions have practically lost their charge in of the adsorption isotherms, (c), it is
the transition to their adsorbed state, so assumed that the contribution of the
that this process resembles the adsorption diffuse-layer adsorption, (2) , is already
of neutral organic molecules, in which subtracted or simply disregarded, that is,
the complete coverage of the surface is the symbol will be used for the amount
often reached. Similar pronounced charge of specifically adsorbed species (called (1)
transfer from adsorbed ions to the metal in Sect. 2.1.11.1).
takes place for many transition metals. Generally the expression for this depen-
Despite of this charge transfer many dence should be derived from the equality
of these systems may be treated as of the electrochemical potential of the ion
‘‘ideally polarizable electrodes,’’ if the in the adsorbed state and in the bulk
adsorbed species are not transformed solution:
into a different component present inside
the bulk phase. The latter condition is µads () = µsol (c) (70a)
violated, for example, in the hydrogen
adsorption at metals of the platinum group In most cases the solution is sufficiently
in which the adsorbed hydrogen atoms dilute to use the ideal-solution approxima-
can be in equilibrium with protons in tion for the bulk value,
solution and hydrogen molecules in gas
µsol (c) ∼
= kB T ln(mc) + constant (70b)
phase or hydrogen dissolved inside the
metal. The latter system corresponds to m being molar fraction of the surface-
perfectly polarizable electrodes, see Ref. [13] active ion in the mixed electrolyte, and
for further discussion. c the overall concentration (m = 1 for the
Another possible complication may arise binary electrolyte). On the other hand the
from a slow approach to the thermody- approximation for the µads () dependence
namic equilibrium leading to hysteresis must take into account the interactions
phenomena, which are also typical for between the adsorbed ions, that is, the
many transition metals. change of the energy to transfer an ion
from the bulk solution to the adsorbed
2.1.11.2 Ion Adsorption Models state as a function of the adsorption level.
For metals of the mercury group, one The first theory of ion adsorption was
can distinguish different regions in the proposed by Stern [247]. His results can
ion adsorption isotherms. For sufficiently be obtained from Eq. (70a) if the elec-
low bulk concentrations (which may be trochemical potential of the ion in the
too low to be studied experimentally), the adsorbed state is represented as the
2.1 Electrochemical Interfaces: At the Border Line 97
sum of the entropy term (with account predicted by the Stern theory. For zero elec-
of the saturation term) and the work to trode charge, the average potential in the
transfer the ion from the bulk solution adsorption plane (ϕGC in Eq. (72)) is equal
into the adsorbed state in the average field to the electrode potential (at the p.z.c.),
of the EDL: Eσ =0 . Then the derivation of the Eq. (72)
leads to an inequality:
µads = kB T ln
(max − ) zi e dEσ =0
<1 (73)
+ zi eϕGC + constant (71) kB T d ln c
zi e is the charge of the species. Stern as- However, the experimental data show
sumed the location of the adsorption ion that the left-hand-site quantity exceeds
at the outer Helmholtz plane separating significantly its upper limit, 1.
the compact and diffuse layer, so that the This ‘‘paradox’’ (which remains unre-
potential in the work term in Eq. (71) solved even if the adsorption plane is
corresponds to the potential difference shifted into the compact layer) represented
across the diffuse layer. The combination a strong stimulus for the development
of Eqs. (70a) and (71) gives Stern’s expres- of the ion adsorption theory. It was
sion for the adsorption isotherm from a shown [249] that it originated from the cal-
binary electrolyte solution for the case of a culation of the work term in Eq. (71) as an
single ionic species being accumulated in average potential in the adsorption plane.
the adsorbed state: The qualitative explanation of experimen-
tal observations can be achieved, if one
−(W̄ads + zi eϕGC ) takes into account that the transfer of the
= c exp
(max − ) kB T ion into the adsorption site is accompanied
(72) by a redistribution of the other adsorbed
where ‘‘the potential of specific adsorp- ions that creates a ‘‘hole’’ for it with the
tion’’ W̄ads originated from the constant diminished repulsive potential. This in-
terms in Eq. (70b) and (71) which re- applicability of the approximation of the
flect the nonelectrostatic contributions to uniform distribution of the adsorbed ion
the energy of transfer into the adsorbed charge along the plane is called discreteness-
state, for example, the partial desolvation of-charge effects, which must be taken into
of the ion. account in the theory of this phenomenon.
This model allows one to explain such Another crucial factor is an ade-
qualitative features of the phenomenon as quate inclusion of these diffuse-layer and
the drastic (exponential) dependence of discreteness-of-charge effects into the
on the electrode charge, as well as the ex- model, which play a marked role if the over-
istence of the aforementioned two regions all electrolyte concentration is well below
in the (c) variation. However, a more 1 M. It is why the adsorption parameters
detailed experimental study has revealed found by the interpretation of experimen-
serious deviations from the predictions tal data with the use of various isotherms
of Eq. (72). For many ions, one can ob- derived mostly for uncharged species
serve the effect of Esin and Markov [248], (Flory–Huggins, Frumkin, Temkin, etc)
a stronger shift of the p.z.c. with the vari- may strongly deviate from their proper
ation of the bulk concentration than it is values.
98 2 Electrochemical Double Layers
The aforementioned diffuse-layer and to minimize the total energy of the system
discreteness-of-charge effects have been after the insertion of an additional ion,
taken into consideration in the the term in Eq. (74) dependent on the
model proposed by Grahame and charge of the specific adsorption, σ1 , was
Parsons [26, 250–252]. First, it was diminished by ‘‘the discreteness-of-charge
assumed (unlike in the Stern model) that factor’’, λ, the value of which should
the specifically adsorbed ions were located lie between 0 and 1. In the original
at the distance from the metal surface (in formulation by Grahame, this factor was
‘‘the inner Helmholtz plane’’) ensuring expressed through the ratio of KO and the
their maximum bond strength, owing to capacitance of the whole compact layer,
the combination of forces of electrostatic KH in Eq. (19). Later, its value was treated
and quantum-mechanical origins. It shows as an unknown parameter to be found
the need for the partial or even complete from the fitting of theoretical predictions
desolvation of the adsorbed species and to experimental data.
its deep penetration into the compact The use of the thus modified Eq. (74)
layer. The position of this adsorption plane and the equality of the electrochemical
depends on all components of the system, potentials (70a) lead to the adsorption
metal, solvent, and adsorbed ion. isotherm:
The second modification compared to ϕGC
ln(mc) = ln + 2B + zi e + ln β̃
the Stern theory was the introduction kB T
of ‘‘the discreteness-of-charge factor’’, λ. σ
The electrostatic part of the work of ion ln β̃ = ln β̃o + zi e KO (75)
kB T
transfer from the bulk solution into the λ
adsorption plane across the EDL field in B = zi 2 e2 KO
2kB T
Stern’s approximation may be represented
as the sum of the contributions of the Here the coefficient, B, is called attraction
compact and diffuse layers: constant, in analogy with the correspond-
ing factor in the Frumkin adsorption
WEDL = zi e(ϕ1 − ϕGC ) + zi eϕGC isotherm. However, it has got the oppo-
= zi e(σ + σ1 )KO −1 + zi eϕGC site sign, that is, it reflects the repulsion
between the adsorbed species, the am-
(74) plitude of which is diminished owing to
In the second expression in this formula, the relaxation of the adsorbed ion ensem-
the potential difference across the exter- ble (factor λ). The entropy term for the
nal part of the compact layer is replaced adsorption state is taken without the ‘‘sat-
by the sum of the electrode and specifi- uration term’’, compare Eqs. (71) and (72),
cally adsorbed charges, with the coefficient since the charging degree for s,p-metals is
inversely proportional to the integral ca- mostly very low with respect to the com-
pacitance of ‘‘the external part of the com- plete coverage.
pact layer’’, KO , that is, between the inner The expression (75) represents a virial
(adsorption plane) and outer (diffuse-layer isotherm for the (c) dependence, with
boundary) Helmholtz planes. a correction for the diffuse-layer effect,
To take into account the adjustment of if the electrical variable is kept constant.
the distribution of the previously adsorbed Unlike the case of the Frumkin isotherm
ions inside the adsorption plane in order in which it was the electrode potential, the
2.1 Electrochemical Interfaces: At the Border Line 99
Grahame–Parsons isotherm uses the elec- Popov [253]. Further discussion on this
trode charge, σ , whose value influences development is given in Chapter 3.1 of
the parameter, β̃, related to the standard this volume.
adsorption energy. Equation (75) predicts
that the isotherms for different electrode 2.1.11.3 Image Potential for Ions in the
charges must be congruent. Diffuse or Compact Layer
This isotherm was used successfully to Within the framework of the GCSG treat-
interpret experimental data for adsorp- ment (Sect. 2.1.4) the ion concentration
tion of various simple inorganic ions, across the diffuse layer is governed by
including weak specific adsorption of the Boltzmann distribution in the aver-
cations. At the same time, significant age electric field created by the electrode
deviations have been observed for some charge (plus the charge of the specifically
other systems, especially for low sur- adsorbed ions in the general case), and the
face coverages as well as for the surface ion countercharges in the diffuse layer:
recharge by adsorbed anions. Moreover,
in some cases in which the predictions ◦ −zi eϕ(z)
ci (z) = ci exp (76)
of Eq. (75) were formally in accordance kB T
with experimental data, the values of the
This formula implies that the presence of
‘‘discreteness-of-charge factor’’, λ, were
a particular ion at point z does not disturb
found outside the expected interval, nega-
this potential distribution. In reality, it is
tive or exceeding 1.
only valid as a zero order approximation.
Another shortcoming of this model
A more accurate formula contains an
is the absence of its derivation from
additional factor:
the theory based on the analysis of the
statistical-mechanical properties of the ◦ −zi eϕ(z) Wim (z)
ci (z) = ci exp −
ensemble of adsorbed ions. Progress in kB T kB T
the latter direction is discussed in the next (77)
subsection. where the ‘‘image potential’’ Wim (z) repre-
All these approaches disregard the effect sents the difference between the energies
of the gradual change of the dielectric of the whole system if the additional ion of
properties of the compact layer as a type i is added either at point z or in the
consequence of the replacement of the bulk solution, as a source of the external
solvent molecules in it by the adsorbed field for the rest of the system. Since this
species. This approximation limits the quantity is related to a single ion, expression
applicability of such theories to moderate (77) contains the Boltzmann constant, kB .
adsorption and to adsorbed species of a Within the diffuse layer the domi-
small size. In the opposite case of large nant contribution is given by forces of
ions (especially, those of organic type), electrostatic origin, analogous to the De-
this effect may become predominant even bye–Hückel ion atmosphere around each
for sufficiently low adsorption charges. charge species in the bulk solution. It
In the simplest approximation, one may was assumed frequently on the intuitive
accept a linear dependence of the compact- ground that the image potential in Eq. (77)
layer capacitance on the charge of the related to the location of the ion in the
specific adsorption, such as in the model vicinity of the metal may be modeled by
proposed by Kolotyrkin, Alexeyev, and its expression for a charged species in the
100 2 Electrochemical Double Layers
dielectric medium near its boundary with (which already appeared earlier, for exam-
a conductor: ple, in Eq. (38a)):
these parameters, the profile, Wim (z), may All these results correspond to the case
be a monotonous function (increasing or that the ion is located within the diffuse
decreasing), or a curve with a minimum, layer far away from the outer Helmholtz
see, for example, the numerical illustra- plane. The conclusions change drastically
tions in Ref. [257]. for charged species inside the compact
For example, at the p.z.c. (σ = 0) the layer, in particular, for specifically ad-
shape of the profile depends on the ratio sorbed ions. This interphasial region rep-
of the Debye screening length in the bulk resents a nonuniform dielectric medium
solution (32), LD , and the product εLH whose effective dielectric constant, ε(z),
this ratio being identical to the ratio of the has got the values strongly reduced in
compact- and diffuse-layer capacitances, comparison with those in the bulk solu-
CH and CGC in Eq. (20). tion. For such systems, the interaction
For a sufficiently low electrolyte con- of the species with the metal or the
centration (so that CH CGC ), the pro- diffuse-layer ions represents a minor cor-
file within the interval, z LD , is close rection, while the dominant contribution
to its form at vanishing concentration, to the image potential is given by the
Eq. (79), in particular there are the repul- short-range interaction with the dielectric
sion and attraction branches, Eqs. (81) and medium [258, 259]. For ions of a suffi-
(78). For even larger distances from the ciently small radius, ro , one can use as a
metal surface (comparable with the Debye qualitative estimate a simple formula for
length), the interaction with the interface the electrostatic contribution to the image
is screened: potential:
(zi e)2 −2z/LD
Wim (z) = − e (82) (zi e)2 1 1
4εz Wel−stat (z) ∼
= − (85)
2ro ε(z) ε
More important changes take place in
more concentrated solutions where CH Because of lowered values of the effective
CGC (which means c 0.1 M for the dielectric constant inside the compact
Hg–water interface). The profile repre- layer (independent of the physical reason
sents a monotonically decreasing curve leading to this reduction), the term given
(pure repulsion of the ion from the in- by Eq. (85) represents a strong repulsion
terface at all distances), in accordance with of ions from this interphasial region. In
the formula: reality, one should add the effect of the ion
desolvation to enter this layer.
(zi e)2 −2z/LD
Wim (z) = e (83) This consideration gives an explanation
4εz of the existence of surface-inactive ions
Thus the repulsive behavior at shorter for which the attractive contribution due
distances is a general feature of the image to the short-range (electron exchange)
potential profile at the p.z.c., whereas, the metal surface–ion interaction is absent
large-distance variation depends on the or insufficiently strong to compensate the
ionic strength of the solution: aforementioned repulsive effect. It may
also be responsible for the strong influence
(zi e)2 εLH − LD −2z/LD of the iodide adsorption on the hydrogen
Wim (z) = e ,
4εz εLH + LD evolution, despite the absence of marked
z min(LD , εLH ) (84) specific adsorption of H3 O+ ions [260].
102 2 Electrochemical Double Layers
One can also derive analytical expres- and it leads to fundamental consequences
sions for the variation of the image for the correlation properties of this
potential for a charged species inside ensemble as well as for the form of the
the compact layer as a function of the adsorption isotherm (see Sect. 2.1.11.6).
bulk electrolyte concentration [254–256]. Intuitively, it seems to be ‘‘natural’’ to
This contribution is relatively small com- use an analog of Eq. (78) based on the same
pared with the short-range interactions, model of the ‘‘image charge’’ formed by the
but its knowledge is of importance, since metal electrons as a response to the field
it represents the change of the single-ion of the adsorbed ion:
(standard) adsorption energy with varia-
1 1
tion of the ionic strength of the solution in U o (R) = (ze)2 −
the adsorption isotherm, see Sect. 2.1.11.6. εR ε(R 2 + 4a 2 )1/2
2a 2
∼
= (ze)2 3 (86)
2.1.11.4 Interaction Between Adsorbed εR
Ions
The deviation of the adsorption isotherm at Here, a is the distance between the center
a uniform electrode surface from the linear of the adsorbed charged species and the
behavior (Henri isotherm) is related to the metal surface. Equation (86) predicts a
interaction between the adsorbed species. rapid decrease of the interaction energy
A substantiated derivation of its form can as a function of the distance between
only be made on the basis of an analysis of adsorbed ions, R, and it means that the
the statistical-mechanical properties of the ensemble of species with such interaction
whole ensemble of adsorbed ions, which, law should correspond to the one with
in turn, requires the knowledge of the short-range interactions. The uncertain
interaction potential between the ions, U , point within this approach is to choose
as a function of their distance, R, along the value of the dielectric constant of the
the surface. This quantity is defined as medium in Eq. (86), ε.
a difference between the energies of the It is tempting to identify it with the value
system, when these two ions are fixed at typical for the compact layer, which would
distance R or are very far from each other. mean that the interaction propagates along
There are numerous types of short- this interphasial region. However, it can
range interactions between adsorbed ions be easily seen from simple calculations
such as exchange-correlation effects, an that the interaction energy must diminish
overlap of their electronic cores, mutual exponentially (as a function of R), if the
polarization, interaction via the electrons interaction between the charges is realized
of the metal, interaction between their along the lines passing inside the compact
solvation shells (which may partially be layer, for example, in the model in which
retained in the adsorbed state), changing of this layer is surrounded by two conducting
the local structure of the compact layer, and media imitating the metal and the diffuse
so on. All these forces decay rapidly with layer. As a result, this model cannot explain
ion-ion distances exceeding the molecular the very large values of the interaction
scale, and the electrostatic contribution constants observed experimentally for ion
becomes dominant. A crucially important adsorption. The alternative attribution of
feature of this long-range contribution is a ε to the bulk-solvent value, also leads to a
surprisingly slow decay as a function of R, very weak interaction.
2.1 Electrochemical Interfaces: At the Border Line 103
It will be seen from the discussion that state, and so on. In the simplest model of
follows that the electrostatic interaction the compact layer as a uniform dielectric
between adsorbed ions follows quite a slab of thickness zH and dielectric con-
different functional form. A detailed anal- stant εH , this parameter is given by the
ysis of this problem may be found in formula:
za ai
reviews [89, 254, 255, 261]. = (88a)
Similar to Sect. 2.1.11.2, let us consider zi zH
first the interaction of two charged species where ai is the distance between the
located at a small distance (atomic scale) adsorbed ion and the metal surface (con-
from the metal–solvent interface. Once sidered in this formula as a point charge),
again, it turns out that the functional form za and zi are the charges of the ion in
of the U (R) dependence is only influenced the adsorbed state and in the bulk solu-
by the same characteristic length, εLH , tion. A somewhat more general model of
which had already appeared in Eq. (79) for the compact layer as a nonuniform dielec-
the image potential, that is the product of tric slab whose local dielectric permittivity
the bulk dielectric constant of the solvent depends on the normal coordinate, ε(z),
and the inverse value of the compact -layer results in the formula (88b) containing the
capacitance, Eq. (80). ratio of the ‘‘effective thickness’’ for the
The binary interaction potential, U (R), overall compact layer, LH ≡ (4πCH )−1 ,
follows indeed the R −3 law within its long- and that of its ‘‘inner part’’ between the
distance limit: bulk metal and the adsorption plane, LI
2εLH 2 (again, the charge of the adsorbed ion
U (R) ∼
= (zi e)2 2 (87) is considered here as the point charge
R3
localized at the distance a from the sur-
but only for very large distances between face) [262]:
the adsorbed ions, R εLH . One should zH
remember that this characteristic distance, za LI dz
= , where LH = ,
εLH (known for each system from inde- z LH −∞ ε(z)
pendent capacitance experiments), is well ai
dz
beyond the atomic scale, for example, it LI = (88b)
−∞ ε(z)
is about 2.4 nm for any ion adsorbed at
the Hg–water interface. The amplitude in Similar expressions can easily be de-
Eq. (87) is several dozen times larger than rived for any particular model of the
the value predicted by Eq. (86). metal–solvent interphasial structure on
The only parameter in the whole U (R) the basis of the general formula for this
expression that is unknown from inde- parameter [254, 261, 263], including a spa-
pendent experiments is the multiplier tial distribution of the electronic charge of
2 , which depends on numerous factors, the ion in the adsorbed state. Generally,
like, location of the adsorbed ion within one may expect that its value should be
the compact layer, the distribution of the between 0 and 1.
dielectric properties across this layer (ana- On the other hand, this parameter, ,
logues of the ratio of the capacitances, can be found for any particular ion from
KH /KO , in the Grahame–Parsons theory, the experimental data on its adsorption,
Sect. 2.1.11.2), partial transfer of the elec- see Sect. 2.1.11.6. It means that the form
tronic charge by the ion in its adsorbed of the binary interaction potential and the
104 2 Electrochemical Double Layers
correlation properties of this ion ensemble unchanged, including the universal func-
(Sect. 2.1.11.5) can be described without tional form of U (R) with a single unknown
any unknown parameters. parameter, , a high intensity of the in-
Equation (87) only represents a limiting teraction (compared to its estimates within
behavior at very large distances between oversimplified model (86)), existence of the
adsorbed ions, R. There exists a wide ‘‘Coulombic’’ behavior of the potential (89)
range of intermediate distances in which and so on.
the interaction potential follows quite a The only qualitative modification is the
different functional law: appearance of the exponential decay at
2 distances exceeding the screening length
U (R) ∼
= (zi e)2 2,
εR within the diffuse layer, which was as-
for zH R εLH (89) sumed to be not too short (much longer
than the compact-layer thickness). As a
The crucial difference of this formula in result the interval of intermediate dis-
comparison to Eqs. (86) and (87) is a very tances in which the R −1 behavior is valid,
slow decrease of the interaction potential Eq. (89), is determined by the inequalities,
within this interval of R. The existence of zH R R∗ , where R∗ is the minimum
this intermediate asymptotic behavior of of the distance εLH , and the screening
U (R) leads to crucial consequences for the length inside the diffuse layer for the val-
correlation properties of the ion ensemble, ues of its charge and the ionic strength
Sect. 2.1.11.5. of the solution, in particular, the Debye
One can demonstrate that this large
screening length of the bulk solution for
increase of the intensity of the interaction
the zero charge of the diffuse layer.
is due to a specific distribution of the lines
The intensity of the interaction depends
of the electric field created by an adsorbed
also on parameter , whose experimen-
ion. Because of a great difference between
tally found values for the adsorption at
the effective dielectric constants in the
s,p-metals are not much smaller than 1 (see
compact layer and in the bulk solvent, the
lines do not propagate along the compact Sect. 2.1.11.6). For such systems, one can
layer. A fraction of these lines determined see that the ‘‘Bjerrum length’’, LB , where
by the factor 1 − , connect the ion with the interaction energy is equal to the ther-
the charges at the metal surface. The mal energy belongs to the aforementioned
remaining lines (their fraction being ) interval of the intermediate distances,
quit the compact layer and propagate zH LB R∗ . The importance of such a
outside of it at a distance R. The interaction sequence of characteristic distances for the
given by Eqs. (87) and (89) is determined correlation properties of the system will be
by the latter fraction of lines. seen in the next section.
A similar analysis was carried out for
the case of an electrochemical system 2.1.11.5 Statistical Mechanics of Adsorbed
in which the ion is adsorbed at the Ion Ensemble
charged metal–electrolyte solution inter- During an extended period the structure
face. The derived analytical expressions of the two-dimensional (2D) ensemble
for the interaction potential become a bit formed by adsorbed ions located in the
more complicated [89], but all the above- inner Helmholtz plane was postulated on
formulated qualitative conclusions remain intuitive grounds.
2.1 Electrochemical Interfaces: At the Border Line 105
Most often, it was assumed that the ion the form of the U (R) dependence in
distribution represents a regular 2D lattice, Sect. 2.1.11.4 confirm the existence of such
hexagonal or square, owing to the repulsive a ‘‘cut-off-disk’’ but its size does not depend
electrostatic forces between the ions. How- on and is markedly different from the
ever, the theoretical studies (confirmed by postulated values.
experimental data) related to a similar 2D These assumptions on the distribution
one-component system of charged species, of ions in the adsorption plane were com-
electrons at the surface of liquid helium, bined with some expression for the binary
have shown that such a regular lattice interaction potential, U (R) (in most cases,
becomes unstable with respect to certain given by Eq. (86)), to get the contribution
collective motions (dislocation formation, from the interactions into the chemical po-
etc) at extremely low temperatures, that is, tential of an adsorbed ion. However, the
under the conditions in which the average final results of such treatment can hardly
interaction energy per ion is still much be considered as reliable.
larger than the thermal energy (contrary A substantiated analysis must be based
to the standard stability criterion for 3D on the dependence of the binary interac-
lattices). The application of this stability tion energy on the distance between ions,
condition for 2D lattices to the ensemble U (R), discussed in Sect. 2.1.11.4. Certain
of adsorbed ions shows that the hypothet- specific features of this system compared
ical regular lattice never formed in these with usual 3D ones originate immediately
systems by electrostatic repulsion within from its two-dimensional character [264,
the experimental temperature range, since 265], e.g., the absence of a long-range crys-
the 2D density is limited by the repulsion talline order at any finite temperature, slow
between the cores of these ions. decay of the autocorrelation function of the
Another hypothesis assumed that the species velocity in the long time limit, and
binary interaction between ions is weak, so forth.
so that the system could be considered Additional effects arise from the modifi-
as close to a 2D ideal gas. However, cation of the form of the binary interaction
the analysis of the function, U (R), has energy, U (R), for the ions at the interface,
revealed that this energy exceeds the compared to the one in the bulk solution.
thermal one up to the distance about One of the most important effects is the
RB ∼ 1 nm, that is, it is strong. great intensity of this interaction, for ex-
The third popular approach was based ample, Eqs. (87) and (89), which may in
on a cut-off-disk model. According to it certain cases be stronger than the same
the electrostatic repulsion between the characteristic in the bulk phase (dielec-
ions leads to the formation of an area tric or solution). Another crucial feature
around each ion practically free from is a slow decrease of this interaction,
other adsorbed ions. The size of this area U ∼ R −1 , in the intermediate range of
was identified with the average distance the distances, zH R R∗ , with a much
between the ions, so that it was dependent steeper diminution outside at greater dis-
on the 2D density, that is, on the ion tances, Eqs. (89) and (87), correspondingly.
adsorption, (as in the previous Sects. This combination makes it impossible to
2.1.11.2–2.1.11.4, this symbol is used for apply standard approaches of statistical
the amount of specifically adsorbed ions, mechanics derived separately for systems
unlike in Sect. 2.1.11.1). The results on with a strong short-range or weak long-range
106 2 Electrochemical Double Layers
binary interaction potential. Besides, one interval are close to those of a 2D one-
must take into account the screening prop- component plasma, despite the absence
erties of the diffuse layer, which depend of the R −1 behavior of U (R) at very long
markedly on the electrode charge and ion distances. It would be of great interest in
force of the solution. the future to perform experimental studies
The statistical-mechanical analysis of of these predictions with the use of small-
correlation properties of such an ion angle scattering techniques, in particular
ensemble at the uniform metal surface of the anomalous behavior at small Fourier
was performed in Ref. [266], see also vectors owing to the plasma correlations.
reviews [254, 255, 261, 267]. There are At even higher system density, ∼ B ,
two characteristic values of the 2D den- the interaction energy between a pair of
sity, = ∗ ∼ (πLB R∗ )−1 and = B ∼ adsorbed ions located at the average dis-
(πL2B )−1 corresponding to the intervals: tance becomes comparable to the thermal
0.05–0.5 nm−2 and 0.5–2 nm−2 , respec- energy, kB T . Then the properties of the
tively. system correspond to those of a 2D ir-
Owing to the rapid decrease of the regular condensed state. Ultimately, when
interaction potential at large distances, the approaching the monolayer coverage de-
system behaves like a 2D dilute gas with termined by the short-range repulsion of
a weak interaction between the species ion cores or/and their solvating shells, a
if the surface coverage is extremely low, typical 2D ‘‘solid’’ is formed, which rep-
∗ . The virial expansion may be used resents a crystal lattice possessing a quasi
to determine all energetic and correlation long-range ordering.
characteristics. However, the coefficients
of this virial expansion turned out to be 2.1.11.6 Ion Adsorption Isotherm
very large (compared with their estimates In a surprising contrast to this complicated
with the use of the Bjerrum length). As picture for the correlation properties of the
a result a noticeable repulsion between ensemble of adsorbed ions (Sect. 2.1.11.5),
adsorbed ions in the intermediate distance the same analysis has resulted in a very
range takes place already at very low ion simple formula for the average interaction
densities, ∼ ∗ . energy, that is, for the contribution to the
Within this transitional interval of den- chemical potential of adsorbed ions, µint ,
sities, ∼ ∗ , the correlation properties which is given as a linear function of the
of the system are subject to a gradual coverage over the whole range of surface
change. For lower coverages, the princi- coverages:
pal contribution to the average interaction
energy is due to pair ‘‘collisions’’ of ions. µads = kB T ln + µint + constant,
On the contrary, at higher densities, µint ∼
= 2vkB T (90)
∗ , new collective plasma-like excitations
arise, leading to the screening of the inter- The coefficient in this linear relation serves
actions at the distance R . This new 2D as an analogue of the attraction con-
screening length depends on the density stant in the Frumkin isotherm (with the
of the system, R ∼ (LB )−1 , that is, the minus sign) or the constant B in the
increase of the coverage results in a more Grahame–Parsons isotherm (75). How-
effective screening. Thus the correlation ever, unlike these models, in which this
properties of this ensemble within this term as well as its dependence on the
2.1 Electrochemical Interfaces: At the Border Line 107
electrical variable (see following text) were electrode potential, the adsorption energy
introduced phenomenologically, Eq. (90) and the attraction ‘‘constant’’ as functions
was derived on the basis of the binary of this variable being modeled by the sim-
interaction potential, U (R), so that the plest approximation (linear or quadratic).
constant, v, is related immediately to this In the theory of Grahame and Parsons,
characteristic: Eq. (75), it is the electrode charge.
∞ According to the derivation of isotherm
−U (R)
v=π 1 − exp RdR (93), the ions are transferred into the
LB kT adsorbed state at a fixed value of the
(91)
electrode potential, E, so that the parameters
The introduction of the expression for
of the isotherm, Wads and v, depend on
the binary interaction potential discussed
this variable, as well as on the overall bulk
in Sect. 2.1.11.4 into this equation leads
concentration, c, in the case of the mixed
to the formula (92) for the ‘‘repulsion
electrolyte:
constant’’, v:
◦
Wads = WEDL + Wim c + Wads ,
zi2 e2
v = 2 v̄, v̄ = (92) WEDL = zi e(E − ϕGC )(1 − )
2kB T (CH + CGC )
In this formula the coefficient, v, is + zi eϕGC (94)
represented as a product of the factor
In other words, the single-ion adsorp-
2 introduced in Sect. 2.1.11.4, and a
tion energy is represented as the sum
function of the compact- and diffuse-layer
of the EDL term (compare Eq. (74)),
capacitances (the GC expression for the the image potential as a function of
latter, Eq. (21a), has been used in this the electrolyte concentration, Wim c =
derivation). Wim (z; σ, c) − Wim (z; σ, 0) and the stan-
The combination of Eq. (70a) and (90) dard adsorption energy, Wads ◦ . The an-
results in the adsorption isotherm (93) for alytical expression for Wim c is given in
a mixed electrolyte solution [267, 268]: Ref. [256, 267, 268]. All these expressions
Wads including the capacitances in Eq. (92) and
ln(mc) = + ln + 2v (93) the diffuse-layer potential difference, ϕGC ,
kB T
in Eq. (94) depend on the electrode po-
where Wads is the single-ion adsorption tential; these dependences correspond to
energy corresponding to the energy of ion zero specific adsorption so that they should
transfer to the surface in the absence of be taken from experimental data for the
other adsorbed ions. surface-inactive electrolyte.
Similar to several other models, isotherm After this treatment the isotherm only
(93) has the virial form, with a linear de- contains two microscopic parameters of
pendence in the interaction term. The the adsorbed ion, whose values cannot
essential difference between them is re- be measured without experiments on
lated to another driving parameter of the its specific adsorption, Wads ◦ and ,
system, which is called the electrical vari- which may generally be dependent on
able and which characterizes the state the electrode potential. The theoretical
of the electrode while the bulk concen- expression for (in particular, its simplest
tration of ‘‘active’’ species changes. The form (88a)) shows that its value is expected
Frumkin isotherm is valid for a fixed to be between 0 and 1; it should diminish,
108 2 Electrochemical Double Layers
0.175
0.15
0.1 0 − 0.1
ϕ
[V]
Fig. 22 Experimental values of the ‘‘repulsion constant’’, v, in
isotherm (93) for iodide adsorption at Bi-water interface (points are
shown with their dispersion) as a function of the electrode potential
in the rational scale, ϕ ≡ E − Eσ =0 . The theoretical curve (solid
line) was calculated from Eq. (92) for = 0.71 with the use of
experimental data for the compact- and diffuse-layer capacitances
of the same interface in a surface-inactive electrolyte solution.
2.1 Electrochemical Interfaces: At the Border Line 109
− 10
−5
ϕ
[V]
Fig. 23 Dependences of the total single-ion adsorption energy,
Wads /kB T (curve 1), and of the sum of the ‘‘specific’’ and image
contributions, (Wads ◦ + Wim c )/kB T(curve 2) for the system,
Bi/H2 O−I− , as a function of the electrode potential,
ϕ ≡ E − Eσ =0 . The values of Wads were found from the fitting of
isotherm (93) to experimental adsorption data for each electrode
potential. Then the values of Wads ◦ + Wim c were obtained with
the use of Eq. (94) and the data for surface-inactive electrolyte
solutions. The EDL term, WEDL , was calculated for the same value
of the adsorption parameter, = 0.71, see Fig. 22.
110 2 Electrochemical Double Layers
Then the value of was determined chemical bonds. It confirms the principal
from the experimental point for the most conclusion of Sect. 2.1.11.4 concerning the
positive potential, with the use of the value binary interaction potential, namely, that
of v̄ found from Eq. (92) and experimental the dominant role is played by the inter-
data for the Bi–water interface in the action along the field lines passing across
background electrolyte solution. This value the diffuse layer, rather than those along the
was used to calculate, already without any compact layer.
unknown parameter, the whole theoretical The values of v for each particular system
curve for v (Fig. 22) as well as for demonstrate a marked variation with the
Wads ◦ + Wim c (Fig. 23) as functions of ϕ. potential, which can be described by
One can notice that a constant value of Eq. (92) with a constant value of , that is,
is sufficient to reach an agreement with all this dependence on the electrode potential
experimental points for v, within their dis- is given by the theoretical expression,
persion, which is very small for the positive v̄(E, c), which does not contain any
potential range. Similarly the single-ion ad- unknown parameter.
sorption energy (after removal of its EDL As another success of the approach, one
component) is practically constant within may notice a large variation between the
the same range of potentials. Thus, in values of v for different systems, up to
accordance with expectations, the micro- a factor of 5, while the estimates of the
scopic characteristics of the adsorbed ion, principal microscopic parameter of the
I− , do not change markedly in this region adsorbed ion, , turned out to vary within
of electrode potentials, despite the change a much more narrow range, from 0.55 to
of the electric field created by the charged 1.0. Moreover, all these values fall inside
metal surface. the expected interval < 1. If one uses
The data at high negative potentials formula (88a) or (88b) to get a simple
show a significant reduction of the ad- qualitative estimate of this parameter, one
sorption energy with the potential. It also may conclude on the basis of these results
corresponds to the qualitative expectations that all these ions are located not too close
concerning the weaker bonding of an an- to the metal surface, and that they retain
ion at negatively charged metal surface. most (if not all) of their electric charges in
Similar results have been obtained for the adsorbed state, za ∼ zi .
the other systems listed in the preceding Within the framework of the hypothe-
text. In particular, the values of the ‘‘re- sis that no charge transfer takes place in
pulsion constant’’, v, were found within the course of the ion transition into the
the limits of 0.6 and 3.5 nm2 per ad- adsorbed state, the value of for each sys-
sorbed ion. These results represent an tem would allow one to determine the ratio
additional confirmation of the predic- of the capacitance between the metal sur-
tion of strong intensity and long-range face and the adsorption plane to the overall
of the binary interaction potential dis- compact-layer capacitance, Eq. (88b), sim-
cussed in Sect. 2.1.11.4. One should keep ilar to the Grahame–Parsons treatment
in mind that the attempt to interpret the in Sect. 2.1.11.2. (the Helmholtz–Stern
same experimental data with the use of model of the compact layer as a uniform
the ‘‘single-image’’ expression (86) would dielectric slab does not correspond to the
result in the values of the ion-metal dis- modern view of the interfacial structure,
tance, ai , well beyond the usual length of see previous sections, so that it is hardly
2.1 Electrochemical Interfaces: At the Border Line 111
possible to use Eq. (88a) to estimate the • for the same electrode, InGa, the
real distance between the adsorbed ion and stronger adsorbed anion, iodide, has
the metal surface.) However, these adsorp- got a smaller value of this parameter,
tion data cannot give any argument in favor than bromide;
of or against such a partial charge transfer • in the case of a weakly adsorbed cation,
in these systems, in other words, the re- cesium, there is no significant differ-
duced value of this adsorption parameter, ence between ethylene and propylene
< 1, represents generally a combined carbonate solutions.
effect of partial charge transfer from the
adsorbed ion to the metal and of the loca- Despite the success in the interpretation
tion of the adsorbed ion inside the compact of these experimental data for small-size
layer, see Eq. (88b) as an illustration. inorganic ions, one should keep in mind
The only exception is the case of the the need for the further development
Hg–H2 O-Cl− interface [267, 271] in which of the theory, in particular to include
the value was close to 1. It allows one the effect of the gradual change of the
to conclude on the absence of a notice- dielectric properties of the compact layer
able charge loss by this ion as well as on with the increase of the surface coverage
close values of the characteristic lengths, by adsorbed ions.
LH and LI in Eq. (88b), that is, of the corre-
sponding capacitances. However, the latter 2.1.12
does not mean that the physical location Electric Field-Induced Modifications of the
Surface Structure
of the adsorbed ion is close to the outer
Helmholtz plane. There are various indi-
Charging the electrode may not only affect
rect indications (in particular, presented in
the polarization of molecular dipoles on
previous sections) that the effective dielec-
the surface, deform the surface electronic
tric permittivity, ε(z), is quite nonuniform
profile, and induce the adsorption of ions.
across the ‘‘compact layer’’, its lower values
It can also induce surface reconstruction
being within the ‘‘tail’’ of the metal elec-
and roughening transitions, that is, modify
trons while the external part of the compact
the atomic structure of the metal surface.
layer possesses a much higher permittivity
In this section, we will briefly describe the
(but still much lower than the one in the
principles of the statistical theory of this
bulk solvent). Then the adsorbed ion can
phenomenon [272–274].
be localized at the distance of 0.1 to 0.2 nm
In ultrahigh vacuum (u.h.v.), clean low-
from the ion skeleton of the metal, that is,
index single-crystal surfaces of gold and
deeply inside the compact layer having a
platinum are known to be reconstructed
thickness of about 0.3 nm, Eq. (12), but
at low temperatures and unreconstructed
can be close to 1.
at temperatures above the critical tem-
The comparison of the values for
perature [275]. Transition from the recon-
different systems have shown expected
structed to the high-temperature unre-
tendencies:
constructed state is called deconstruction.
• this parameter for adsorbed iodide is This term does not necessarily imply that
greater at the Bi electrode than at InGa the surface structure at high temperature
(the adsorption energy is greater for resumes the full 1 × 1 unreconstructed
InGa); symmetry. Indeed the deconstruction is
112 2 Electrochemical Double Layers
solutions. At potentials more negative than emerges by removing atoms from holes,
−0.3 V versus. s.c.e., they report the ap- which grow into missing rows, and redepo-
pearance of 1 × 2 reconstruction (with sition of these atoms into the added rows.
small impurity of 1 × 3 and 1 × 4 blocks). Thus qualitatively the STM results of
Unreconstructed patterns emerge at po- Behm’s and Weaver’s groups agree with
tentials larger than zero; Hamelin [278] each other. Simultaneously Ocko and
reported a threshold value of 0.1 V. coworkers [280] examined the same sur-
Since the p.z.c. for this system is about faces in 0.1 M perchloric acid and alkali
−0.02 V [278], at zero charges the surface halide salts solutions by a ‘‘more objective’’
is not reconstructed in contrast to the same in situ synchrotron diffraction method. For
surface in u.h.v. at room temperature. The a meticulous discussion of this work, see
authors of Ref. [277] speculate, that this the review [273]; here we just stress that
might be due to some adsorption of cations the results, although less unambiguous in
used to stabilize the reconstructed surface a variety of details, were qualitative similar
in u.h.v., which plays a role similar to neg- to the results of STM studies.
ative charging. Between −0.3 and 0 V the
interpretation of the images is unclear. 2.1.12.2 Driving Force for Reconstruction
The same surfaces were studied by The structure and the shape of the metal
Magnussen and coworkers [279] in 0.1 M crystal are determined by the net bal-
acid solutions: HClO4 , H2 SO4 – pure and ance of forces. In the bulk of the noble
with traces of Cl− . The Cl− -free so- metal, the stabilizing force is the com-
lutions showed no dependence on the pressive stress of s-p electrons balanced
nature of electrolyte, but even small ad- by repulsive interaction of the filled d-
ditions of Cl− lifted reconstruction. The shells. The s-p electrons play the role of
reconstructed state was observed at poten- a ‘‘glue’’, which is as follows. They always
tials from 0.05 down to −0.3 V, although ‘‘like’’ to spread out along the surface.
less ordered than in u.h.v. A window However, at the (110) surface the external
of disordered and roughened states was filed of the skeleton ions is very inho-
found between 0.05 and 0.25 V, while at mogeneous. The electrons will have to
more positive potentials the 1 × 1 unrecon- form rather inhomogeneous distributions,
structed phase prevails. For slow variation which are energetically costly. This forces
of potential, the authors were able to follow surface atoms to rearrange to form tightly-
the initial steps of reconstruction. They re- packed facets that mimic the lower energy
ported the transformation beginning near (111) structure. On (110) surfaces, this is
0.2 V via a slight roughening of the step achieved by missing row (Fig. 24) recon-
edges; further potential decrease induced struction [281, 282]. The force that pushes
roughening and zigzag step structure of the surface to reconstruct is stronger when
the reconstructed surface with a simul- more electrons are present in the surface
taneous growth of anisotropic holes and layer. This happens when the surface is
islands one atom wide but up to 10 nm negatively charged.
long in the 110 direction! When averaged The field-induced reconstruction was
over a big terrace, the concentration of studied in the ground state Kohn–Sham
islands was found to be nearly the same electron density functional simulations for
as the concentration of holes, in accor- fixed atomic structures of the surface lay-
dance with the idea that the reconstruction ers [283, 284]. Their energies are compared
114 2 Electrochemical Double Layers
1.0
D
k BT Jy−1
0.5
(1 × 2) (1 × 1)
0.0
−1.0 − 0.5 0.0 0.5 1.0
a (σ − σ 0) Jy−1
Fig. 25 Ising model phase diagram. The reconstructed (1 × 2) and
unreconstructed (1 × 1) regions are separated by Ising disordered
phase (D).
116 2 Electrochemical Double Layers
1.0
r
R
D
r
i
k BT Jy−1
0.5
(1 × 2) (1 × 1)
0.0
− 0.50 − 0.25 0.00 0.25 0.50
a (σ − σ 0) W0−1
Fig. 26 Phase diagram of the Kornyshev and Vilfan theory
exhibits a line of an Ising order-disorder transition and two lines
of roughening transitions, with disordered, D, and rough, R,
phases in between.
2.1 Electrochemical Interfaces: At the Border Line 117
based on a free Fermion analogy, having • Oil–water interface [289, 290], impor-
again utilized the relationship of the tant for oil extraction and catalysis.
charge dependence of the basic param-
eters of the theory with E. Not going The size of this chapter will not allow us to
any further into the details of the theory consider all these systems. We will there-
(for tutorial presentation see the review fore discuss the last one, well studied in
article [273]), we draw in Fig. 26 the re- electrochemistry, that is the interface be-
sulting phase diagram. While the Ising tween two immiscible electrolytes (ITIES).
transition has a second order, the roughen- This is a biomimetic system and a model
ing transitions presumably are of Koster- system for reaction kinetics, but also an
actual medium for phase transfer catal-
litz–Thouless universality class (see the
ysis. In this system, one of the solvents
discussion in Ref [273]).
is usually water, and the other one is a
In Ref. [273] the role of field-induced
low polar organic liquid (oil), such as, for
adsorption of anions together with a sim-
example, nitrobenzene or dichloroethane.
ple model of its incorporation into the
This interface separates ‘‘hydrophilic’’ and
theory is thoroughly discussed. Together
‘‘hydrophobic ions’’, such as simple alkali-
with the analysis of the possible effects
halides and organic ions, respectively.
due to the presence of water [273], all When two such salts are dissolved in this
this provides a framework for under- system, one composed of hydrophilic and
standing the interesting phenomenon of the other one of hydrophobic ions, they
electrochemically induced missing row form back-to-back EDLs and the interface
reconstruction. can be polarized [289, 290]; standard elec-
trochemical methods can be used for the
2.1.13
characterization of this interface. The re-
Soft Electrochemical Interfaces
sulting electric field across the interface
affects a set of phenomena that occur at
Electrochemistry of soft interfaces is a the interface. A controllable variation of
rapidly progressing interdisciplinary field. the potential drop across this interface can
The most important systems here are as shed extra light on these phenomena.
follows: First models of liquid–liquid interfaces
• Polymer electrolyte–aqueous electro- in electrochemistry treated them as flat
lyte interfaces in the porous interior and sharp [289, 290]. The nonlinear capac-
of proton conducting or ion exchange itance due to two back-to-back ionic double
membranes for separation, desalina- layers at such an interface is given simply
tion, and fuel-cell applications [287]. by [289]
• Interfaces between aqueous phase and 1 1 1
= + (105)
the volumes confined by amphiphilic CGC C1 C2
molecules [288]. In vitro, these refer (1/2)
to lipid vesicles and micelles, lipid εi κi 4π 2 σ 2
Ci = 1+ , i = 1, 2
lamellae, cubic and hexagonal phases, 4π (εi κi )2
Langmuir–Blodgett (LB) films, which (106)
are important in colloid science and in where σ is the surface density of the charge
extraction technology. In vivo, these are accumulated in each double layer per unit
the surfaces of biological membranes. area of a flat interface.
118 2 Electrochemical Double Layers
can be easily done in consistence with the the harmonic approximation breaks down,
already used harmonic approximation for as they, essentially, reflect the electric field-
Fcw [ξ(k)] (Eq. 109). Then all characteristics induced destruction of the surface.
of the interface are obtained in a closed Indeed the potential-dependent FT of the
form. For voltage drops smaller than kB T , height–height correlation function now
the results look relatively compact [308, reads
309], but for larger voltages they are very kB T
cumbersome [295]. g(k, V ) = (122)
γ [k 2 + kgr
2 ] − f (k, V )
The result for the electric capacitance
reads where f (k, V ) is a very complicated
but elementary function of its argu-
C = CGC (V )R(κ1 , κ2 , V ) (121)
ments [295], which grows with |V |. For
where V = eU/kB T , CGC (V ) is the GC moderate |V |, it is proportional to V 2 ;
capacitance (Eqs. (105) and (106)) and for instance, for small wave vectors the
R(κ1 , κ2 , V ) is the so-called roughness expression is particularly simple,
function, the factor in which all the 1 2 2 (0)
deviations from GC are accumulated; f (k, V ) ≈ k V CGC (k κ1 , κ2 )
2
when γ → ∞, R → 1. The general (123)
(0)
expression for R̃(κ1 , κ2 , V ) [295] is too where CGC ≡ CGC |σ =0 .
cumbersome to repeat here, but its The harmonic approximation breaks
typical graph is shown in Fig. 27. This down when the denominator in Eq. (122)
graph shows that the roughness function approaches zero, but the tendency towards
increases rapidly with V because the dramatic enhancement of fluctuations is
surface area increases due to a dramatic correct. This translates into the depen-
enhancement of the amplitude of capillary dence of the mean square height of
waves. The steepest wings in Fig. 27 2 , for which there is a
corrugation, dcw
should be ignored, because in this region full expression [295], but here we quote
20
18
16
14
1
12
2
R
~
10 3
8
6
4
2
0
−15 −10 −5 0 5 10 15
eE (kBT )−1
Fig. 27 Roughness function versus potential. ε2 = 10;
γ = 30 dyn cm−1 ; κ2−1 , nm: (1) 1; (2) 2; (3) 5.
122 2 Electrochemical Double Layers
only its simplified version for the case of capacitance curves show a strong asym-
moderate V : metry as a function of the potential. The
V 2 kB T (0) discrepancy between experimental results
2 (0) 2
dcw ≈ (dcw ) + C and the GC theory has stimulated theoret-
4πγ 2 GC
ical work that went beyond the classical
κ1 κ2
× ln (124) scheme. The first step in this direction
(κ1 + κ2 )kgr has been done in Refs. [317–319], in
(0) which the ‘‘mixed boundary layer’’ was
where dcw is the root mean square
introduced and the effect of overlapping
height at the p.z.c. Equation (124) shows
of the two space-charge regions on the
unambiguously that charging the interface
double-layer capacitance has been consid-
enhances surface corrugation. The general
ered. The authors used a quasi-chemical
expression deviates from this law for
approximation (QCA) [317, 318] and MC
V > 4 showing a stronger enhancement.
The dispersion law is also modified by simulations [319] in order to reveal the
charging the interface. It now reads [295, density profiles of the ions (the Pois-
308], son–Boltzmann equation has been solved
numerically, assuming, for computational
ω2 = −2iω(k) − ωc2 (k) simplicity, that the dielectric constants of
k the two solvents are equal to each other
+ f (k, V ) (125) and do not vary with the distance from
ρ1 + ρ2
the interface). Numerical studies of the
that is, the charging reduces the frequency capacitance [317–319] reveal the effect of
of capillary waves. the mixed layer. They, however, do not dis-
However, such a simple theory has its close general laws for the capacitance of
own limitations. One of them is that relax- this interface, leaving open the following
ation of the Debye plasma cannot follow questions:
the high frequency capillary waves, and
their contribution to surface corrugation • Which characteristics determine the
will not be ‘‘seen’’ by electrolyte ions. In observed capacitance dependence on
Ref. [295], one can find a discussion of the the nature of ions?
yet unsolved dynamic aspects of the prob- • Which parameters determine the sign
lem and suggestion for a correction, taking of the deviation of the capacitance from
into account the incomplete adiabaticity. the Gouy–Chapman result and the
asymmetry of the capacitance curves
2.1.13.2 Ion Penetration into an as a function of potential?
‘‘Unfriendly Medium’’ • What information on the ionic profiles
The treatment of interfacial capacitance in and interfacial structure in general can
terms of the capacitance of two ‘‘back- we gain from the capacitance data?
to-back’’ GC double layers works fairly
well for some electrolytes, while it fails Furthermore, smearing of the interface
for the majority of others [289, 311–316]. between two solvents should cause a
In contrast to the predictions of the sim- smooth variation of dielectric properties
ple GC theory, it was found that the across the interface (if it were sharp,
capacitance of ITIES depends on the na- similar effects would emerge due to
ture of the ions [311, 313–318]. Often the nonlocal dielectric polarizability [92–94,
2.1 Electrochemical Interfaces: At the Border Line 123
3.0
0.3
2.5
C/C 0GC
2.0 0
Gouy-Chapman
1.5
−0.3
1.0
−4 −2 0 2 4
E (k BT )−1
Fig. 28 The effect of penetration of a hydrophobic positive ion into the
aqueous solution on the potential dependence of the normalized
0
capacitance, C/CGC . Curves correspond to the GC result obtained for
L1 = L2 = L3 = 0, and to the indicated values of the parameter
+∞
L+ +
2 = −∞ dz[exp(−g2 (z)) − θ(z)] in nm. The figure is plotted for the
case of no penetration of other ions through the interface, and no
smearing of the dielectric profile. ε1 = 80, ε2 = 10,
κ1 = κ2 = 0.3 nm−1 .
12. O. A. Petrii, G. A. Tsirlina, Electrode poten- 31. O. Stern, Z. Elektrochem. 1924, 30, 508.
tials, Thermodynamics and Electrified In- 32. D. C. Grahame, Chem. Rev. 1947, 41, 441.
terfaces, Encyclopedia of Electrochemistry 33. A. Hamelin in Modern Aspects of Electro-
(Eds.: E. Gileadi, M. Urbakh), Wiley-VCH, chemistry (Eds.: B. E. Conway, R. E. White,
Germany, Vol. 1, Chap. 1. forthcoming. J. O’ M. Bockris), Plenum Press, New York,
13. B. B. Damaskin, O. A. Petrii, Specific ad- 1985, p. 1, Vol. 16.
sorption, Thermodynamics and Electrified 34. M. A. Vorotyntsev in Modern Aspects of
Interfaces, Encyclopedia of Electrochemistry Electrochemistry (Eds.: J. O’ M. Bockris,
(Eds.: E. Gileadi, M. Urbakh), Wiley-VCH, B. E. Conway, R. E. White), Plenum Press,
Germany, Vol. 1, Chap. 3. forthcoming. New York, 1986, p. 131, Vol. 17.
14. O. A. Petrii, Electrochim. Acta 1996, 41, 35. D. C. Grahame, J. Am. Chem. Soc. 1954, 76,
2307. 4819.
15. M. A. Vorotyntsev, Yu. A. Ermakov, V. S. 36. D. C. Grahame, Techn. Report no 14 to
Markin et al., Russ. J. Electrochem. 1993, 29, the office of Naval Research of Febr. 18,
513. 1954. Deposited at the Library of Congress,
16. M. A. Vorotyntsev, A. A. Rubashkin, J. P. Washington, D.C.
Badiali, Electrochim. Acta 1996, 41, 2313. 37. R. Parsons, F. G. R. Zobel, J. Electroanal.
17. A. N. Frumkin, Electrocapillary Phenomena Chem. 1965, 9, 333.
and Electrode Potentials, Commerce typogra- 38. M. A. Vorotyntsev, in Advances in Science
phy, Odessa, 1919. & Technology, VINITI, Moscow, 1984, p. 3,
18. A. N. Frumkin, Potentials of Zero Charge, Vol. 21.
Nauka, Moscow, 1979. 39. R. Gonzalez, F. Sanz, Electroanalysis 1997,
19. S. Trasatti in Comprehensive Treatise of 9, 169.
Electrochemistry (Eds.: J. O’ M. Bockris,
40. B. B. Damaskin, Sov. Electrochem. 2001, 37,
B. E. Conway, E. Yeager), Plenum Press,
745.
New York, 1980, p. 45, Vol. 1.
41. A. N. Frumkin, N. B. Grigoryev, Sov. Elec-
20. E. Lust, Double layers at single crystal
trochem. 1968, 4, 533.
and polycrystalline electrodes, Thermody-
42. A. A. Kornyshev, M. A. Vorotyntsev, Surf.
namics and Electrified Interfaces, Encyclo-
Sci. 1980, 101, 23.
pedia of Electrochemistry (Eds.: E. Gileadi,
43. A. A. Kornyshev, M. A. Vorotyntsev, Can. J.
M. Urbakh), Wiley-VCH, Germany, Vol. 1,
Chap. 2.4. forthcoming. Chem. 1981, 59, 2031.
21. S. Trasatti, O. A. Petrii, Pure Appl. Chem. 44. A. A. Kornyshev, W. Schmickler, M. A.
1991, 63, 711. Vorotyntsev, Phys. Rev. B 1982, 25, 5244.
22. S. Trasatti, E. Lust in Modern Aspects 45. B. B. Damaskin, N. V. Nikolaeva-Fedoro-
of Electrochemistry (Eds.: R. E. White, vich, Zh. Fiz. Khim. 1962, 36, 1483.
B. E. Conway, J. O’ M. Bockris), Kluwer 46. R. Payne, J. Electroanal. Chem. 1975, 60, 183.
Academic/Plenum Publishers, New York, 47. W. R. Fawcett, M. Opallo, Angew. Chem.,
London, 1999, p. 1, Vol. 33. Int. Ed. 1994, 33, 2131.
23. D. C. Grahame, J. Chem. Phys. 1948, 16, 48. W. R. Fawcett in Electrocatalysis (Eds.:
1117. J. Lipkowski, P. N. Ross), Wiley-VCH, New
24. D. C. Grahame, B. A. Soderberg, J. Chem. York, 1998, p. 323.
Phys. 1954, 22, 449. 49. A. N. Frumkin, Z. Phys. Chem. 1933, 164,
25. M. A. Vorotyntsev, J. P. Badiali, G. Inzelt, J. 121.
Electroanal. Chem. 1999, 472, 7. 50. A. N. Frumkin, G. M. Florianovich, Dokl.
26. D. C. Grahame, R. Parsons, J. Am. Chem. Akad. Nauk SSSR 1951, 80, 907.
Soc. 1961, 83, 1291. 51. G. M. Florianovich, A. N. Frumkin, Zh. Fiz.
27. A. N. Frumkin, R. V. Ivanova, B. B. Khim. 1955, 29, 1827.
Damaskin, Dokl. Akad. Nauk SSSR 1964, 52. A. N. Frumkin, in Advances in Electrochem-
157, 1202. istry and Electrochemical Engineering (Ed.:
28. B. M. Grafov, V. V. Elkin, Sov. Electrochem. P. Delahay) Intersci. Publ., New York,
1990, 26, 1015. Vol. 1, 1961, p. 65; Vol. 3, 1963, p. 287.
29. G. Gouy, J. Phys. Radium 1910, 9, 457. 53. L. I. Krishtalik, Electrochim. Acta 1965, 13,
30. D. Chapman, Philos. Mag. 1913, 25, 475. 1045.
2.1 Electrochemical Interfaces: At the Border Line 127
54. P. Delahay, Double Layer and Electrode Ki- Electrodes, Tartu University, 1981, p. 146,
netics, Intersci. Publ., New York, London, Vol. VI.
Sydney, 1965. 76. M. A. Vorotyntsev, Sov. Electrochem. 1981,
55. K. Asada, P. Delahay, A. Sundaram, J. Am. 17, 472.
Chem. Soc. 1961, 83, 3396. 77. M. A. Vorotyntsev, Sov. Electrochem. 1981,
56. O. A. Petrii, A. N. Frumkin, Dokl. Akad. 17, 835.
Nauk SSSR 1962, 146, 1121. 78. M. A. Vorotyntsev, Sov. Electrochem. 1981,
57. A. N. Frumkin, O. A. Petrii, N. V. Nikolaeva- 17, 162.
Fedorovich, Electrochim. Acta 1953, 8, 177. 79. J. Lecoeur, C. R. Acad. Sci. (Paris) 1976,
58. R. R. Dogonadze, A. M. Kuznetsov, M. A. 283C, 651.
Vorotyntsev, Croat. Chim. Acta 1972, 44, 80. A. Hamelin, Sov. Electrochem. 1982, 18,
257. 1413.
59. A. N. Frumkin, N. V. Nikolaeva-Fedorovich, 81. L. I. Daikhin, A. A. Kornyshev, M. I. Ur-
N. V. Berezina et al., J. Electroanal. Chem. bakh, Phys. Rev. E 1996, 53, 6192.
1975, 58, 189. 82. L. I. Daikhin, A. A. Kornyshev, M. I. Ur-
60. D. I. Dzhaparidze, V. V. Shavgulidze, Sov. bakh, Electrochim. Acta 1997, 19, 2853.
Electrochem. 1973, 9, 1390. 83. L. I. Daikhin, A. A. Kornyshev, M. I. Ur-
61. N. V. Fedorovich, M. D. Levi, A. V. bakh, J. Chem. Phys. 1998, 108, 1715.
Shlepakov, Sov. Electrochem. 1977, 13, 89. 84. M. Abramovitz, I. Stegun, (Eds.), Handbook
62. R. R. Nazmutdinov, I. V. Pobelov, G. A. of Mathematical Functions, Dover, New
Tsirlina et al., J. Electroanal. Chem. 2000, York, 1965.
491, 126. 85. B. B. Mandelbrot, The Fractal Geometry of
63. R. R. Nazmutdinov, G. A. Tsirlina, Yu. I. Nature, Freeman, New York, 1982.
Kharkats et al., J. Phys. Chem. B 1998, 102, 86. E. Lust, A. Jänes, V. Sammelselg et al., Elec-
677. trochim. Acta 1988, 43, 373.
64. R. R. Nazmutdinov, G. A. Tsirlina, O. A. 87. E. Lust, A. Jänes, V. Sammelselg et al., Elec-
Petrii et al., Electrochim. Acta 2000, 45, 3521. trochim. Acta 2000, 46, 373.
65. D. J. Gavaghan, S. W. Feldberg, J. Elec- 88. A. A. Kornyshev in The Chemical Physics
troanal. Chem. 2000, 491, 103. of Solvation (Eds.: R. R. Dogonadze,
66. L. I. Krishtalik, Charge Transfer Reactions E. Kalman, A. A. Kornyshev, J. Ulstrup),
in Electrochemical and Chemical Processes, Elsevier, Amsterdam, 1988, p. 355, Part C
Plenum Press, New York, London, 1986. Solvation Phenomena in Physical, Chemical
67. R. J. Hunter, Zeta Potential in Colloid Sci- and Biological Systems.
ence. Principles and Applications, Academic 89. M. A. Vorotyntsev, A. A. Kornyshev, Electro-
Press, London, New York, 1981. statics of Media with the Spatial Dispersion,
68. J. Lyklema, Fundamentals of Interface and Nauka, Moscow, 1993.
Colloid Science, Academic Press, London, 90. M. A. Vorotyntsev, V. Yu. Izotov, A. A.
Fundamentals, Vol. 1, 1993; Solid-liquid Kornyshev, Sov. Electrochem. 1983, 19, 364.
Interfaces, Vol. 2, 1995. 91. A. A. Kornyshev in The Chemical Physics
69. G. Valette, A. Hamelin, J. Electroanal. of Solvation (Eds.: R. R. Dogonadze,
Chem. 1973, 45, 301. E. Kalman, A. A. Kornyshev, J. Ulstrup),
70. D. Leikis, K. Rybalka, E. Sevastyanov et al., Elsevier, Amsterdam, 1985, p. 77, Part A
J. Electroanal. Chem. 1973, 46, 161. Theory of Solvation.
71. A. N. Frumkin, J. Res. Inst. Catalysis 92. M. A. Vorotyntsev, A. A. Kornyshev, Sov.
Hokkaido Univ. 1967, 15, 61. Electrochem. 1979, 15, 560.
72. M. A. Vorotyntsev, J. Electroanal. Chem. 93. A. A. Kornyshev, M. A. Vorotyntsev, J. Uls-
1981, 123, 379. trup, Thin Solid Films 1981, 75, 105.
73. N. B. Grigoryev, Dokl. Akad. Nauk SSSR 94. A. A. Kornyshev, J. Ulstrup, Chem. Scr.
1976, 229, 647. 1985, 25, 58.
74. I. A. Bagotskaya, B. B. Damaskin, M. D. 95. A. A. Kornyshev, J. Electroanal. Chem. 1986,
Levi, J. Electroanal. Chem. 1980, 115, 189. 204, 79.
75. V. E. Kazarinov, M. A. Vorotyntsev, M. Ya. 96. A. A. Kornyshev, J. Ulstrup, J. Electroanal.
Kats in Double Layer and Adsorption at Solid Chem. 1985, 183, 387.
128 2 Electrochemical Double Layers
97. J. D. Cole, Perturbation Methods in Applied I. Prigogine, S. A. Rice), Wiley, New York,
Mathematics, Plaisdell, Waltham, 1968. 2001, p. 337.
98. R. G. Watts-Tobin, Philos. Mag. 1961, 6, 133. 125. M. A. Vorotyntsev, A. A. Kornyshev, Sov.
99. W. R. Fawcett, Isr. J. Chem. 1979, 18, 3. Electrochem. 1984, 20, 1.
100. N. F. Mott, R. G. Watts-Tobin, Electrochim. 126. A. A. Kornyshev, M. A. Vorotyntsev, J. Elec-
Acta 1961, 4, 79. troanal. Chem. 1984, 167, 1.
101. S. Levine, G. M. Bell, A. L. Smith, J. Phys. 127. J. W. Halley, B. Johnson, D. Price et al.,
Chem. 1969, 73, 3534. Phys. Rev. B 1985, 31, 7695.
102. I. L. Cooper, J. A. Harrison, J. Electroanal. 128. V. I. Feldman, A. A. Kornyshev, M. B.
Chem. 1975, 66, 85. Partenskii, Solid State Commun. 1985, 53,
103. W. R. Fawcett, J. Phys. Chem. 1978, 82, 1385. 157.
104. B. B. Damaskin, A. N. Frumkin, Electrochim. 129. A. V. Schlepakov, I. A. Bagotskaya, Elek-
Acta 1974, 19, 173. tokhimija 1982, 18, 26.
105. R. Parsons, J. Electroanal. Chem. 1975, 59, 130. L. M. Doubova, I. A. Bagotskaya, Elek-
229. tokhimija 1977, 13, 64.
106. B. B. Damaskin, J. Electroanal. Chem. 1977, 131. E. K. Patjarv, U. V. Palm, Elektokhimija
75, 359. 1976, 12, 806.
107. A. A. Kornyshev, Electrochim. Acta 1989, 34, 132. R. Payne, J. Am. Chem. Soc. 1967, 89, 459.
1829. 133. G. Vallette, A. Hamelin, J. Electroanal.
108. Ya. I. Frenkel, Z. Phys. 1928, 51, 232. Chem. 1973, 45, 301.
109. M. D. Lang, Solid State Phys. 1973, 28, 225. 134. V. A. Panin, K. V. Rybalka, D. I. Leikis,
110. N. D. Lang, W. Kohn, Phys. Rev. B 1971, 3, Elekrokhimija 1972, 8, 390.
1215. 135. D. Price, J. W. Halley, J. Electroanal. Chem.
1983, 150, 347.
111. I. E. Inglesfield in The Chemical Physics of
136. M. B. Partenskii, V. Dorman, P. C. Jordan,
Solid Surfaces and Heterogeneous Catalysis,
Int. Rev. Phys. Chem. 1996, 15, 153.
Elsevier, Amsterdam, 1981, p. 183, Vol. 1.
137. A. M. Brodsky, M. I. Urbakh, Usp. Fiz.
112. J. Harris, A. Liebsch, Vacuum 1983, 33, 655.
Nauk 1982, 138, 413.
113. A. K. Theophilou, A. Modinos, Phys. Rev. B
138. A. M. Brodsky, L. I. Daikhin, M. I. Urbakh,
1972, 6, 801.
Usp. Khim. 1985, 54, 3.
114. M. B. Partenskii, Ya. G. Smorodinskii, Fiz.
139. A. M. Brodsky, M. I. Urbakh, Prog. Surf. Sci.
Tverd. Tela 1974 16, 644. 1984, 15, 121.
115. P. G. Dzhavakhidze, A. A. Kornyshev, G. I. 140. A. M. Brodsky, M. I. Urbakh, Electrodynam-
Tsitsuashvili, Solid State Commun. 1984, 52, ics of Metal Surfaces, Nauka, Moscow, 1989.
401. 141. A. A. Kornyshev, M. I. Urbakh, J. Elec-
116. J. R. Smith, Phys. Rev. 1969, 181, 522. troanal. Chem. 1987, 235, 11.
117. P. Gies, R. Gerhardts, Phys. Rev. B 1985, 31, 142. J. D.-E. McIntyre in Advances in Electrochem-
6843. istry and Electrochemical Engineering (Eds.:
118. P. Gies, R. Gerhardts, Phys. Rev. B 1986, 33, P. Delahay, C. N. Tobias), Wiley, New York,
982. 1973, p. 61, Vol. 3.
119. W. Kohn, L. J. Sham, Phys. Rev. 1965, 140, 143. A. B. Ershler, A. M. Foontikov, V. N. Alex-
1133. eev et al., Elektrochimiya 1980, 16, 1171.
120. W. Schmickler, D. Henderson, Prog. Surf. 144. A. G. Foontikov, Dissertation, A. N.
Sci. 1986, 22, 323. Frumkin Institute of Electrochemistry of
121. V. I. Feldman, M. B. Partenskii, M. M. the Academy of Sciences of the USSR,
Vorobjev, Prog. Surf. Sci. 1986, 23, 3. Moscow, 1981.
122. J.-P. Badiali, Ber. Bunsen-Ges. Phys. Chem. 145. A. A. Kornyshev, M. I. Urbakh, J. Elec-
1987, 91, 270. troanal. Chem. 1988, 257, 305, 1987, 235,
123. S. Amokrane, J.-P. Badiali in Modern As- 11.
pects of Electrochemistry (Eds.: B. E. Conway, 146. P. G. Dzhavakhidze, A. A. Kornyshev,
J.O’M. Bockris, R. E. White), Plenum Press, A. Liebsch et al., Phys. Rev. B 1992, 45, 9339.
New York, 1992, p. 22, Vol. 1. 147. P. G. Dzhavakhidze, A. A. Kornyshev,
124. J. W. Halley, S. Walbran, D. Price A. Liebsch et al., Electrochim. Acta 1991, 36,
in Advances in Chemical Physics (Eds.: 1835.
2.1 Electrochemical Interfaces: At the Border Line 129
summarize the main points. It considers of the ions in the bulk of the solution, kB
a planar metal electrode in contact with is the Boltzmann constant, and T is the
an ionic solution, and starts from the temperature. The Debye length LD , which
following assumptions: is familiar from the Debye–Hückel theory
for electrolyte solutions, is given by
• the metal is a perfect conductor, and its
excess charge is distributed evenly on 1/2
εε0 kT
the surface; LD = (2)
2(ze0 )2 n0
• the solvent is a dielectric continuum,
characterized by a dielectric constant ε; The GC capacity has a pronounced mini-
• the ions are point particles, whose mum at the pzc, and rises rapidly on both
distribution is determined by the Pois- sides (see Fig. 1). It compares well with
son–Boltzmann equation. experimental data for low electrolyte con-
For the case of a z–z electrolyte, explicit centrations and for low to medium charge
expressions for the differential capacity densities, where the structure of the sys-
and for the profile of the electrostatic tem, the size and chemical nature of the
potential can be derived. We give the solution, and the electronic properties of
expression for differential capacity C per the metal, do not matter.
unit area: Considering the simplicity of the GC
theory, it works surprisingly well–so well
dσM εε0 ze0 (φ − φpzc ) that it still serves as the reference point
CGC = = cosh for all further advances in double layer
dφ LD 2kT
(1) theory, and new results are often stated as
where φ is the electrode potential and corrections to the GC theory.
φpzc is the potential of zero charge (pzc), A cursory review of the experimental
at which the charge density σM on the data shows that the GC theory overes-
electrode vanishes. n0 is the concentration timates the capacity for high electrolyte
3.0
2.0
[F m−2]
10−1 M
C
1.0
10−2 M
10−3 M
0.0
−0.10 −0.06 −0.02 0.02 0.06 0.10
(f − f pzc)
[V]
Fig. 1 The Gouy–Chapman capacity for several ionic
concentrations of a 1 : 1 electrolyte.
2.2 Electrical Double Layers: Theory and Simulations 135
concentrations and for high charge densi- can be obtained by measuring the capacity,
ties. In a real solution, the ions and the at a given charge density, for several
solvent molecules have a finite size, which electrolyte concentrations, and plotting
prevents their centers from coming arbi- the inverse of the experimental capacity
trarily close to the metal surface, and which versus the calculated GC capacity; this is
also limits the amount of charge that can known as a Parsons and Zobel plot [4].
be packed into the first few layers. To ac- This should result in a straight line, and
count for these effects, Stern [3] suggested the Helmholtz capacity can be obtained
that the first layer of the solution is free of from the intercept. For some metals, in
ions, and that this inner or compact layer particular for mercury, this procedure
contributes a term to the capacity, which works well, while for others, such as
is in series with the GC term: gold [5], the plot is curved, which may
1 1 1 indicate that there is always at least weak
= + (3) specific adsorption on these metals; of
C CH CGC
course, it may also indicate that Eq. (3)
where the Helmholtz capacity CH is does not hold for these systems.
attributed to the inner layer; CH is The Helmholtz capacity always depends
sometimes called the inner layer or compact strongly on the charge density σM ; dif-
layer capacity, but since Eq. (3) also ferent metals may have quite different
holds in models that do not assume capacities, but typically there is a maxi-
the existence of an inner layer, we shall mum, also known as the hump, near the
not use that terminology. In the absence pzc (see Fig. 2). Since the strength of
of specific adsorption, the Helmholtz the electric field increases on both sides
capacity is expected to be independent of of the pzc, the occurrence of the hump
the electrolyte concentration, and also of can be explained by dielectric saturation.
the nature of the ions. Experimentally it Watts-Tobin [6] elaborated this idea into
80
70
Ag(111)
60
50
[µF cm−2]
CH
40
30
20 Hg
10
0
−12.0 −8.0 −4.0 0.0 4.0 8.0 12.0
s
[µC cm−2]
Fig. 2 The Helmholtz capacity for Ag(111) and Hg in contact with
an aqueous solution; after Ref. [77] with permission from the author.
136 2 Electrochemical Double Layers
a quantitative model: he treated the in- in which each point is occupied by one
ner layer as a two-dimensional lattice of water molecule, which can form hydrogen
dipoles, which could be oriented either bonds with its nearest neighbors and
parallel or antiparallel to the surface nor- take up a fairly large number (25) of
mal. In effect, this is equivalent to the Ising orientations. The statistical mechanics of
model, which was evaluated in the mean this ensemble was treated in the quasi-
field approximation (MFA). As expected, chemical approximation, which is much
this model predicts a capacity hump at better than the mean-field approximation
the pzc, but otherwise it does not agree that was employed in the Ising-like models
well with the Helmholtz capacity of mer- discussed above.
cury, on which the early theories were However, all these early monolayer
focused, or of any other metal. In mer- models suffer from two defects: the
cury, the Helmholtz capacity rises again at connection of the inner layer to the rest of
charge densities above the hump. This the solution is not clear and the electronic
increase was explained [7] by the pres- structure of the metal is neglected – it just
ence of two water species in the inner acts as a perfect conductor. Nowadays, it
layer: single water molecules and wa- is generally recognized that there are three
ter clusters composed of two molecules, contributions to the double-layer capacity
which have a smaller dipole moment. on metals:
A high electric field can break up the
1. the capacity of the diffuse layer, which
clusters, and the capacity increases again
is described by the GC theory;
since the average dipole moment becomes
2. a contribution from the boundary layer
larger. Parsons [8] worked out the statisti-
of the solution, which extends over a
cal mechanics of this four-state model (two
distance of several solvent diameters
orientations for each water species), and by
toward the bulk;
an appropriate choice of the system param-
3. the electronic response of the metal
eters he obtained reasonable agreement
surface to the high field at the interface.
with the data for mercury. Later this model
was elaborated by other researchers, who A treatment of these effects requires
added further details and increased the more advanced methods than the GC
number of adjustable parameters, so that theory does; therefore, we shall briefly
the mercury data could be fitted perfectly; review the methods that have been used in
such work has been critically reviewed by the double-layer theory before discussing
Guidelli [9]. However, if this kind of model the latter two contributions in greater
were to represent the data for other met- detail.
als, all parameters would have to be fitted
again, so the physical significance of these 2.2.3
models was dubious from the beginning. Theoretical Methods and Principles of
Modern theories have abandoned the Computer Simulations
idea that the water dipoles can take up only
two orientations, though it occasionally Double-layer theory is complicated by the
still crops up in the literature. A significant fact that it involves the theory of liquids
conceptual progress was achieved in the and of solids, and of the interactions be-
work of Guidelli [10, 11]. He modeled the tween them. While the theory of the bulk
inner layer as a two-dimensional lattice, of these phases is quite well developed,
2.2 Electrical Double Layers: Theory and Simulations 137
together with Eq. (4) it forms a closed several simple models of water [12, 13];
system of equations for the unknown func- the results are all numerical.
tions cij (1, 2) and hij (1, 2), which can be The MSA, HNC, RHNC, and similar
solved at least in principle. In the spe- schemes were first developed to treat bulk
cial case in which the interaction is small, fluids, and then applied to interfaces by
βvij (1, 2) < 1, the total correlation func- using the following trick: one spherical
tions hij are also small, and the HNCs particle from the ensemble is singled out
simplify to the mean spherical approxima- and its radius is taken to be infinitely large
tion (MSA): or, if that limit is not tractable numerically,
much larger than the radii of all other
cij (1, 2) = −βvij (1, 2) (6) molecules. This large particle represents
the surface.
which results in a linear theory. This
is a useful approximation for weakly
2.2.3.2 Computer Simulations
interacting particles in the vicinity of
Even comparatively simple models turn
the pzc, where the Coulomb forces are
out to be difficult to treat by the methods
weak.
of statistical mechanics. In such cases,
A better approximation, which has be-
computer simulations may be used to
come quite popular recently, is the reference
explore the consequences and predictions
hypernetted chain (RHNC) approximation.
of the model, even though they do not
It makes use of a reference system consisting
give the same detailed understanding as
of a fluid of neutral hard spheres, which
analytical methods do. This is not the place
is the only potential for which the bridge
to describe these methods in any detail; the
function d(1, 2) in Eq. (2) is known with
book by Allen and Tildesley [14] provides
sufficient accuracy. If we denote by the sub-
an excellent introduction to the field. There
script 0 a property of the reference system
are three different methods.
and set h = h0 + h, c = c0 + c, v =
v0 + v, and d = d0 + d ≈ d0 , Eq. (5)
can be written as Monte Carlo simulations The basic me-
thod works in the following way: starting
c(1, 2) = h(1, 2) from a given configuration, a new con-
figuration is tentatively chosen at random
− log[g0 + h(1, 2)]
from a certain region of phase space, for
+ log g0 − βv(1, 2) (7) example, by moving one particle a cer-
tain distance. If the energy of the new
where g0 ≡ h0 + 1 is only a function configuration is lower than that of the
of the separation r ≡ |r1 − r2 | between previous one, it is accepted, and a new
the two particles. The pair-interaction Monte Carlo (MC) step is initiated. Other-
potential v0 of the reference system is wise a random number r is chosen from
just the spherically symmetrical part of the interval (0, 1). If r < exp −E/kT ,
v(1, 2). Equation (7) is exact except for the where E is the change in energy, the
replacement of the true bridge function new configuration is accepted, otherwise
d(1, 2) of the system under study by the it is discarded. It can be shown that
bridge function d0 (1, 2) of the reference averaging over a large number of config-
system. The RHNC has been applied to urations that have been produced in this
2.2 Electrical Double Layers: Theory and Simulations 139
way leads to correct thermodynamic quan- bring them up again as we discuss specific
tities [15]. There are various variants of this models.
method; they mainly provide information Real systems contain a practically infi-
about the structure and thermodynam- nite number of particles, while the ensem-
ics of an ensemble, but a few have bles that are used in simulations contain
been designed to investigate the dynamics typically a few hundred, at best a few
as well. thousand, particles. In order to mimic an
infinite system, cyclic boundary conditions
Molecular dynamics The orbits of the can be imposed. In electrochemical investi-
particles in an ensemble are obtained gations they are usually imposed in the two
from Newton’s second law: at each step directions parallel to the electrode surface.
the forces on all particles, and hence In the perpendicular direction the dimen-
their accelerations, are calculated, and sions should be so large that a region
from these their movements during a with bulklike properties is obtained, so
small time step, in which the forces that bulk and surface properties can be
can be taken as constant, are obtained. contrasted.
Then the forces are calculated again,
and the system moves another step 2.2.4
forward. During an initialization period, Electronic Structure Calculations
the system must be equilibrized to the
desired temperature, and then it can be In spite of the explosion of computing
sampled to obtain statistical properties. power during the last decade, it is still
Obviously, this method gives both the impossible to calculate the electronic
dynamics and the thermodynamics of the structure of semi-infinite solids. Three
ensemble. different approaches have been taken to
deal with this problem.
Car–Parinello method In ordinary simu-
lations the electronic structure of the par- Simple model systems A good method in
ticles is taken as fixed, and the molecules such situations is a simplification of the
interact through given potentials. In the model so as to make it mathematically
Car–Parinello method [16], at each step of tractable. The art of the theorist consists
the simulation the electronic structure of in keeping the most relevant features of
the system is recalculated, usually through the system and discarding the others.
density-functional methods [17, 18], and Since in the electric double layer the
from these the forces on the particles are electronic properties of the metal surface
obtained. In principle, this is the most play the major role, the jellium model [18],
exact but also the most time-consuming which disregards the details of the atomic
method, so that only small ensembles can structure and focuses on the distribution
be considered. Sometimes this method of the electrons, has been widely applied.
is simplified by restricting the electronic While this model does not give quantitative
structure calculations to a part of the en- results for particular systems, it has
semble. greatly improved our understanding of
All of these methods have been applied the electronic effects; we will discuss
to electrochemical systems, and we shall it later.
140 2 Electrochemical Double Layers
the charge distributions e0 q(r) in the dou- help in understanding the incorporated
ble layer and e0 q0 (z) in the metal, and can effects.
be written in the form
βe02 [q0 (z) + q(r)] 2.2.5.2 Ensembles of Hard Spheres
[q0 (z ) + q(r )] In the GC theory the size of the molecules
βH [q(r)] = dr dr was disregarded, so it is a natural im-
8πε|r − r |
∞ provement to treat them as hard spheres.
an
+ [q(r)]n dr (8) Actually, there are two different models
ρb n! that use hard spheres: in the primi-
n=2
tive model the ions are considered as
where β = 1/kB T , and ρb is the bulk charged hard spheres, while the solvent
concentration of the ions. The first term is a dielectric continuum; in the civilized
is the electrostatic energy in the systems, or nonprimitive model both the ions and
and all other terms, in particular structural the solvent molecules are considered to
and entropic terms, have been written be spheres, but the latter have a dipole
as an expansion in powers of q(r). moment at their center. The primitive
The coefficients an are dimensionless model is mainly of academic interest:
quantities. ions and solvent molecules are roughly
The simplest way to treat this Hamilto- of the same size and should be treated on
nian is the MFA. The statistical average the same footing. However, one predic-
q(z) is then the value of q(r) for which
tion from the primitive model deserves
the functional derivative δH [q(r)]/δq(r)
attention: it suggests that at low tem-
vanishes. For the Hamiltonian of Eq. (8)
peratures the interfacial capacity should
this gives
increase with temperature, while the GC
∞
theory and most model calculations predict
an
βV (z) + q(z)(n−1) = 0 (9) a decrease. In this context, it is useful to
ρb
n=1 define an effective or reduced temperature
where V (z) is the average electrostatic through T ∗ = kT σi (4πεε0 )/e02 , where σi
potential. If one assumes that the average is the diameter of the ions, which is as-
charge distribution is related to the average sumed to be equal for the two species.
potential via a Boltzmann distribution and Low temperature means T ∗ < 1, so that
that the metal is a classical conductor (i.e. aqueous solutions are high-temperature
q0 (z) ∝ δ(z)), then all the coefficients an systems because of their high dielectric
can be determined, and one obtains the constant. Somewhat surprisingly, molten
nonlinear GC theory. salts are low-temperature systems by
So far, this is nothing but an ingenious this definition. This result is not only
method for obtaining a well-known result. based on simulations within the primi-
However, this approach can be extended tive model [21] but also on more general
in various ways by including other effects. arguments based upon pressure balance
Stafiej and coworkers discuss explicitly at the interface [22]; it may explain why
ion–ion interactions, chemical interac- the capacity of gold electrodes in contact
tions with the wall, and nonlocality. The with a frozen, highly concentrated aque-
main value of this method lies in the fact ous solution has been found to increase
that the calculations are transparent and with temperature [23].
142 2 Electrochemical Double Layers
We now focus our attention on the developed into a power series in κ; the first
civilized model. In the electrochemical two terms are
context the basic system consists of hard
1 1 1 1 ε−1
spheres in contact with a charged hard = + σi + σs
C εε0 κ 2 λ
wall; there are three types of hard spheres: (10)
solvent molecules of diameter σs , which The next term is of the order of κ. Within
have a dipole moment µ at their center, the MSA, the dielectric constant ε is related
and ions with a diameter σi and ions to the dipole moment µ and the density ρs
with a charge number ±1 – note that both of the solvent through
kinds of ions have the same diameter.
The integral equations for this model ε−1 4πµ2 9
= ρs (11)
have been solved analytically, but in the ε 3kT λ (λ + 1)2
2
MSA [24, 25] only, so that the results are
valid for small excess-charge densities on and the parameter λ is obtained from the
the electrode, that is, in the vicinity of dielectric constant via
the pzc.
λ2 (1 + λ)4 = ε (12)
The solution is implicit and consists of
a pair of nonlinear algebraic equations, The first term in Eq. (10) is just the GC
which can be solved numerically. In the capacity at the pzc; the second term is
limit of low concentrations, where the independent of the ionic concentration,
Debye inverse length κ = 1/LD tends to and can be identified with the Helmholtz
zero, explicit results can be derived. The capacity. However, in this model the
distribution of the ions gi (x) and of the Helmholtz capacity is not caused by a
dipoles gs (x) as a function of the separa- single monolayer of solvent with special
tion x from the electrode show oscillations, properties, like in the Stern model, but
which are caused by the packing of the par- results from an extended boundary layer.
ticles (see Fig. 3); they die out far from the It depends on the dielectric properties of
surface. the solvent and on the diameters of the
The interfacial capacity C is of particular particles. Since λ ε, the influence of
interest since it is measurable. For low the ions on the capacity is predicted to be
ionic concentrations, its inverse can be small. This is in line with the experimental
2
[kT]
e0f
1.25
0.75
n n −1b
Metal
0.50
Effective
0.25
image plane
0.00
−3.0 −1.5 0.0 1.5 3.0
x
[Å]
Fig. 4 Normalized electronic density profile at the surface of
jellium with a bulk density of 15.3 atomic units. The upper dashed
curve is for a surface charge density σM = −0.1 C m−2 , the lower
dashed curve for σM = 0.1 C m−2 . The full curve is for an
uncharged surface. The arrow gives the position of the effective
image plane for an uncharged surface.
Trasatti [28] has noticed that in aqueous A major breakthrough in our under-
solutions the Helmholtz capacity CH of standing of the electric double layer
the simple sp metals, taken at the pzc, came with the introduction of the jel-
correlates with their electronic densities lium model, which had already been used
(see Fig. 5). This indicates that the surface extensively in the theory of metal sur-
polarizability of simple metals increases faces [31, 32], into electrochemistry [34,
with their electronic densities, a trend that 36]. In this model the lattice of posi-
can be explained by the jellium model tively charged metal ions is represented
presented below. by a constant positive background charge,
which drops abruptly to zero at the sur-
2.2.6.2 The Jellium Model face. The electrons are modeled as an
As early as 1928, Rice [29] attempted to inhomogeneous electron gas, which inter-
estimate the contribution of the metal to acts with the positive background. Explicit
the interfacial capacity within the frame- calculations for this model are usually
work of the Thomas–Fermi theory. How- based on density-functional theory, which
ever, he obtained a positive contribution of has already been mentioned above. The
the metal to the capacity, which was prac- quantum-mechanical self-interactions of
tically the same for all metals. Only some the electron gas – the exchange and the
50 years later, Kornyshev and cowork- correlation energies – are treated in the lo-
ers [30] showed that Thomas–Fermi-like cal density approximation, which is based
models are unsuitable for double-layer the- on the following idea: these interactions
ory because they do not give a realistic are known for an electron gas of con-
description of the work functions and the stant electronic density. As a first ap-
surface potentials of metals. proximation, one can assume that in an
2.2 Electrical Double Layers: Theory and Simulations 145
4
Hg
3 Tl
[m2 F −1]
C H−1
Cd
2
In
Zn
1
Ga
0
8 16 24 32
n × 1000
[a.u.]
Fig. 5 Inverse Helmholtz capacity of aqueous solution in contact
with sp metals; experimental values from Trasatti [28]. (The dashed
line is based on a model calculation by Schmickler and
Henderson [26].)
inhomogeneous gas the exchange and cor- in front of the geometrical surface. Thus
relation energies at each point take on a positive value of xim indicates that the
the same values that they would have effective plate separation is smaller than
in a homogeneous gas with the same the physical separation between the two
density. While this approximation is sim- surfaces. Typical values for xim are of the
ple, it has the advantage that it fulfills a order of 0.5 Å; for a normal plate capacitor
number of sum rules that the exact expres- such a small shift of the apparent plate
sion must also obey. Most importantly, position is quite negligible. However, for
it gives good results for the electronic high ionic concentrations the double layer
structures of surfaces, and only for com- can be considered as a capacitor with a
plicated interactions must one resort to plate separation of a few Ångstrøms, so
better mathematical treatments such as that a shift of this order of magnitude has
the generalized gradient approximation. a noticeable effect.
The electronic polarizability of the jel- The relation between the position of
lium surface, which is the relevant quantity the image plane and the metal capacity
for the interfacial capacity, can be ex- is given by
pressed in terms of the effective position ∂χ 1
of the image plane, which has the meaning xim = −ε0 = −ε0 (15)
∂σM Cm
that its name suggests: a test charge placed
in front of the jellium surface experiences As one might expect, the polarizability
an image force as if the image plane were increases with the electronic density (see
a distance xim in front of the surface (see Fig. 6), which qualitatively explains the
Fig. 4). In a metal capacitor the effective corresponding increase of the Helmholtz
positions of the plates lie at a distance xim capacity, though for jellium in vacuum
146 2 Electrochemical Double Layers
2.0
1.5
x im
1.0
−10 [Å]
0.5
−5 25
0 20
sM
[µC cm−2] 5 15 n × 103
10 [au]
10
Fig. 6 Variation of the effective position of the image plane with the
electronic density and with the surface charge density, calculated
within the jellium model.
this dependence is relatively weak. It also reproduce trends and orders of magni-
depends on the surface charge density, tudes for polycrystalline materials. In an
and is much higher when there is a improved version the positive background
considerable excess of electrons on the charge is replaced by a lattice of pseu-
surface. dopotentials so that single crystals can be
As already mentioned, the distribu- represented. Usually, the pseudopotentials
tion of electrons at the jellium surface are averaged parallel to the surface in order
entails a surface dipole moment. There- to keep the calculations one-dimensional.
fore, at the pzc a water molecule situated Such models predict sizable differences
within the electronic tail experiences a for the surface polarizabilities, measured
positive field that tends to orient the in terms of xim , of different single-crystal
molecule with its oxygen end toward surfaces, and may thereby explain why the
the metal surface. Conversely, the water interfacial capacity varies with the crystal
molecules tend to enhance the electronic orientation [40, 42]. However, the calcu-
spillover. As a result, the work func-
lated values depend more strongly on the
tion for jellium in contact with water
orientation than do the experimental val-
is lower than in the vacuum. Model
ues; this indicates that the contributions
calculations for this effect have been per-
of the metal and of the solvent are not
formed by several authors [34, 37, 39–41].
independent.
While the results depend on the details
of the model, it is generally agreed that
the larger the overall change in the 2.2.6.3 Calculations for Metal Clusters
dipole potential, the higher the electronic Modern quantum-chemical methods ac-
density. count for the chemical properties of
Jellium is a simple, structureless model the atoms much better than jellium-type
for a metal, and can therefore only models can, but they are usually limited
2.2 Electrical Double Layers: Theory and Simulations 147
1.26
1.25
1.25
1.25
[Å]
x im
1.24
1.24
1.23
−20 −10 0 10 20
sM
[µC cm−2]
Fig. 7 Effective position of the image charge for a mercury cluster as
a function of the surface charge density. (Data taken from Ref. [44].)
Fig. 8 Metal–solvent
separation ‘‘s’’ as a function of
the surface charge density for
two different choices of the
pseudopotentials [52, 53].
2.0
(a)
(b)
[A°]
s
1.5
−10 0 10
sM
[µC cm−2]
[µF cm −2]
C
50
Solvent
Solvent
0
−10 0 10
sM
[µC cm−2]
curves is the same for reasonable choices in the sense that different experimental
of these pseudopotentials. capacity data give almost the same solvent
To arrive at a complete double-layer capacities.
model, Amokrane and Badiali proceeded Bérard and coworkers [54] have applied
in a semiempirical manner. Assuming the RHNC to a simple water model and
that the contributions of the metal and combined this with the jellium model. The
the solvent to the inverse capacity are solution and the jellium were assumed to
additive (see Eq. 14), they obtained the interact electrostatically. This leads to a
solvent capacity from experimental data coupled set of equations for the density
for the Helmholtz capacity of Ag(111). In profiles of the electrons and the parti-
this way they obtained a solvent capacity cles in the solution, which was solved
for water, which shows a pronounced self-consistently through an iterative pro-
maximum near the pzc (see Fig. 9); this cedure. A critical issue in this coupling is
curve can be fitted to a model in which the the distance of closest approach of the
solvent is represented by a layer of dipoles. solvent molecules to the jellium edge,
This procedure can be carried fur- modulated by the repulsion of the inhomo-
ther by taking experimental data for the geneous electron gas from the orbitals of
Helmholtz capacities of other metals and the solvent molecules. Bérard and cowork-
extracting the metal contribution by as- ers [54] allow the solvent spheres to come
suming that the solvent part is not affected into contact with the jellium edge. More-
by the metal. In this way, Amokrane and over, owing to the presence of the solvent,
Badiali arrive at a consistent interpreta- the electron density spills further out of the
tion of all experimental data. However, metal, dropping to 0.1% of the bulk value at
their theory hinges on the calculations for 3 Å. At an uncharged metal, the resulting
Ag(111). In fact, the variation of the cal- strong coupling causes the solvent dipoles
culated metal capacity is so strong that it in the layer contacting the metal to have a
practically determines the solvent capacity small net orientation normal to the metal
150 2 Electrochemical Double Layers
−10 −5 0 5 10
z
[Å]
surface, with the positive ends out. This such complications increase the computer
is in agreement with the results of several time.
computer simulations that we will discuss The interaction of water with the metal
below. has been modeled in various ways: hard-
wall, Lennard–Jones, and Morse poten-
2.2.7.2 Molecular Dynamics Simulations tials have been used, as well as potentials
Realistic models for solvents such as derived from ab initio cluster calculations.
water are too complicated to be treated Whatever interaction is chosen, the pres-
by anything but computer simulations, ence of the metal always induces a layering
and molecular dynamics has usually been of the water molecules, similar to that seen
the method of choice. Most studies have in the hard-sphere ensemble. As an ex-
been limited to pure water in contact ample, we show the density profiles for
with a metal surface, but recently a the oxygen and hydrogen atoms obtained
few simulations have been performed for by Foster and coworkers [56] in Fig. 10.
concentrated aqueous solutions. We will The corresponding simulations were per-
not attempt to survey this very active field, formed in the absence of an external field,
but give a few representative results. that is, for the pzc, and in the z-direction the
There are a variety of semiempirical water was contained between two identical
models for water; most of them model walls, while periodic boundary conditions
the molecule by a few point charges were applied in the xy directions. As a
with a Lennard–Jones or similar potential result of the layering effect, the density pro-
superimposed. For example, the popular files exhibit a series of oscillations, whose
rigid SPC/E model [55] consists of three amplitude diminishes toward the bulk. In
point charges of −0.8476e0 on the oxygen the first two layers the profiles for oxygen
and 0.4238e0 on the hydrogen sites (e0 is and hydrogen differ somewhat: on an av-
the proton charge) and a Lennard–Jones erage, the oxygen atoms are a little closer
potential centered at the oxygen atom. to the electrode surface than the hydrogen
The resulting dipole moment is somewhat atoms. Therefore the dipole moment of
larger than that of the isolated water the water molecules has a small net ori-
molecule in the gas phase in order to entation along the direction of the surface
account for the polarizability. Other water normal, which produces a surface dipole
models have a built-in polarizability, but potential χ. The magnitude of this effect
2.2 Electrical Double Layers: Theory and Simulations 151
[V]
χ
0
−1
−2
−10 −5 0 5 10
z
[Å]
Fig. 11 Surface potential χ as a function of the distance from the center
of the simulation box. (Data taken from Ref. [44].)
depends on the interaction between the the water model employed and on the in-
metal and water; Fig. 11 shows the dipole teraction of the first layer with the metal
potential for a simulation of water in con- surface. Xia and Berkowitz [61] suggest
tact with a mercury electrode. The total that at very high fields water undergoes
potential drop between the electrode sur- a phase transition to an icelike ordered
face and the bulk is of the order of several structure.
tenths of an electron volt, which agrees The simulations reported above have
qualitatively with estimates of Trasatti [57] been performed for pure water. Modeling
based on experimental data. the double layer requires the introduction
The detailed structure of water at the of ions, which are usually represented as
surface is often discussed in terms of charged spheres with a Lennard–Jones
potential. Given the limited size of the
the bilayer model proposed by Doering
ensembles that can be handled, a statisti-
and Madey [58] to explain water adsorp-
cally meaningful number of ions results
tion on metal surfaces in the vacuum at
in a high concentration; typically of the
low temperatures. As the name suggests,
order of 2 M. At present, a few such sim-
this model proposes two layers of water
ulations have been reported [62, 63], and
that are hydrogen-bonded to each other. more can be expected in the future as
Obviously, in the liquid state and at ambi- computing power grows. The results ob-
ent temperatures such a bilayer must be tained so far are still tentative, but generally
strongly disturbed, though a vestige seems seem to be in line with physicochemical
to appear in some simulations [59]. Simi- intuition and the notions developed by
lar, but more icelike, structures have been Grahame [64, 65] and Parsons [66], though
suggested by Krämer and coworkers [60]. a few details are surprising. Thus, the sol-
When the electrode surface is charged vation of the ions plays a central role in
or, equivalently, an electric field is intro- the double-layer structure. In particular,
duced into the simulation box, the water the fact that the anions tend to be less
molecules tend to align with the field; the strongly solvated than the cations induces
extent to which this happens depends on an asymmetry with respect to the pzc. As
152
4
Oxygen density
Ionic densities
3 3
=0 =0
2 2
2 Electrochemical Double Layers
1 1
0 0
2.5
2.5
0.0 0.0
0 5 10 15 20 25 0 5 10 15 20 25
z z
[Å] [Å]
Fig. 12Distribution of the ions at an idealized metal surface with various surface charge densities. Dotted lines: cations (Na+ ), full lines: anions (Cl− ).
For comparison, the figure (a) shows the distribution of the oxygen atoms at the pzc.
2.2 Electrical Double Layers: Theory and Simulations 153
an example, we show in Fig. 12 the den- and interaction potentials derived from
sity profiles of Na+ and Cl− obtained by quantum-chemical calculations.
Spohr [63] at the pzc and at surface charge The system is investigated by molecular
densities of σ = ±9.9 µF cm−2 . At the pos- dynamics, performing a self-consistent
itively charged surface the first peak for calculation for the electronic subsystem
the weakly solvated anions is closer to the at each step. The presence of the solvent
metal (in this case mercury) surface than is found to have a pronounced effect on
the peak for the cation is at the negative the electronic distribution on the metal
surface. However, at the pzc the sodium surface. In particular, the simulations
ions are closer to the surface because their clearly show the formation of image
solvation shell can be accommodated be- charges on the metal surface.
tween the first and second water layers. This and similar studies from this group
The total potential drop at the interface is provide a very detailed description of
determined both by the distribution of the the interface, to which we cannot give
ions and by the orientation of the water; full justice in this limited space. Of
these two contributions tend to cancel each course, such comprehensive simulations
other, so there is a complicated interplay. are presently of a preliminary nature, but
While it is still too early to obtain the they do indicate the important physical
interfacial capacity from such simulations, processes that occur at the interface. We
first estimates give a reasonable order of may expect more work of this type to be
magnitude [63]. performed in the future.
the yardstick with which the surface area uneven distribution of the surface charge
is measured. Therefore, must also be considered: the charge
accumulates on the surface protrusions.
lim R̃(κ) = 1 and lim R̃(κ) = R Then the roughness function becomes
κ→0 κ→∞
(18) quite involved, and depends also on the
where R is the geometrical roughness electrode potential [70].
of the electrode, which is defined as the
ratio between the true and the apparent 2.2.9
surface areas. If the z-direction is taken Liquid–Liquid Interfaces
as perpendicular to the average surface,
then, for a moderately rough electrode, the In a certain sense, interfaces between two
true surface is given by a single-valued liquids are simpler than those between
function: a metal and a liquid, since they do not
z = ζ (x, y) (19) involve solid-state properties. However, for
a long time the structure of liquid–liquid
When the roughness is small it can interfaces has remained a matter of
be treated as a perturbation, and the controversy. Essentially, there are two
roughness function can be expressed different views: one holds that the interface
through the Fourier transform (FT) of is sharp and contains a compact layer
the surface. This theory is simple only of solvent molecules into which the ions
near the pzc; away from the pzc the cannot penetrate. The other view posits the
existence of an extended boundary layer in where φi∞ is the value of the poten-
which the two solutions mix. tial in the indicated phase far from
During the last few years, the latter view the interface. The capacity can be de-
has received more supporting evidence. composed into a series combination of
Already the early experimental work of two GC capacities Ci , one for each
Girault and Schiffrin [71], who determined interface:
the surface excess of water at the interface
εi ε0 zi e0 (φ s − φi∞ )
with 1,2-dichloroethane, had indicated the Ci = cosh (21)
existence of a mixed boundary layer. LiD 2kT
Recent X-ray scattering experiments [72]
where φ s is the potential at the idealized,
indicate an average interfacial width of the
flat interface (cf. Eq. 1). The latter must be
order of 3 to 6 Å. These experiments are
calculated from the Poisson–Boltzmann
in line both with model calculations based
equation [77].
on the density functional formalism [73]
The Verwey–Niessen model gives a rea-
and with computer simulations [74, 75].
sonable first approximation, but at a closer
Accordingly, the interface is best visualized
glance significant deviations are observed
as rough on a molecular scale as indicated
even at low ionic concentrations, where
in Fig. 13.
this theory is expected to hold. Surpris-
So far, there have been no molecular dy-
ingly, at low (≤10−2 M) concentrations
namics simulations with a larger number
the experimental capacity is often higher
of ions for these interfaces, and the double-
than that predicted by Verwey–Niessen
layer theory rests on analytical calculations
theory – this is just the opposite to the be-
and on simulations of the lattice gas model
havior of metal–solution interfaces, where
(LGM).
the capacity is always lower than the
GC value.
2.2.9.1 Verwey–Niessen Theory These deviations were first explained by
The basic system consists of two immisci- the presence of a compact, ion-free layer
ble solvents, each of which contains a salt at the interface; this is known as the mod-
that is badly solvable in the other solvent. ified Verwey–Niessen model. Obviously,
When a potential difference is applied be- the presence of an ion-free layer can only
tween the two solutions, opposing space reduce the capacity, so the theory had to
charge layers form on both sides of the be modified further. For a few systems
interface and give rise to a capacity. The a consistent interpretation of the experi-
simplest model consists of two diffuse dou- mental capacity was achieved [78–80] by
ble layers back to back, which are described combining this model with the so-called
by the GC theory – in the context of liq- modified Poisson–Boltzmann (MPB) the-
uid–liquid interfaces this model is known ory [81], which attempts to correct the GC
as the Verwey–Niessen theory [76]. theory by accounting for the finite size
The interfacial capacity per unit area at of the ions and for image effects, while
the interface between two solutions labeled the solvent is still treated as a dielectric
(1) and (2) is defined as continuum. The combined model has an
adjustable parameter, so it is difficult to
dσ judge whether the agreement with exper-
C= (20)
d(φ2∞ − φ1∞ ) imental data is significant. The existence
156 2 Electrochemical Double Layers
of a compact solvent layer is just a postu- which these authors handle by a pertur-
late, and is difficult to reconcile with the bation expansion treating the roughness
extended boundary layer discussed in the as a small parameter. Following Daikhin
previous section. and coworkers [69] (see previous section),
Pecina and Badiali express the deviation
2.2.9.2 Capillary Waves from the Verwey–Niessen theory through
Since liquid–liquid interfaces are rough, a roughness function R̃(LiD , εi ), which here
the ideas for rough electrodes presented depends on the dielectric constants of
the two solutions and on their Debye
above can be applied, though with some
lengths. Explicit expressions for the rough-
modifications. Again, if this roughness
ness function can be given for simple
is not too large, the interface can be
surface corrugations. While Pecina and
represented by a function ζ (x, y) at a
Badiali consider small deviations from the
particular moment of time. The two-
pzc, the nonlinear case has been treated by
dimensional Fourier components A(q) of
Urbakh and coworkers [83].
this surface function are known as capillary
The concept of capillary wave explains
waves; q is a two-dimensional wave vector.
the increase of the interfacial capacity
Thus, a rough surface can be considered as
beyond the Verwey–Niessen limit, and
a superposition of capillary waves, which
thus helps in understanding the structure
are present at all wavelengths. Of course,
of the interface. A quantitative interpre-
these cannot be observed as traveling
tation of experimental data within this
waves, but only as a fluctuating surface
model is, however, difficult since there are
roughness.
other causes for the deviations from Ver-
The energy stored in a capillary wave
wey–Niessen theory besides the surface
is proportional to the surface area it roughness.
creates. This entails a prediction for the
distribution of the amplitudes:
2.2.9.3 Lattice Gas Model
kT The Ising model is one of the most
A(q)A∗ (q) = (22)
γ |q|2 versatile models in physics and chemistry,
and in the equivalent form of the LGM
where the angular brackets denote ther- it has also been applied to liquid–liquid
mal expectation values. This relation has interfaces. In this case it is based upon a
indeed been observed in computer simu- three-dimensional, typically simple cubic
lations of liquid–liquid interfaces [74, 75]. lattice. Each lattice site is occupied by one
The concept of capillary waves can be of a variety of particles. In the simplest
used to explain how the surface roughness case the system contains two kinds of
increases the interfacial capacity beyond solvent molecules, and the interactions
the Verwey–Niessen value. For this pur- are restricted to nearest neighbors. If we
pose, Pecina and Badiali [82] have solved label the two types of solvent molecules
the linear Poisson–Boltzmann equation S1 and S2 , the interaction is specified by
across the interface between two solu- a symmetric 2 × 2 matrix wij , where each
tions with different dielectric constants element specifies the interaction between
and Debye lengths separated by a corru- two neighboring molecules of type Si
gated surface. A major difficulty is the and Sj . Whether the system separates
boundary condition at the rough interface, into two phases or forms a homogeneous
2.2 Electrical Double Layers: Theory and Simulations 157
2.0
1.5
Particle density
Anions Cations
1.0
Cations
Anions
0.5
0.0
−4 −2 0 2 4
z L−1D
Fig. 14 Normalized distribution c(x)/c0 of the ions for a potential
drop of e0 φ/kT = 1. The full lines give the distribution of the
majority ions; in the region x < 0 these are the anions, in x > 0
these are the cations. The dashed lines give the distributions of the
counterions. The calculations have been performed in the
quasi-chemical approximation for equal Debye lengths and
dielectric constants in both solutions [84].
158 2 Electrochemical Double Layers
2.0
1.5
(3)
[F m −2]
(2)
1.0
C
(1)
0.5
0.0
−0.25 −0.15 −0.05 0.05 0.15 0.25
∆f
[V]
Fig. 15 Interfacial capacity in the lattice gas model for various
ion–solvent interactions. The data were obtained from an MC
simulation [85]. The crosses denote the Gouy–Chapman capacity.
2.2 Electrical Double Layers: Theory and Simulations 159
22. D. Henderson, J. Chem. Phys. 2000, 112, 50. J. W. Halley, B. Johnson, D. Price et al., Phys.
6716. Rev. 1985, B31, 7695.
23. A. Hamelin, S. Röttgermann, W. Schmickler, 51. J. W. Halley, B. Johnson, D. Price et al., Phys.
J. Electroanal. Chem. 1987, 230, 281. Rev. 1988, 38, 9357.
24. S. L. Carnie, D. Y. C. Chan, J. Chem. Phys. 52. S. Amokrane, J. P. Badiali, Electrochim. Acta
1980, 73, 2949. 1989, 34, 39.
25. L. Blum, D. Henderson, J. Chem. Phys. 1981, 53. S. Amokrane, J. P. Badiali, J. Electroanal.
74, 1902. Chem. 1989, 266, 21.
26. W. Schmickler, D. J. Henderson, J. Chem. 54. D. R. Bérard, M. Kinoshita, X. Ye et al., J.
Phys. 1984, 80, 3381. Chem. Phys. 1994, 101, 6271.
27. D. Boda, D. Henderson, J. Chem. Phys. 1998, 55. H. J. C. Berendsen, J. R. Grigeria, T. P.
109, 7362. Straatsma, J. Phys. Chem. 1987, 91, 6269.
28. S. Trasatti, J. Electroanal. Chem. 1981, 123, 56. K. Foster, R. Raghavan, M. Berkowitz, Chem.
2031. Phys. Lett. 1989, 162, 32.
29. O. K. Rice, Phys. Rev. 1928, 31, 105. 57. S. Trasatti in Trends in Interfacial Electro-
30. A. A. Kornyshev, W. Schmickler, M. A. chemistry (Ed.: A. F. Silva), Reidel, Dordrecht,
Vorotyntsev, Phys. Rev. B 1982, 25, 5244. 1986.
31. J. R. Smith, Phys. Rev. 1969, 181, 522. 58. D. L. Doering, T. E. Madey, Surf. Sci. 1982,
32. N. D. Lang, W. Kohn, Phys. Rev. B 1970, 1, 123, 305.
4555. 59. J. P. Valleau, A. A. Gardner, J. Chem. Phys.
33. N. D. Lang, W. Kohn, Phys. Rev. B 1971, 3, 1987, 86, 4162.
1215. 60. A. Krämer, M. Vossen, F. Forstmann, J.
34. J. P. Badiali, M. L. Rosinberg, J. Goodisman, Chem. Phys. 1997, 106, 2792.
J. Electroanal. Chem. 1983, 143, 73.
61. X. Xia, M. L. Berkowitz, Phys. Rev. Lett. 1995,
35. J. P. Badiali, M. L. Rosinberg, J. Goodisman,
74, 3193.
J. Electroanal. Chem. 1983, 150, 25.
62. M. R. Philpott, J. N. Glosli, S. B. Zu, Surf.
36. W. Schmickler, J. Electroanal. Chem. 1983,
Sci. 1995, 335, 422.
150, 19.
63. E. Spohr, Electrochim. Acta 1999, 44, 1697.
37. W. Schmickler, D. J. Henderson, J. Chem.
64. D. C. Grahame, Chem. Rev. 1947, 41, 441.
Phys. 1984, 80, 3381.
65. D. C. Grahame, R. Parsons, J. Am. Chem.
38. W. Schmickler, D. J. Henderson, Prog. Surf.
Sci. 1986, 22, 323. Soc. 1961, 83, 1291.
39. W. Schmickler, Chem. Phys. Lett. 1983, 99, 66. R. Parsons in Comprehensive Treatise of Elec-
135. trochemistry (Eds.: J. O’M. Bockris, B. E. Con-
40. E. Leiva, W. Schmickler, Proc. Indian Acad. way, E. Yeager), Plenum Press, New York,
Sci. 1986, 97, 267. 1980, Vol. 1.
41. M. I. Rojas, E. P. M. Leiva, J. Electroanal. 67. D. L. Price, J. W. Halley, J. Chem. Phys. 1995,
Chem. 1991, 303, 55. 102, 6603.
42. E. Leiva, W. Schmickler, Surf. Sci. 1993, 291, 68. B. B. Mandelbrot, The Fractal Geometry of
226. Nature, Freeman, San Francisco, 1982.
43. W. H. Mulders, J. H. Sluyters, J. H. van 69. L. I. Daikhin, A. A. Kornyshev, M. Urbakh,
Lenthe, J. Electroanal. Chem. 1989, 261, 273. Phys. Rev. E 1996, 53, 6192.
44. R. R. Nazmutdinov, M. Probst, K. Heinz- 70. A. Kornyshev, private communication.
inger, Chem. Phys. Lett. 1994, 222, 101. 71. H. H. Girault, D. H. Schiffrin, J. Electroanal.
45. J. Böcker, R. E. Nazmutdinov, E. Spohr et al., Chem. 1983, 150, 43.
Surf. Sci. 1995, 333, 372. 72. A. M. Tikhonov, D. M. Mitrinovic, M. Li
46. J. P. Badiali, M. L. Rosinberg, F. Vericat et al., et al., J. Phys. Chem. B 2000, 200, 6336.
J. Electroanal. Chem. 1985, 158, 82. 73. D. Henderson, W. Schmickler, J. Chem. Soc.,
47. W. Schmickler, D. J. Henderson, J. Phys. Faraday Trans. 1996, 92, 3839.
Chem. 1984, 80, 3381. 74. I. Benjamin, Chem. Rev. 1996, 96, 1449.
48. W. Schmickler, D. J. Henderson, J. Phys. 75. I. Benjamin, Annu. Rev. Phys. Chem. 1997,
Chem. 1985, 82, 2925. 48, 407.
49. D. Price, J. W. Halley, J. Electroanal. Chem. 76. E. J. W. Verwey, K. F. Niessen, Philos. Mag.
1983, 150, 347. 1939, 28, 435–446.
2.2 Electrical Double Layers: Theory and Simulations 161
Phase α
+ − + −
−
− −
− + −
− − − − − − − −
− −
Interface Koenig surface
+ + + + + + + +
+
+
+
Phase β −
−
+ −
(a) (b)
ace
surf
nig
C αj C αj Koe
C kβ C kβ
C
(c) x (d) x
Fig. 1An arbitrary scheme of the structure of a perfectly polarizable interface between two
immiscible liquids (a, b, c, d) and an ideally polarizable interface (b, c).
If phases α and β contain at least one properties of some interfaces are very close
common ion that can freely pass across the to being ideal.
interface, the interface may be called re-
versible or nonpolarizable (Fig. 2). Although 2.3.2
the latter term is commonly accepted, it Verwey–Niessen Model
does not correctly describe the state of the
system. A reversible (nonpolarizable) in- Verwey and Niessen first described the
terface can be partially polarizable or com- EDL at ITIES as two noninteracting diffuse
pletely nonpolarizable. A completely non- layers, one at each side of the interface [9].
polarizable interface containing at least Both solvents were assumed to be struc-
one common ion can pass a high current tureless media with macroscopic dielectric
in either direction without causing a devi- permittivities, and potential distribution in
ation of the interfacial potential difference the EDL was defined by Gouy–Chapman
from the equilibrium value. Although, theory.
in practice we encounter neither ideally Gavach and coworkers [10] extended
polarizable nor completely nonpolarizable the Verwey–Niessen model (MVN) by
interfaces, under certain conditions the introducing an ion-free transition layer
164 2 Electrochemical Double Layers
α c kβ
c αj
c βj
C
c αk β
(b) X
εα ε αh ε β εβ
h
φ αd
∆βαφ h ∆βαφ
∆αβ g
φ dβ
α β
−δ α 0 δβ
Fig. 3 Distribution of the electric potential in the electric
double layer at the interface between two immiscible electrolyte
solutions.
φ φ
X X
(a) (b)
φ φ
We will denote surface charge density From the Gouy–Chapman theory for a
β
in the diffuse layers as qdα , qd and in the 1 : 1 electrolyte, it follows that:
compact layer as qh . From electroneutrality 2ε0 RT F φd
of the interphase, we have: qd = − εκ sinh (4)
F 2RT
β
qdα + qh + qd = 0 (2) where κ is the Debye length,
At the potential of the free zero charge 2F 2 c0
(fzcp), the compact layer has not adsorbed κ= (5)
εε0 RT
ions and the charges of the diffuse layers
are equal to zero: and c0 is the bulk electrolyte concentration
β
in the corresponding phase.
φdα = φd = 0 and αβ φh = αβ φ f zcp (3) If a Helmholtz layer does not have
charges, potential drops in the two diffuse
The dipolar potential αβ φh also needs
layers have a simple relation:
the index fzcp because of its dependence
on total interfacial potential. Usually, the β
F φdα F φd
dependence is weak and Eq. (3) is used to εα κ α sinh + εβ κ β sinh =0
2RT 2RT
determine the dipolar potential. (6)
166 2 Electrochemical Double Layers
1 2 3
RE1 H2O (α ) A+B− (β) A+C− (α ) RE2
M+Cl− oil H2O
a1 a2 a3
∆Agα φ RE1 ∆αβ φ ∆φD −∆Agα φ RE2
Scheme 1
2.3 Electrochemical Double Layers: Liquid–Liquid Interfaces 169
25
20
[mN m−1]
Br−
γ
15
F−
Cl−
10
Fig. 5 Comparison of electrocapillary
curves for the interface between
nitrobenzene solution of 0.1 M
hexadecyltrimethylammonium
tetraphenylborate and aqueous solu-
tion of 0.05 M LiF (◦), LiCl (∇), and 5
−0.2 0 0.2
LiBr (). (From Ref. [15], reproduced W−
by permission of The Chemical Society E O−− Epzc
of Japan.) [V]
170 2 Electrochemical Double Layers
14
10
0.1 0.2 0.3 0.4
W−
E O+
[V]
27 0.4
26
0.3
[mN m−1]
[F m−2]
C
25
γ
0.2
24
23 0.1
0.15 0.20 0.25 0.30 0.35 0.40 0.45
−E
[V]
Fig. 7 Interfacial tension and differential capacity as functions of applied
potential difference. The medium: 0.1 M LiCl in water, 0.1 M
tetrabutylammonium tetraphenylborate in the nitrobenzene at 25 ◦ C.
(Experimental points are taken from Refs. [17, 18].)
50
40
30
[mV]
E min
20
10
0
0.000 0.025 0.050 0.075 0.100
C
[M]
Fig. 8 Displacement of the differential-capacity minimum on
the potential axis as a function of electrolyte concentration in
the water-nitrobenzene system. The medium: 0.1 M LiCl in
water and 0.1 M tetrabutylammonium tetraphenylborate in
nitrobenzene. (Constructed from data of Samec and
coworkers [19].)
where six coefficients A+ and A− can Ions can be adsorbed specifically if the
be determined from the six boundary main contribution to their interaction with
conditions: the interface (ions, dipoles) is caused by
φ(−∞) = w non-coulombic short-range forces. Specific
org φ
adsorption cannot be explained using
φ(∞) = 0 only the theory of diffuse double layer.
Specifically adsorbed ions penetrate to the
φ(x2w − 0) = φ(x2w + 0)
(34) compact layer and form a compact or loose
org org
φ(x2 − 0) = φ(x2 + 0) monolayer. The surface passing through
the centers of specifically adsorbed ions is
εw φ (x2w − 0) = εh φ (x2w + 0)
usually called the inner Helmholtz plane. If
εh φ (x2
org org
− 0) = εorg φ(x2 + 0) several kinds of specifically adsorbed ions
174 2 Electrochemical Double Layers
are present, each ion can have its own and the potential dependence of the ca-
inner Helmholtz plane. pacitance is strongly influenced by the
It was found while studying the mech- adsorption isotherms owing to the interfa-
anism of electron transfer (ET) across the cial ionic association.
interface between two immiscible liquids When there is specific adsorption of
that specific anion adsorption occurs at ions dissolved in phase α the condition
the octane–water interface in the presence of electroneutrality can be written as
of metalloporphyrins. This adsorption in-
creases in the order of Cl− , Br− , I− , and q β = −q α = −(qiα + qdα ) (36)
is caused by coordinative bonding of the
anions as ligands of the porphyrin metal where qi is the charge of the in-
atoms. ner Helmholtz plane. The separation of
Specific ion adsorption at the po- qα from qiα and qdα cannot be done
larizable nitrobenzene–water interface without introducing a model of the
containing monolayers of phosphatidyl- interface.
choline, phosphatidylserine, octaethylene Although capacitance of the interface
glycol monodecyl ethers, tetraethylene can be calculated as before using the
glycol monodecyl ether, and hexade- equation
cyltrimethylammonium (HTMA+ ) was β
dαβ φ dαβ φi dφdα dφd
studied in detail. HTMA+ exhibited no = + + (37)
specific adsorption in the potential range dq dq dq dq
in which the aqueous phase was pos- the second term in the right side of
itive [21], whereas a strong adsorption the equation is not the diffuse-layer
occurred in the potential range in which capacitance since the charge of a diffuse
the electric potential in the aqueous phase layer in phase a is equal to
was negative with respect to the nitroben-
α,j
zene (Fig. 9). qdα = −qβ − qi (38)
Another example of specific ion ad- j
sorption was discussed in terms of the
formation of interfacial ion pairs between The diffuse-layer capacitance is equal to
ions in the aqueous and the organic
dqα
phase. The contribution of specific ionic Cd = − (39)
adsorption to the interfacial capacitance dφdα
can be calculated using the Bjerrum the-
and one can write
ory of ion-pair formation. The results
α,j
show that a phase boundary between two d qi
immiscible electrolyte solutions can be de-
j
scribed as a mixed solvent region with C −1 = (C i )−1 + (Cdα )−1 1 +
dq β
varying penetration of ion pairs into it,
depending on their ionic size. The capac-
+ (Cd )−1
β
itance increases with increasing ionic size (40)
in the order Li+ < Na+ < K+ < Rb+ <
Cs+ . Yufei et al. [22] found that significant The drop of a potential in the compact layer
specific ion adsorption occurs at the inter- depends on the surface charge qα and the
α,j
face between two immiscible electrolytes charge of the inner Helmholtz plane qi .
2.3 Electrochemical Double Layers: Liquid–Liquid Interfaces 175
−
+
−
+ −
+ +
+ − +
− +
−
+ −
+ −
−
+
−
+
−
− +
−
NB W
+
φO
φW
OHP OHP
(NB) (W)
Fig. 9 Schematic representation of the double-layer structure of the interface between
nitrobenzene and aqueous solutions in the presence of the specific adsorption of
hexadecyltrimethylammonium ions [21]. (Reproduced by permission of the Chemical
Society of Japan.)
It shows that or
dφi 1
≡
∂φi dq β Ci
dφi = dq β
∂q β α,j
qi ∂φi α,j
∂φi dqi
= +
∂φi α,j
∂q β q α,j ∂q
α,j
α,k dq
β
i q β ,qi
+ α,j
dqi (41) i
Oil A
A
pQ Q
In all these equations, X designates the for this state with real particles A, O, W
mole ratio of corresponding substances. becomes
Substituting these equations into (44), one
NAs
obtains: XsA = (54)
NAs + NOs + Nws
pµ0,s 0,b 0,s 0,b 0,b
Q + rµA − µB − pµo − pnw µw NOs
XsQ = (55)
(XbA )r Xsb NAs + NOs + Nws
+ RT ln = RT ln
(Xbo )p (Xbw )pnw (XsQ )p Nws
Xsw = (56)
(50) NAs + NOs + Nws
Using the standard Gibbs free energy of
adsorption and, we can obtain:
XsA Xs
sb G0 = µ0,s 0,b 0.b
B − rµA + pµo XsB = ; XsQ = s o s (57)
XsA
+ Xos XA + Xo
0,s
+ pnw µ0,b
w − pµQ (51)
The adsorption isotherm can then be
one obtains the adsorption isotherm: presented in the form
XsB (XbA )r sb G0 XsA
= b p b pn exp − (Xs + Xso )p−1
(XsQ )p (Xo ) (Xw ) w RT (XsQ )p A
(52) (XbA )r sb G0
We considered the 2-d solution of surfac- = exp − (58)
(Xbo )p (Xbw )pnw RT
tant B in the solvent of quasi-particles Q,
in which the mole ratios were defined as
In the past, the adsorption isotherm was
Ns NQs presented in terms of the fraction θ of the
XsB = s B s ; XsQ = s (53) surface actually covered by the adsorbed
NB + NQ NB + NQs
surfactant.
Some authors prefer another set of If we introduce η as the ratio of areas
definitions when real particles in the occupied in the interface by the molecules
interface are considered. The equation of surfactant and oil, the mole fractions
178 2 Electrochemical Double Layers
CH2
CH H CH3
H3C CH2CH3
NH N
H H
H3C N HN
CH3
H
HH
CH3OOC
Water O
O
O
Octane Octane Octane Octane Octane
P
h
y
Octane t
phase o
l
organic (β) phase with its image in the uniform charge distribution and can con-
aqueous phase (α) sitting on the same tinuously pass between the two phases. For
axis perpendicular to the dividing plane is a point charge at long distances from the
given by: interface, electrostatic Gibbs free energy
can be written as
εα − εβ (ze0 )2
F (h) = − (71) (ze0 )2 εα − εβ 2a
εα + εβ 16πε0 εβ h2 G = 4+
32πε0 εα a εα + εβ h
where h is the distance from the interface. (72)
If εα > εβ , the charge in the nonpolar where a is an ionic radius, h is the distance
phase is attracted to its image, but if from the interface and ze is the charge of
εα < εβ there is repulsion between the an ion.
charge in an organic phase and its image. The potential φ for the spherically
From Eq. (71) it follows that charges in symmetrical charge distribution is
the organic phase are attracted to the
ze0
oil–water interface. Image forces attract
4πε 0 εα r1
the diffuse layer on the organic side,
making it thinner, and repel and thicken ze0 εα − εβ
φ= + (in region α)
the aqueous diffuse layer. 4πε0 εα r1 εα + εβ
The Kharkats–Ulstrup model [25] incor-
ze0 2εβ
(in region β)
porates the finite radius of an ion a, which 4πε0 εβ r1 εα + εβ
is assumed to have a fixed spherically (73)
2.3 Electrochemical Double Layers: Liquid–Liquid Interfaces 181
a a
h h h h
εβ εα
εβ εα
P P
(a) (b)
Using standard procedures of calculations, (1 + h/a)(1 − 2 h/a)
×
Kharkats and Ulstrup obtained equations 1 + 2 h/a
for the electrostatic part of the free Gibbs
a 2h
energy of a finite-size ion. + ln 1 +
2h a
When h > a
2 2
(ze0 ) 2εβ h
(ze0 )2 εα − εβ 2a + 1−
G = 4+ 16πε0 εβ a εα + εβ a
32πε0 εα a εα + εβ h (75)
2 If h = 0 and center of an ion is at the
εα − εβ
+ interface, electrostatic Gibbs energy will
εα + εβ
be equal to
2 a 2h + a
× + ln (ze0 )2
1 − (2h/a)2 2h 2h − a G(h = 0) = (76)
(74) 4πε0 (εα + εβ )a
The first term in (74) is the Born solvation
The Figure 13 illustrates the effect of
energy, and the second is the interaction
image forces on the electrostatic Gibbs
with the image. For a charge located in
energy of an ion at the oil–water interface.
region β, the electrostatic free energy is ob-
Image forces play a significant role in
tained from (74) by exchanging εα ⇔ εβ .
electric double-layer effects. The excess
Kharkats and Ulstrup [25] obtained the
surface charge density is
expression for the electrostatic free energy
in region 0 ≤ h ≤ a: hc
g(h)
q = ec0 exp − − 1 dh
ha kB T
(ze0 )2 2h (77)
G = 2+
32πε0 εα a a where
εα −εβ 2h εα −εβ 2 (ze0 )2
+ 4− + g(h) = G(h) − (78)
εα +εβ a εα +εβ 8πε0 εβ a
182 2 Electrochemical Double Layers
Sequoia S. A.)
20
ε 2 = 20
12
ε2 = 50 ε1 = 78
−6 −2 0 2 6
h /a
dψ(0) e02 2 0
ψ(x) = ψ(0) + x (91) κ02 = lim κ 2 = zi ni (99)
dx x→∞ ε0 εkT
i
Additional functions in Eq. (173) are
The boundary conditions for these equa-
4πd 3 0 tions are
ζi (x) = 1 + i ni
3 1. ψ(x), ψ (x) → 0, if x → ∞
x+d 2. ψ (0+) = −δ/ε0 ε (100)
+π n0i
d 3. ψ(x), ψ (x) are continuous at x > 0
i max ,x−d
2
This set of equations completes the
× [(x − x)2 − a 2 ]g0i (x ) dx (92) mathematical formulation of the problem,
but the equations are very cumbersome
F
L1 (ψ) = [ψ(x + d) + ψ(x − d)] and can only be solved numerically. The
2 major difficulty is related to the finite ion
F − 1 x+d size. If we consider the limiting case of
− ψ(x ) dx (93) point charges, the equation for potential in
2d x−d
2(1 + κx) exp(−κd) the double layer would be:
+2(1 − κd) exp(−κx)
d2 ψ 1 0
=− ni zi e0
a
dx 2 ε0 ε
i
+2κ dx r exp(−κr) dr
2 −1
zi e0 zi e0
x × exp − ψ(x) +
kT 8πε0 ε
f
+ [exp(−κd)
f
κx
× κ0 − κ + exp(−2κx)
+2κ(d − x) exp(−κx)
2x
− exp(−κ(2x + d))] (101)
F= ,
2[exp(−κd)(1 + κ(x + d))
In comparison with the Gouy–Chapman
+ exp(−κx)]
theory, the MPB equation for point
charges contains two improvements in
f
− [2κx exp(−κx)+(1+κd)
that the Debye–Huckel parameter κ de-
κx
pends on distance, and the ion image is
×(exp(−2κx) − 1) exp(−κd)] screened.
d In solving the MPB equations, differ-
≤x≤d (94)
2 ent estimates of ζi (x) were considered.
Depending on the type of ζi (x), the equa-
(1 + f δ2 )
F= , if x ≤ d (95) tions were called MPB1, MPB2, MPB3 and
(1 + κd)(1 − f δ1 ) so on.
1 The MPB theory overcomes certain lim-
F0 = lim F = (96)
x→∞ 1 + κ0 d itations of the Gouy–Chapman model by
taking into account ionic volume, the ionic
1
δ2 = exp[κ(d − 2x)] sinh κd (97) atmosphere, and image forces. Results
2κx of both theories coincide at low elec-
(κd cosh κd − sinh κd)δ2 trolyte concentrations and small surface
δ1 = (98) charges, but beyond this limiting case,
(1 + κd) sinh κd
186 2 Electrochemical Double Layers
there is a very significant qualitative and that MPB4 describes the experimental
quantitative discrepancy. For instance, the results at the nitrobenzene/water and
MPB theory predicts a thinner diffuse dichloroethane/water phase boundaries. It
ion layer and higher capacitance than the reproduces the Monte Carlo (MC) cal-
Gouy–Chapman model. Near a charged culation more accurately than the MVN
interface, MPB predicts considerable ad- theory for 1 : 1 electrolytes over a wide
sorption of co-ions. The major difference range of conditions including variations
is that the decaying potential distribution in the image forces, ion size, the inner-
of the Gouy–Chapman model transforms layer potential distribution, electrolyte con-
into the decaying-oscillating solution of the centration, surface charge density and
MPB theory, with oscillations beginning solvent effects. They used the equa-
at high electrolyte concentrations when tion for electrostatic potential from the
κ0 d > 1.24. This oscillation was also pre- MPB4 model for the inner layer at ITIES
dicted by Stilinger and Kirkwood [27] who when x < a/2:
estimated a different value κ0 d > 1.03.
q aq
Other modern theories predict oscillatory ψ(x) = ψ(0) + xψ (0) − xλc|q aq |1/2
behavior of the electrostatic potential, but εh
under different conditions. In the HCA (102)
theory, the condition is κ0 d > 1.23, which where ψ(x) is the electrostatic potential at
is very close to MPB theory. The Bo- coordinate x, ψ(0) is the electrostatic po-
golyubov–Born–Green–Ivon theory puts tential at the surface, ψ (0) is the derivative
the condition at κ0 d > 1.72. Therefore, os- of ψ(0), c is the electrolyte concentra-
cillatory behavior is a consistent finding of tion, qaq is the surface charge density
all theories. It is worthwhile to mention of the aqueous phase, λ is the parame-
that damped oscillation in the MPB theory ter determined by fitting the experimental
is due to a fluctuating potential and its values of the Galvani potential differences
existence does not depend on the excluded and εh is the dielectric permittivity of
volume. the compact layer. The third term in
One of the major drawbacks of MPB Eq. (102) represents the dependence on
theory is that it does not take into ac- electrolyte concentration, surface charge
count the discreteness of a solvent. In the density, and the solvent effect on the inner-
diffuse layer, the discreteness of solvent layer potential distribution. These effects
molecules strongly influences ion–ion can be ascribed to ionic penetration into
interactions at small distances. Further- the opposite phase and ion–ion correla-
more, an understanding of the structure tions across the interface. Cui et al. [29]
and properties of the compact layer is obtained good agreement between theo-
impossible without an adequate model retical calculations and experimental data
of solvent molecules and their interac- and came to the conclusion that the
tions with the electrode and between structure of the 1,2-dichloroethane–water
themselves. interface is similar to that of the wa-
Application of MPB model to ITIES ter/nitrobenzene interface, except that the
was made first by Torrie and Val- effects of the image force and ion size
leau [28]. Cui and coworkers [29] applied are more pronounced and the inner-layer
the modified Poisson–Boltzmann the- potential drop plays a more important
ory (MPB4) to the ITIES and found role.
2.3 Electrochemical Double Layers: Liquid–Liquid Interfaces 187
the basis of relationships, developed for the edl and adsorption parameters (e.g.
the liquid Hg surface without any con- Eσ =0 , interaction parameter in the adsorp-
sideration of the structure of the solid tion (Frumkin’s) isotherm, maximal Gibbs
surface. But the theoretical analysis of the adsorption max , adsorption equilibrium
electrochemical characteristics for the PC constants, and so forth), obtained for the
solid electrodes, even for simple metals, is PC electrodes, often may be apparent [1–5,
considerably complicated in comparison 7, 15, 16, 20–22].
with that for Hg. One of the reasons is It may be noted that for a PC sur-
both the differences between properties of face, there are complications in defining
individual faces of the metal and the in- the PZC because one does not have
fluence of various defects of the surface zero-charge density of all crystal faces si-
structure [1–10]. In the beginning of the multaneously, exposed on PC electrode.
1970s, by a careful experimental study of The work function is different for vari-
the adsorption of pyridine on the solid ous homogeneous regions (planes), but if
bismuth drop electrode of known PC sur- the grains are in contact, the electrochem-
face structure and of tetraalkylammonium ical potential of the electron has a single
ions on PC Zn, the phenomenon of split- value. Thus, the Volta or outer potential
ting the adsorption–desorption maximum is different for various faces, so differ-
was discovered [1, 20, 21] (Fig. 1). Already ent surface charge densities are present.
the first investigations showed that split- Similarly, the surface structure of solid
ting the maxima was due to the PC nature metal makes the electronic structure more
of the surface, and also to the nature of complicated than that for a liquid metal or
the adsorbing particle and the solvent, to for a single-crystal plane. Certainly, the use
the surface charge density σ , and to the so- of the density properties depending only
lution concentration. On account of that, on distance perpendicular to the interface
140 60
120 50
4
100 3 40
[µF cm−2]
[µF cm−2]
80
C
30
60
20
40 2 1
1
20 10 2
3
0 0
−2 −1.8 −1.6 −1.4 −1.8 −1.3 −0.8
E (SCE) E (SCE)
(a) [V] (b) [V]
Fig. 1 Differential capacitance versus electrode potential dependence: (a) for PC Zn in
0.1 M KI containing TBAI (M): (1) 0, (2) 3 × 10−4 , (3) 3 × 10−3 , (4) 3 × 10−2 ; (b) for PC Bi
in 0.1 M KF containing pyridine (M): (1) 0, (2) 0.1, (3) 0.2 (updated from Refs. [1, 20, 21]).
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 191
charge σj at individual faces j may be dif- with σ = 0. The same conclusions hold
ferent at E = const and this may change for PC Au (σAu PC = −0.03 C m2 ), PC Bi
the form of the capacitance curve near (σBi = −0.01 C m−2 ), and for other PC
PC
the diffuse-layer capacitance minimum [1]. electrodes [5].
Important results were obtained in a pio- Mathematical simulation of CPC , E
neering paper by Valette and Hamelin [25, curves shows that the shape of the diffuse-
26]. They compared experimental capaci- layer capacitance minimum depends on
tance curves for a PC Ag electrode and Ag the difference of Eσ =0 for individual faces,
faces. It was found that the capacitance of exposed on the PC electrode surface, and
a PC Ag electrode can be obtained by the their fractions, as well as on the shape of
superposition of the corresponding Cj , E partial Cj , E curves (Fig. 3). The results of
curves for individual faces exposed at the experimental capacitance studies at two-
PC surface: plane model PC Bi electrodes were in
agreement with these conclusions [5, 15,
CPC (E) = θj Cj (E) (5)
16]. Thus it has been shown that the
j
potential of the diffuse-layer capacitance
where θj is the fraction of the face on minimum for a PC electrode with notice-
the PC electrode surface. A weighed sum able difference in Eσ =0 values for various
of C, E curves for the faces was found planes does not correspond to the PZC
to be similar to the C, E curve for a PC of the whole surface and j θj σj = 0 at
electrode, and thus all main Ag faces are Emin .
exposed on the surface. The diffuse-layer
capacitance minimum potential Emin PC for
2.4.1.5.2 Parsons and Zobel Plot Method
a PC Ag electrode was only slightly less Substantial contributions to the interpre-
negative (30 mV) than the PZC of the tation of the experimental data for solid
Ag(110) face (i.e. for the face with the
electrodes have been made by Leikis
more negative value of Eσ =0 ). The diffuse-
et al. [28] and by Valette and Hamelin [25,
layer capacitance minimum for PC Ag was
26]. Both approaches are based on the
wider and less deeper than for the Ag
same model: the value of Eσ =0 and of the
faces.
inner-layer capacitance per unit ‘‘true’’ sur-
Thereafter, the influence of the crys-
face area, Ci , are assumed to be constant
tallographic inhomogeneity of PC and
over the whole PC surface and the value
monocrystalline electrodes (with various PC was identified by E
of Emin σ =0 . Thus, the
surface defects) has been discussed for
GCSG model is considered as applicable
various metals in many papers [1–5, 7,
to the capacitance characteristics related
15, 16, 22] (Fig. 2). Bagotskaya et al. [27]
to the unit of ‘‘true’’ surface area, which
showed that integration of the partial C,
differ by a factor R −1 from those per unit
E curves from Eσ =0 of each face to
‘‘apparent’’ surface area
the minimum potential of the CPC , E
PC , on the PC electrode, taking
curve, Emin Creal (E) = R −1 Cgeom (E) and
into account the fraction of plane, gives
PC = −0.04 C m−2 at E PC . At E PC , the
σAg min min σreal (E) = R −1 σgeom (E) (6)
Ag(110) plane has a positive σ and other −1 −1
[Ci (σreal )] = [Creal (σreal )]
planes have a negative σ , thus, at the PC
Ag surface there are no surface regions − [Cd (σreal )]−1 (7)
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 193
20
1
3
15
[µF cm−2]
C
4 5
10
7
9
10
5
−1.6 −1.1 −0.6
E vs SCE
[V]
Fig. 2 C, E curves (ν = 210 Hz) for (BiDE )R (1, 4, 8), Bi(111)EP (2, 6, 10),
(BiPC )MP CE (3, 7), and Bi(111)CE (5, 9) in aqueous solution of NaF, M: (1,
2) – 0.1; (3–6) – 0.01; (7–10) – 0.002.
where Cd (σ ) is the diffuse-layer capac- shown that the test is somewhat more
itance obtained according to the Gouy complicated [1–5, 7, 29]. The PZ plots
theory [1–5, 7]. Extrapolation of Cgeom −1 for Hg–surface-inactive electrolyte solu-
−1 −1
versus Cd to Cgeom = 0 provides the tion interfaces at σ = 0 and σ < 0, albeit
−1 usually linear, exhibit reciprocal slopes that
inner-layer capacitance as R −1 Ci,geom . The
are somewhat greater than unity. The main
inverse slope of the PZ plot at EminPC is
reason for this is the experimental error in
identified by R. In the absence of ion- measuring C, as well as the hyperbolic
specific adsorption and for ideally smooth form of C in the GCSG model [15, 16,
surfaces, these plots should be linear 22, 29].
with the unit slope. However, data for The error in Ci is the total differential of
Hg and single-crystal face electrodes have Eq. (2):
194 2 Electrochemical Double Layers
40 40
[µF cm−2]
[µF cm−2]
C
C
4
20 20
1 2 3 1 2 3
15
1
2
3
[µF cm−2]
4
C
10
0.75 1.00
−E (SCE)
(c) [V]
Fig. 3 Theoretical C, E curves (1, 2, and 3) for single-crystal planes and for a model PC
surface (4) calculated by the superposition of the C, E curves at E = const with
θ1 = θ2 = θ3 = 1/3, (1) faces with high hydrophilicity; (2, 3) faces with weak
hydrophilicity. (1, 2) Eσ =0 (max) = 0.4 V and (3) Eσ =0 (max) = 0.09 V.
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 195
0.2
1
2
0.16
3
4
[cm2 µF−1]
0.12
1/C(exp)
0.08
0.04
0
0 0.04 0.08 0.12 0.16
1/Cd(theor)
[cm2 µF−1]
Fig. 4 (Ci )−1 versus (Cd )−1 dependence for wedge-shaped two-plane electrode: (1) 0.5(011̄) +
0.5(1̄01̄); (2) 0.8(011̄) + 0.2(111); (3) 0.3(011̄) + 0.7(111); and (4) 0.5(011̄) + 0.5(111) with
Eσ =0 = 15 mV (1) and Eσ =0 = 70 mV (2–4) (Ci , inner-layer capacitance; Cd , diffuse-layer
capacitance) (updated from Refs. [15, 16]).
196 2 Electrochemical Double Layers
fR was assumed to be the actual surface factor R of solid PC surfaces and to test
roughness factor independent of cel and σ . the GCSG theory on the basis of Eqs. (6)
The more dilute the solution and the larger and (7). For each cel , a set of Ci , σ curves
the difference between Eσ =0 of individ- was calculated for various R values and
ual planes (i.e. the homogeneous regions) the optimum value of R was selected on
exposed at a PC surface the higher the the basis of the assumption that near
values of fCR have been established. Thus, Emin the Ci , σ curve must be smooth.
fCR decreases as |σ | increases. A compar- The experimental values of R were found
ison of the data for F− , BF4 − , and ClO4 − to increase as cel decreased from 1.40 to
solutions shows that R increases in the or- 1.80. This was explained by the fact that R
der F− < BF4 − < ClO4 − with increasing is a complex quantity, being R = fR fCR .
weak specific adsorption [5, 15, 16, 22]. Later, the influence of the crystallographic
Thus, the experimental value of R de- inhomogeneity of PC and monocrystalline
pends on the surface charge density and electrodes (with various surface defects)
sometimes on the electrolyte concentra- was discussed in many papers [1–5, 7, 15,
tion cel . The real R cannot depend on 16, 22, 25–29] (Fig. 5).
σ and on cel . Experimental PZ plots for
single-crystal faces in the region of Eσ =0 2.4.1.5.4 Polycrystalline Electrode Models
(−3 < σ < 3 µC cm−2 ) often show slopes The edl models for PC electrodes may
increasing with |σ |, that is, the apparent be classified in two groups [5, 7, 10,
R decreases as |σ | rises [1–5, 7, 15, 16, 27, 30]. The first group considers the
22]. These findings indicate that the ex- PC electrode surface as one consist-
perimental R (the inverse slope value of ing only of relatively large homogeneous
the PZ plot) is not a real measure of the (monocrystalline) regions with linear pa-
actual R. However, the practically unit rameter y ∗ 10 nm, which corresponds
slope of the C −1 , Cd−1 -plots for Hg, Bi, to macropolycrystallinity. Within these
Cd, Sb, Ag, and for other systems with surface regions, both the compact and the
correlation coefficients better than 0.996 diffuse layers at different homogeneous
provides convincing evidence both for the areas can be viewed as independent ones,
validity of the GC theory and for the lack and accordingly
of experimentally detectable deviations of Xj Cij Cdj
app
the roughness factor from unity. Slopes of CPC = R (9)
Cij + Cdj
C −1 , Cd−1 plots much lower than unity very j
near Eσ =0 (−0.5 ≤ σ < 1 µC cm−2 ) can
where Cij and Cdj are the inner- and
be interpreted as deviations from the sim-
diffuse-layer capacities for the plane j ,
ple GC theory caused by the geometrical
respectively. Therefore, this is the so-called
roughness and energetical inhomogeneity
model of independent diffuse layers (IDL)
of the solid electrode surface [2–5, 15, 16,
(Fig. 6).
22, 25–29].
In terms of the second group of mod-
els, the PC surface consists only of very
2.4.1.5.3 Surface Roughness and Shape of small crystallites with linear parameter
Inner-layer Capacitance Curves In 1973, y ∗ , whose sizes are comparable with the
Valette and Hamelin [25, 26] proposed an- edl parameters in moderate electrolyte
other method to determine the roughness concentrations. For such electrodes, the
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 197
140
120
100 1
3 2
80
[µF cm−2]
Ci
60
4
40
20
0
−12 −8 −4 0 4
σ
[µC cm−2]
Fig. 5 The inner-layer capacitance versus electrode charge dependence for
the PC Cd electrode at different values of roughness factor: (1) 1; (2) 1.08;
(4) 1.15; (3) extrapolation by an approximate equation determined between
−1.0 and −10 µC cm−2 .
CH, n + 1(σ n + 1)
CH, m(σ m)
(b)
This model is known as the model of surface regions (y ∗ > 10 nm), and be-
the common diffuse layer (CDL) [27]. tween the large homogeneous areas there
As shown in Ref. [30], both models can are aggregates, which consist of many very
describe only some limiting cases and the small crystallites, whose sizes are probably
exposition for the total capacity of the PC smaller than the effective Debye screening
electrode (equivalent circuit) investigated lengths. Therefore, the structure of edl on
depends on the relationship between a real PC electrode can be expressed by the
the three lengths: (1) the characteristic relations
size of the individual planes at a PC app
CPC
electrode surface y ∗ ; (2) the effective
screening length in the bulk of the
n
Xj Cij Cdj
=R + Xm+1
layer near the face j ldj ≡ ld (σj ) =
diffuse
j =1
Cij + Cdj
ld / 1 + (σj /σ ∗ )2 , where ld is the Debye
screening length and σ ∗ = εRT /2πF ld ;
m
Cd (σ̄ )R Xj Cij
(3) εlij , where ε is the bulk dielectric
j =n+1
constant of the solvent and the length lij = × (11)
m
(1/4)πCij is determined by the capacity of C (σ̄ ) + R Xj C
dj ij
the compact layer of the face j . According j =n+1
to the theoretical analysis, the CDL model
n
m
is valid for PC electrodes with very small Xm+1 + Xj = 1, Xj = 1,
grains (y ∗ < 5 ÷ 10 nm) first of all with j =1 j =n+1
a moderate difference in Eσ =0 for the m>n
faces ( Eσ =0 = 0.1 ÷ 0.15 V) and for very
dilute electrolyte solutions (c ≤ 0.01 M) The results of computer simulation of
exp
near the point of total zero charge. Thus, many experimental CPC , E curves for
if the value of y ∗ is much less than two of various bismuth solid drop electrodes
the three lengths: y ∗ εli1 ∼ εli2 , ld1 , ld2 , show that 10 to 30% of the whole surface
the CDL model is valid. For other cases, of bismuth solid drop electrode is covered
the model of IDL would be valid [30]. with small crystallites (y ∗ < 10 nm), and
The real PC electrode usually consists ¯
the standard deviation ( C)
exp
of CPC , E
app
of the comparatively large homogeneous curves from CPC , E curves is smaller if
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 199
Eq. (11) is used instead of Eqs. (9) or (10). smallest characteristic length of surface
The data for the wedge-shaped two-plane inhomogeneity, R̃(κ) can be expressed as
electrode show that the share of small
crystallites at the surface of this electrode 1 − H 2
R̃(κ) ≈ R (14)
is not more than 5 to 10% [15, 16, 22]. 2Rκ 2
1
2
3
2 4
R (κ )
5
6
~
7
8
9
1
0 0.4 0.8 1.2
κ
[nm−1]
Fig. 7 Roughness function versus inverse Debye length plots of Bi(111)CE (1); Bi(111)ECE
(2); Bi(DE)R (3); and for various theoretical simulation models: sinusoidal corrugation
(SC) – h = 30, lSC = 50 (4); random Gaussian roughness (RGR) – lRGR = 50, h = 30 (5);
SC − h = 5.0, lSC = 2.5 (6); RGR − h = 2, lRGR = 1 (7); SC − h = 0.85, lSC = 2.5 (8); and
RGR − h = 0.9, lRGR = 2.0 (9) (all parameters in nm) (l, characteristic lateral correlation
length; h, root-mean-square height of surface roughness).
has been worked out in Ref. [34]. Far value R̃max > R, and the crossover occurs
from Eσ =0 the effects of nonlinear screen- at ER ≈ 65 mV (T ≈ 298 K).
ing have been found, which lead to the This model has been used for inter-
charge dependence of PZ plot, observed pretation of the experimental capacitance
experimentally in many works [1–5, 7, data of rough Cd electrodes (Fig. 8). In
25–33]. Thus, the roughness factor of PC the region of moderate Er , the rough-
surface depends on the effective diffuse- ness function rises with the decrease of
layer thickness κeff (Er ) = κ cosh(eβEr /2), the effective Debye length leff . At higher
where eβEr is the dimensionless electrode Er , the surface roughness function lev-
potential, Er = Ex − Eσ =0 is the ratio- els off to the constant value, that is, to
nal electrode potential, β = (kB T )−1 (kB the so-called geometrical surface rough-
is the Boltzmann constant and e is the ness value. In the region of Eσ =0 there
elementary charge); Ex is an electrode are deviations from the theory caused
potential. According to theoretical calcu- by the energetic inhomogeneity of elec-
lations, R̃(κ, E) is minimal close to Eσ =0 trode surface. It is to be noted that the
and rises to the limiting value at mod- new effective Debye length–dependent
erate and high σ . According to Ref. [34], surface roughness theory needs future cor-
for a fixed value of κ, the effect of sur- rections taking into account the role of
face structure on the capacitance increases crystallographic inhomogeneity of the ge-
with Er . However, in contrast to the ometrically rough electrode surface, that
case of low ER , R̃(κ, E) could be a non- is, the dependence of the local surface
monotonous function of κ with a maximal charge density on the crystallographic
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 201
1.9
1.6
R (κ , E)
1.3
~
1
2
1.0
3
4 5
0.7
−0.5 −0.4 −0.3 −0.2 −0.1 0.0
E vs SCE
[V]
Fig. 8 Experimental roughness function R̃(κ, E) versus E curves for electrochemically etched
Cd(0001) for the fixed electrolyte (NaF + H2 O) concentrations: (1) 0.05 M; (2) 0.02 M;
(3) 0.007 M; (4) 0.002 M; (5) 0.001 M.
Note: Abbreviations: TBN+ : tetrabutylammonium ion; SHE: standard hydrogen electrode; PC: polycrystalline; EP: electrochemically polished; CP:
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes
The edl at PC Ag–aqueous solution certain precautions are taken with respect
interface has been discussed by Leikis to the applied potential and the electrolyte
et al., Valette and Hamelin, Vorotyntsev, composition. The surface reconstruction
and Trasatti and Lust [2–5, 7, 25–28]. of Au(110) is more rapid than that of
A diffuse-layer minimum was found at Au(111) and Au(100). A comprehensive
Emin = −0.96 ± 0.02 V (SCE), but Emin review of reconstruction phenomena at
and Ci do not correspond to Eσ =0 single-crystal faces of various metals has
and Ciσ =0 (see also Sect. 2.4.1.5). Emin been given by Kolb [6] and Gao et al. [42].
values for PC Ag–nonaqueous electrolyte The edl at a PC Au–solution interface
interface, depending on the solvent nature, has been studied mainly by Clavilier and
have been discussed in Ref. [5]. Nguyen Van Houng, Hamelin, Borkowska
et al., and others [1–7, 43]. A diffuse-layer
2.4.2.2 Gold minimum was found at Emin = −0.04 ±
Au crystallizes in the same system (fcc) 0.02 V (SCE) for PC Au|H2 O (NaF, KClO4 ,
as Ag. Experimental data obtained by HClO4 ), but Emin obtained for PC Au
cyclic voltammetry (CV), electrochemi- does not correspond to σ̄PC = i θi σi = 0
cal impedance, STM, AFM, SXRS, and and Ci at Emin does not correspond to
other techniques (see also Chaps. 14–20 Ciσ =0 [2–5, 7]. The value of Emin de-
in Vol. 3) show that the edl and inter- pends noticeably on the solvent nature
facial properties depend strongly on the and Emin (versus bis-biphenylchromium
crystallographic structure of the electrode (BBCr(I/0)) scale) rises in the order
surface, the surface charge density, as well of solvents MeOH < H2 O < DMSO <
as the chemical composition of the solvent dimethylformamide < propylene carbo-
and the electrolyte [2–11] (Table 1). Most nate < acetonitrile [5, 43].
of the results before 1980 were obtained
with gold faces isolated by a noncontami- 2.4.2.3 Copper
nating resin and cycled before use in dilute Cu crystallizes in the fcc system. The exper-
Na2 SO4 or H2 SO4 solutions until a stable, imental data for single-crystal Cu|H2 O and
reproducible CV was obtained. From 1980 for PC Cu are controversial [2–5, 7, 44, 45].
onward, most of the results were obtained The first studies with Cu(111), Cu(100),
after a flame annealing treatment of Au Cu(110), and PC Cu in surface-inactive
planes, applied in a way similar to that orig- electrolyte solutions show a capacitance
inally used for ordered Pt electrodes [41]. minimum at E less negative than the
The surface cleanliness achieved by this positive limit of ideal polarizability of Cu
method allowed investigations to be made electrodes. More reliable values of Eσ =0 for
in a meaningful way already during the Cu single-crystal faces have been obtained
first or initial scans of the CV. by Lecoeur and Bellier [44] with EP Cu(111)
Over the past 15 years it has been and Cu(100) (Table 1). Foresti et al. [46]
demonstrated by a variety of in situ and have found Eσ =0 = −0.93 ± 0.01 V (SCE)
ex situ techniques that flame-annealed Au for the Cu(110)–aqueous solution inter-
faces are reconstructed in the same way face, and the validity of the GCSG model
as the surfaces of samples prepared in has been verified. As for Zn single-crystal
ultrahigh vacuum (UHV) [2–11] and that electrodes, reliable values of Eσ =0 have
the reconstructed surfaces are stable even been obtained indirectly from the depen-
in contact with an aqueous solution if dence of the adsorption–desorption peak
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 205
Tab. 2 Electrical double-layer parameters for Pb [5, 48] and Sn [5, 49, 50] in
various solvents
a Fractional
exponent, dimensionless parameter.
b Fractal
dimension, dimensionless parameter.
Note: Abbreviations: PC: polycrystalline; EG: electrochemically grown; EP: electrochemically polished; DMF:
N,N-dimethylformamide; i-PrOH: iso-propanol; TU: thiourea.
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes
207
208 2 Electrochemical Double Layers
The inverse capacitance of metal phase, and this can be explained by the dif-
shown in Fig. 9, obtained according to the ferent surface states of the Bi faces.
Amokrane–Badiali model [15, 16, 19, 22], Emin for PC Bi lies between Eσ =0
depends on the crystallographic structure values for various planes with σP C ≤
as well as on the chemical nature of the 0.01 µC cm−2 .
metal studied. The admittance of Bi single-crystal
faces and PC Bi in acetonitrile and
2.4.2.8 Bismuth various alcohols including LiClO4 solu-
Bi crystallizes in a rhombohedral sys- tions has been studied, and Eσ =0 has
tem. Electrochemically polished Bi single- been found to depend on the nature of
crystal faces were first used for edl the solvent. Maximal Eσ =0 for vari-
studies by Frumkin, Palm and cowork- ous faces decreases in the order of sol-
ers [1, 20, 21]. The Emin (Table 4) was vents: acetonitrile > H2 O > methanol >
observed to be shifted 30 mV toward 2-propanol > ethanol [5, 15, 16].
more negative potentials compared with Ci values increase as the atomic density
Bi solid drop electrode and EP PC Bi. of the face decreases, except for Bi(111),
Later, electrochemically polished Bi(011̄), in good agreement with the theoreti-
Bi(21̄1̄), Bi(001), and Bi(101) faces have cal calculations by Leiva and Schmick-
been studied [5, 15, 16]. The difference ler, predicting the lowest interfacial ca-
between Eσ =0 for the above faces and pacitance for the most densely packed
the Bi(111) face is comparatively big planes [18]. The capacitance of a metal
1 0.03
0.03 2 1
3
0.02 0.02 2
5 6
4 3
[nm]
[nm]
lM
lM
0.01 7
0.01
8 0
−0.01
−0.02 −0.01
−10 −5 0 −10 −5 0
σ σ
[µC cm−2] [µC cm−2]
Fig. 9 The dependence of the ‘‘effective’’ thickness of the potential drop region in the metal surface
layer, calculated by the Amokrane–Badiali model [19] (at constant capacitance of solvent monolayer,
CS ), on the charge density for aqueous solution on different electrodes (a): 1, Sb(111); 2, Sb(001); 3,
Bi(111); 4, Hg; 5, Bi(001); 6, Cd (0001); 7, Cd(112̄0); and 8, Ga; and for Bi(001) in various solvents
(b): 1, methanol; 2, H2 O; 3, acetonitrile (updated from Ref. [15, 16]).
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 209
Tab. 4 Electrical double-layer parameters of Bi and Sb [5, 15, 16, 22] electrodes in various solvents
Note: Abbreviations: AN, acetonitrile; MeOH, methanol; EtOH, ethanol; i–PrOH, isopropanol;
SbDER , an antimony drop electrode with remelted surface; PC, polycrystalline.
phase, CM , has been calculated according The influence of the surface pretreat-
to the Amokrane–Badiali model [19]. The ment of Bi single-crystal faces has been
thickness of a thin metal surface layer studied, and a noticeable dependence of
lM = 1/4πCM as a function of σ decreases Eσ =0 on the surface structure has been
in the order Bi(111) > PC Bi > Bi(101) > established [5, 15, 16, 22].
Bi(21̄1̄) > Bi(011̄) > Bi(001), in agree-
ment with other data. The value of lM max
2.4.2.9 Antimony
rises in the order AN < H2 O < MeOH < Sb crystallizes in the same rhombohedral
EtOH (Fig. 9b), which has been explained system as Bi. Sb(111) exhibits an appre-
by slight deviation of experimental systems ciably more negative value of Eσ =0 com-
from the Amokrane–Badiali model [19]. pared with the other planes (Table 4). The
210 2 Electrochemical Double Layers
Tab. 5 Electrical double-layer parameters of Fe [5, 56–58] and Ni [5, 59] in various solvents
nature of the anion [5, 41, 61] (see also et al. have studied the stepped Pt(111) sur-
Chaps. 14–17 and 19 in Vol. 3). faces [66]. The dependence of the PZTC
For Pt(111) in the aqueous solutions of for Pt(111) surfaces in acidic aqueous so-
hydrofluoric acid, Wagner and Moylan [62] lution having increasing densities of the
have estimated Eσ =0 = 0.22 V (versus re- ordered monoatomic steps in the (111) to
versible hydrogen electrode) by comparing (110) and (111) to (100) zones is evaluated
voltammetric curves and high-resolution from CO ‘‘charge-displacement’’ measure-
electron energy loss spectroscopic data of ments, with the objective of elucidating the
water + H+ coadsorption from the gas influence of the electrochemical double
phase. Iwasita and Xia [63] prepared Pt layer on the large step-induced changes in
single crystals according to the Clav- surface potential known for the clean un-
ilier et al. method [41]. According to the charged surface in UHV [67]. The PZTC
data of Fourier transform infrared (FTIR) values in 0.1 M HClO4 and 0.5 M H2 SO4
reflection adsorption spectra, the water ori- solutions decrease noticeably upon in-
entation changes from hydrogen down to creasing the (110) step density, whereas
oxygen down at 0.29 V (SHE), and this smaller effects are found for the (100)
has been taken to indicate that EQ=0 for steps. The location of the EQ=0 values
Pt(111) is close to this value. within the so-called hydrogen region, how-
Hamm et al. [64] treated a Pt(111) sur- ever, complicates the interpretation of the
face in a UHV chamber by sputtering and EQ=0 – step density dependences owing
annealing until the surface was clean and to the presence of faradaic charge asso-
well ordered. A very narrow potential re- ciated with potential-dependent hydrogen
gion, 0.13 < E < 0.28 V (SCE), is left to adsorption [66].
bare double-layer charging with a capaci- A detailed study of N2 O reduction on
tance C = 110 ± 13 µF cm−2 . Thus, only a variety of single-crystal Pt-group metal
in this region, a Pt(111) surface is free electrodes shows that N2 O reduction cur-
from surface adlayers. Thereafter the im- rent maxima occurred exactly at Clavilier’s
mersion method (IM) was used, and the Q, EQ=0 . Therefore the N2 O reduction data
E-plot has been found to go through zero have been used for obtaining the EQ=0
at E ≥ 0.80 V (SHE). Assuming a small values (Table 5). The importance of a lo-
dependence of C on E, a somewhat more cal value of EQ=0 has been emphasized,
positive Eσ =0 ∼ 1.1 V (SHE) has been esti- especially with respect to reconstructing
mated. However, it is to be noted that these metal surfaces such as Pt(100) and Pt(110),
values of Eσ =0 correspond to the potential which can be prepared in a variety of crys-
region, where the Pt surface is oxidized [5]. tallographic states [5, 65–68].
Clavilier et al. [65] have studied CO Pt and Pd electrodes with a renewed
adsorption on electrochemically faceted surface obtained by cutting with a ruby
Pt(111) and Pt(110) electrodes, and from knife under the working nonaqueous
the charge transients, these authors have solution have been studied by Petrii
provided a ‘‘definite’’ determination of et al. [69, 70] and by other investigators
EQ=0 . However, electrochemically faceted (Table 6). A more detailed discussion of
Pt(111) electrodes have a PC surface struc- data obtained is presented in Ref. [5].
ture, and thus the value of EQ=0 for
(111) Electrical double layer at solid metal
such electrodes lies between EQ=0 for alloy–electrolyte interface have been stud-
(110)
terraces and EQ=0 for steps [5]. Gomez ied, but one of the features of alloys is that
214 2 Electrochemical Double Layers
their bulk and surface compositions are, Zn and Cd both crystallize in the hcp
as a rule, substantially different because system. For both metals, the basal (0001)
one of the components is more surface- face is the most dense, and as for fcc
active than the other. It is to be noted metals, Eσ =0 is more positive than for the
that the surface composition of solid alloys other faces [15, 16, 22, 48] (Fig. 11). For
can change with time as well as through Sn planes the variation of Eσ =0 with the
surface process (selective oxidation or dis- crystal face is practically negligible [49, 50].
solution, surface migration and diffusion Being semimetals, Bi and Sb show
of components). Therefore, for the more anomalies in the correlation of Eσ =0
detailed discussion the future experimen- with the surface atomic density [15, 16,
tal studies are inevitable; however, a short 22], explained in terms of face-specific
review of experimental data is given in space charge effects. However, the definite
Ref. [5]. dependence of Eσ =0 on the reticular
density of plane has been established in
2.4.3 a good agreement with general tendency.
General Correlations The data discussed above indicate that
the difference in Eσ =0 between the most
2.4.3.1 Zero-charge Potential and dense and the most open surface depends
Reticular Density of Planes on the ‘‘softness’’ of the metal surface,
Figure 10 shows the graphical dependence which to a first approximation can be
of the PZC (or PZTC for Pt planes) on the measured by the melting point of the
crystallographic orientation of the surface given metal [2–5]. Eσ =0 is highest for
for fcc metals. The plots exhibit a typical Au (Tf = 1063 ◦ C) and lowest for Sn
pattern with minima and maxima, which (Tf = 231.9 ◦ C).
fall at the same angle for all fcc metals,
and Eσ =0 values are linearly correlated 2.4.3.2 Potential of Zero Charge, Work
with the coordination number of surface Function, Interfacial Parameter, and
metal atoms [2–5]. Inner-layer Capacitance
Eσ =0 values are correlated with the According to the basic concepts of elec-
density of atoms on the given surface trochemistry (see also Chap. 1 in Vol. 1),
and are given in Fig. 11. The PZC is the electrode potential E measured with
more positive for dense surfaces and respect to the constant reference electrode
more negative for open surfaces. Pb can be expressed as
also crystallizes in the fcc system and
therefore the same dependence of Eσ =0 E(vs. ref.) = + δχM + δχS + χion
on the crystallographic orientation should + const (19)
be expected, but Eσ =0 varies exactly in the
(S)
other direction [48]. A possible explanation where δχM = χM − χM is the perturba-
can be sought in the surface mobility of Pb tion of the electron distribution at the
atoms at room temperature, which may surface of the metal (χM is the surface
lead to extensive surface reconstruction potential of the bare metal surface in vac-
(M)
phenomena [5]. The data for Ni(001), for uum), δχS = χS − χS is due to the effect
Ni(110) [59], and for Pt planes (EQ=0 ) [68] of any solvent dipolar layer preferentially
are in good correlation with those for Au, oriented at the phase boundary (χS is the
Ag, and Cu planes. surface potential at the free surface of the
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 215
0.8
0.4
Au
0 Ag
[V vs SHE]
Eσ =0; EQ =0
Cu
Pt
−0.4
Ni
−0.8
Au
0.5
Pt
Eσ =0; EQ =0
Rh
[V (SHE)]
0 Ir
Sb
Bi
Ni Cu
−0.5
Ag
Cd Zn
−1
0 0.5 1 1.5 2
Atomic density∗10−15
[atoms cm−2]
Fig. 11 The dependence of zero-charge potential (PZTC values for Pt metals) on the atomic density
for various single-crystal electrodes (metals noted in figure) (updated from Ref. [5]).
216 2 Electrochemical Double Layers
solvent), and χion accounts for any possi- quantities are as a rule obtained in different
ble charge separation across the interface. laboratories with different techniques.
At the PZC in surface-inactive electrolyte More than with Eσ =0 , the problem is with
solutions, the ionic specific adsorption the selection of values for . The values
is absent ( χion = 0) and Eq. (19) mod- of Eσ =0 have been obtained more recently
ifies to and are usually known with accuracy to
±0.01 V or better, the measurements of
Eσ =0 (vs. ref.) = + δχM + δχS + const are rather occasional and have an
= + χ + const (20) accuracy from ±0.05 to ±0.25 eV [4, 5].
Thus, experimental data do not ensure
Since the two perturbation terms (δχM an appropriate accuracy for X and
and δχS ) are specific to the given interface the uncertainty may outweigh the value
and are experimentally inseparable, they itself. The best way to proceed is to
have grouped into a single quantity X plot Eσ =0 versus for many electrodes
and, according to Trasatti, is called the and to derive information about X
interfacial parameter [1, 4, 5, 71, 72]. term from graphical correlations using
Thus, the difference in Eσ =0 for different statistical analysis.
metals can be written as Eσ =0 = + Figure 12 shows a dependence of Eσ =0
X if the same reference electrode on . For PC sp metals, the average
is used. work function values, defined as M =
M
The main problem in Eσ =0 versus j θj j , have been used [5, 73]. Hg is
correlation is that the two experimental taken as a reference surface and a straight
Ir(111) Pt(111)
Pd(111)
5.5
(111)
(100)
(100)
(110)
(110) Ni
Au
5
[eV]
Φ
(100)
(111)
(110) (100) Cu
Fe Sb
(110)
4.5 Ag Sn
Zn (111) Hg
(100) Bi
(110)
Cd Pb
Tl In Ga
4
−1 −0.5 0 0.5
Eσ = 0; EQ = 0
[V (SHE)]
Fig. 12 The dependence of the electron work function (into UHV) on the zero-charge potential for
various PC (filled marks) and monocrystalline (open marks) electrodes (metals and faces noted in
the figure). ( ) Straight line of the unit slope through the point for Hg. For Pt-metals potentials
of zero total charge are shown.
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 217
line of unit slope is drawn through its principle, those for singular planes, specif-
point. The data show that all metals lie ically Au(111), for which ‘‘congruent’’ pairs
on the left of the straight line, thus all of data exist for and Eσ =0 [4, 5].
of them have an X term more negative Eσ =0 values for PC d-metals are not very
than that of Hg and there are no metals reliable, except Fe and Ni, and therefore
on whose surface the contact with water the points for PC Fe and Ni only as well
produces a lower than on Hg. The as for single-crystal planes are included
horizontal distance (along the Eσ =0 axis) in Fig. 12 as broadly representative of d-
of each metal point from the line of unit metals. X values for these electrodes are
slope measures X with respect to Hg. higher than those for sp and sd metals.
Thus, X measures the relative value Pt metals have never found a def-
of the contact potential difference at the inite position in Eσ =0 , correlation.
metal–water interface [4, 5]. Recently, with the improvement achieved
All sp metals, except Zn, can be in the preparation and control of sur-
apparently gathered in a single group faces, the PZTC has been estimated
and there is a clear trend within the indirectly (0.22 < EQ=0 < 0.35 V (SHE))
group for Eσ =0 to become more negative close to the PZC of PC Pt. In view of
as decreases. Thus, X increases the heterogeneity of Pt surfaces, this close-
ness may be puzzling [5]. However, if the
as decreases. Zn has much higher
PZTC estimated around 0.30 V (SHE) is
value of X than the other sp metals
taken for Pt(111) as EQ=0 , the point of Pt
with comparable work function. Since
would be located further from the line for
X ≡ δχ M + δχ S , the major contribution
d-metals, with a high value of X that is
is probably in δχ S .
not justified by the known behavior of the
In view of the large heterogeneity effect
Pt surface [5].
on the value of Eσ =0 for sd metals, the
The value of Eσ =0 for Pt(111), obtained
data for single-crystal planes are shown in
directly by the IM [64], is much more
Fig. 12. It is remarkable that each metal positive than others reported [62, 63, 68]. It
forms a separate group in which the is in the right direction with respect to a PC
faces are aligned along apparently parallel surface, even though it is an extrapolated
straight lines with the same slope as for sp value and probably corresponds to the
metals. X varies in the sequence Au < oxidized Pt(111) surface [5]. If this value
Ag < Cu and for the same metal with the is taken as PZTC and 6.1 eV assumed to
crystal face (111) < (100) < (110). These be for Pt(111), then a linear dependence
relations are in good agreement with the seems to be valid for Hg, Au(111), and
work function drop values, caused by the Pt(111), and X increases in the order
adsorption of H2 O molecules, obtained Hg < Au(111) < Pt(111). The value of
from the UHV data [4, 5]. Eσ =0 given by Frumkin et al. [1, 14] would
It is to be noted that in many cases, the imply = 5.4 eV for the PC Pt, which
preparation of a single-crystal face is not is within the usual range of experimental
followed by a check of its perfection by data. However, there remains the puzzling
means of appropriate spectroscopic tech- aspect that while EQ=0 refers to a
niques, and therefore we actually have surface with hydrogen adsorbed on it,
‘‘nominal’’ single-crystal faces. The only values do not include contributions from
values of X of high reliability are, in adsorbed H [5].
218 2 Electrochemical Double Layers
Data for single-crystal planes of sp and of a solvent as the two phases are
metals have not been plotted in Fig. 12 brought in contact (see also Chap. 14 in
as there are no correct values in the Vol. 3). The reorientation of a solvent
literature. For that reason the so-called molecule is possible if there are orienting
electrochemical work function values ∗ forces acting on it, which in the absence
have been obtained according to Frumkin of an electric field (σ = 0) can only be
et al. and Trasatti’s conception discussed the short-range chemical interactions. So,
in Refs. [1, 71, 72]. According to the data in X can also be interpreted in terms of
Fig. 13, there is a good linear correlation metal–water interaction, and the concept
between ∗ and Eσ =0 values for single- of ‘‘hydrophilicity’’ has been introduced [1,
crystal planes of sp metals, with the same 4, 5, 71, 72]. However, a quantitative com-
slope as for PC sp metals. The dependence parison (of X with ) with reliable
of X ∗ on the crystallographic orientation recent data is possible only for Ag(110),
of plane is weak, but X ∗ seems to Cu(110), and Pt(111), and the gas phase
increase with the decrease of the reticular data confirm the sequence observed elec-
density of planes, thus in the same trochemically. Thus, varies in the
direction as for fcc sd metals. order Au < Ag < Cu ( for PC Au is
X and measure change in the compared with X for the (110) face);
structure of the surface layers of a metal also, decreases Cu(110) > Pt(111) >
1
4.6 32
4
6
9
11 7 5
12
14
15 8
10
[eV]
30 13
Φ∗
31 19 17
4.2 20 16
32 23
26 24 18
33 21
28 22
27 25
29
3.8
−1.0 −0.7 −0.4 −0.1
Eσ =0
[V (SHE)]
Fig. 13 Dependence of the electrochemical work function ∗ on the pzc Eσ =0 for
various single-crystal and PC electrodes: 1, Sb(21̄1̄); 2, Sb(001); 3, Sb(011̄); 4, PC-Sb; 5,
Hg; 6, Sb(111); 7, Bi(21̄1̄); 8, Bi(1̄01̄); 9, Bi(011̄) and Sn(001); 10, Bi(001); 11, Sn(110);
12, PC-Bi; 13, Sn(11̄0); 14, PC-Sn; 15, Bi(111); 16, Pb(110) and Pb(112); 17, Pb(100); 18,
PC-Pb; 19, Pb(111); 20, In(Ga); 21, Ga; 22, Tl(Ga); 23, Cd(0001); 24, PC-Cd; 25, Cd(101̄0);
26, Cd(112̄0) and Zn(0001); 27, PC-Zn; 28, Zn(101̄0); 29, Zn(112̄0); 30, Zn(0001); 31,
PC-Zn, 32, Zn(101̄0); 33, Zn(112̄0). The values of ∗ 1–29 have been calculated
according to the equation Eσ =0 (versus SHE, V) = 1.27∗ (eV) − 5.91; and 30–33
according to equation Eσ =0 (versus SHE, V) = ∗ (eV)–5.01 (updated from Ref. [71]).
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 219
0.04
(111)
Sb Bi (111)
(211) (211) (011)
(001)
(011) (001) (112)
0.03
Hg (110) Pb
(110) Cd
(001) (0001)
(110) Sn (100)
[nm]
(1010)
0.02
li
Au
(111) (1120)
(111) (001)
In(Ga)
0.01 Ag (110) Zn
(0001)
Ga
(1010)
(1120)
0
0 0.1 0.2 0.3
∆X; ∆X ∗
[V]
Fig. 14 Dependence of the inner-layer thickness on the interfacial parameter X for various
polycrystalline sp metals (filled marks) and single-crystal sp and sd metal electrodes, noted in
figure (updated from Refs. [5, 71]). For single-crystal sp metals, X ∗ values have been used.
220 2 Electrochemical Double Layers
0
Au(111)
Ag(111)
Ag(100)
2 Hg
−10
Hg
[kJ mol−1]
∆(∆G 0A)
3 Sb
4 Bi
In(Ga)
5 Sn
−20
Pb
6 Ga
Cd
Zn(0001)
−30
−0.1 0.1 0.3 0.5
∆X
[V]
Fig. 15 Dependences of the difference ( G0A ) for various alcohols on the interfacial
parameter X for various metals. The number of each plot is equal to the number of carbon
atoms of given aliphatic compound (updated from Refs. [5, 71]). ( G0A ) = G0A − G0sol
where G0A is the Gibbs energy of adsorption of an organic compound at a metal–electrolyte
interface and G0sol is the increase of Gibbs energy of organic compound adsorption, caused by
the addition of electrolyte into the solution.
222 2 Electrochemical Double Layers
−10
22
21
−14
20 16
[kJ mol−1]
∆(∆G 0A)
19
18
17
12 10 2
15 9 4
−18 7 1
14 13 5
11 8 6
3
−22
0 0.01 0.02 0.03 0.04
li
[nm]
Fig. 16 Dependence of the difference ( G0A ) for cyclohexanole adsorption on the thickness of
inner layer at σ = 0 in the solution of surface-inactive electrolyte for various electrodes: 1,
Sb(111); 2, Bi(111); 3, Sb(21̄1̄); 4, PC-Bi; 5, Bi(1̄01̄); 6, Sb(001); 7, Bi(21̄1̄); 8, Bi(011̄); 9, Bi(001);
10, Pb(001); 11, Pb(110); 12, Hg; 13, PC-Pb; 14, Sn(001); 15, Pb(111); 16, PC-Cd; 17, Cd(0001);
18, Cd(101̄0); 19, Cd(112̄0); 20, Zn(0001); 21, Zn(101̄0); 22, Zn(112̄0) (updated from Ref. [71]).
factors are behind this phenomenon. X deviation from the general plot, which is
and water–metal interaction strength are mainly caused by the more pronounced in-
parallel, and these data confirm the hy- fluence of the capacitance of metal phase of
drophilicity sequence Hg < Au < Ag and these electrodes in comparison with other
the crystal-face sequence for fcc metals Bi and Sb single-crystal planes.
(111) < (100) < (110). Therefore the hy-
drophilicity scales of Popov et al., Silva References
et al., and Valette cannot be sustained on
the basis of sound experimental and theo- 1. A. N. Frumkin, Zero Charge Potentials,
retical arguments [5]. Nauka, Moscow, 1979.
This order of metals is in good agree- 2. A. Hamelin in Modern Aspects of Electro-
chemistry (Eds.: B. E. Conway, R. E. White,
ment with the data in Fig. 16, where the
J. O’M. Bockris), Plenum Press, New York,
dependence of ( G0A ) on the inner-layer 1985, pp. 1–102, Vol. 16.
thickness is presented. To the first approx- 3. A. Hamelin, T. Vitanov, E. Sevastyanov et al.,
imation, this plot can be considered linear J. Electroanal. Chem. 1983, 145, 225–264.
and the adsorption activity of organic com- 4. S. Trasatti, L. M. Doubova, J. Chem. Soc.,
pound at the electrode–solution interface Faraday Trans. 1995, 91, 3311–3325.
5. S. Trasatti, E. Lust in Modern Aspects of Elec-
decreases as the thickness of double layer
trochemistry (Eds.: R. E. White, B. E. Conway,
decreases, thus as the hydrophilicity of J. O’M. Bockris), Kluwer Academic/Plenum
electrode increases [71]. Data for Bi(111) Publishers, New York, London, 1999,
and Sb(111) planes show a noticeable pp. 1–216, Vol. 33.
2.4 Electrical Double Layers. Double Layers at Single-crystal and Polycrystalline Electrodes 223
Sample
z
y
x
Piezoelectric scanner
First measurements of the electrostatic with the DLVO theory, named after Der-
double-layer force with the AFM were jaguin, Landau, Verwey, and Overbeek [12,
done in 1991 [9, 10]. The electrostatic 13]. The DLVO theory uses a similar ap-
double layer depends on the surface charge proach as in the Gouy–Chapman (GC) [14,
density (or the surface potential) and the 15] and Debye–Hückel [16] models of the
ionic strength. A brief introduction to the electric double layer. To explain the coagu-
theory of the electrostatic force is given lation of sols after the addition of salts, the
in Chap. 4. The electrostatic double-layer DLVO theory takes explicitly into account
force is in many cases responsible for the interaction between colloidal particles.
the stabilization of dispersions. An AFM The interaction is assumed to consist of
experiment can be regarded as directly two contributions: the van der Waals at-
probing the interaction between a sample traction and an electrostatic double-layer
surface and a colloidal particle (or the AFM repulsion. At low salt concentration, the
tip). Since the AFM tip is relatively small, repulsion is strong enough to keep the
this interaction can be probed locally. The colloidal particles apart. With increasing
lateral spacial resolution can be of the order salt concentration, the electrostatic repul-
of few nanometers. sion is more and more screened, and at a
certain concentration, the van der Waals
2.5.2 attraction overcomes the repulsive electro-
Surface Forces in Aqueous Electrolyte static barrier and coagulation sets in.
Van der Waals forces arise from cor-
2.5.2.1 DLVO Forces relations between time-dependent dipoles
At the end of the 19th century, it was in atoms or molecules (for a review see
well known that many colloids in aque- Refs. [17–19]). The time dependence oc-
ous medium coagulate after the addition curs either because a permanent dipole is
of salt. It was even known that di- or changing its orientation in space or be-
trivalent ions are much more efficient in cause the dipole itself is changing with
destabilizing dispersions than monovalent time. The dominant term in the van
ions [11]. The explanation for this behavior der Waals force is usually the second
was eventually given in a quantitative way frequency-dependent term, also known as
2.5 Analyzing Electric Double Layers with the Atomic Force Microscope 227
the dispersion force. Since all materials one of the surface charges is reversed. For a
are polarizable, a van der Waals force oc- more complete description of double-layer
curs between all atoms and all surfaces. forces see Refs. [18, 19].
The force between two atoms is relatively
weak and short-ranged, decreasing pro- 2.5.2.2 Early Surface Force Measurements
portional to D −7 for small separations The development of the DLVO theory
and D −8 for larger separations where the stimulated an interest in measuring forces
force is ‘‘retarded’’ owing to weaker corre- between the surfaces to verify the theory
lations between the atoms. Calculation of (review Ref. [21], see also Ref. [22]). One
the forces in many atom systems is com- of the earliest direct force measurements
plex because the forces between individual were made between two polished glass
atoms are not additive, and the modern bodies [23, 24]. One glass body was fixed,
treatment of van der Waals forces, known the other was mounted to a spring. The
as Lifshitz theory, is a combination of distance between the two glass surfaces
quantum electrodynamic theory and spec- and the deflection of the spring were
troscopic data [20]. Fortunately, assuming measured. Multiplication of the deflection
additivity is a reasonable approximation in with the spring constant gave the force.
many cases, some simple equations can be Using these simple early devices and
derived. after overcoming severe problems, it
The electrostatic double-layer force is the was possible to verify the theoretically
other contribution to the DLVO force. This predicted van der Waals force for glass
force arises because of surface charges interacting with glass across gaseous
at interfaces. Water has a high dielectric medium [25, 26]. Derjaguin, Rabinovich,
constant. Thus, surface dissociation or and Churaev measured the force between
adsorption of a charged species in water two thin metal wires. In this way they
is very common. The surface charge is could determine the van der Waals force
balanced by dissolved counterions that for metals [27].
are attracted back to the surface by the In these early measurements two prob-
electric field, but spread away from the lems became obvious: the minimal dis-
surface to increase the entropy. Together tance accessible (≈20 nm) was rather lim-
the ions and charged surface are known ited and experiments could not be done
as the electric double layer. When another in liquid environment. In more recent
surface approaches, the double layer is experiments these problems could be over-
perturbed, and the resulting force is come and the minimal distance could
known as the double-layer force. When be decreased to ≈2 nm with improved
the approaching surface charges have devices [28, 29]. Historically, however, im-
the same sign, the concentration of ions provements were made by using other
between the surfaces always increases and types of devices.
results in a repulsive force. The origin of One inherent problem of all techniques
the surface charge is dissociation, so for to measure surface forces directly is
finite dissociation constants, the surface surface roughness. The surface rough-
charge is influenced by the interaction ness over the interacting areas limits
with another surface. This phenomenon is the distance resolution and the accu-
called charge regulation, and results in less racy of how well zero distance (contact)
repulsive forces, or even attractive forces, if is defined. Practically, zero distance is
228 2 Electrochemical Double Layers
the distance of closest approach because level [30–32]. The SFA contains two
if a protrusion is sticking out of one crossed atomically smooth mica cylin-
surface, the other surface cannot ap- ders of approximately 1 cm radius between
proach any further. Two methods can which the interaction forces are measured.
be used to reduce this problem of sur- One mica cylinder is mounted to a piezo-
face roughness: either atomically smooth electric translator. With this translator, the
surfaces are chosen or the interacting distance is adjusted. The other mica sur-
areas are reduced. The first approach face is mounted to a spring of known and
was chosen with the development of adjustable spring constant. The separation
the surface forces apparatus (SFA) and between the two surfaces is measured by
the measurement and analysis of surface means of an optical technique using mul-
interaction forces (MASIF), while the ad- tiple beam interference fringes. Knowing
vantages of the AFM are small interacting the position of one cylinder and the separa-
areas. tion to the surface of the second cylinder,
the deflection of the spring and the force
2.5.2.3The Surface Forces Apparatus can be calculated.
The development of the SFA (Fig. 2) was With the SFA, the main predictions of
a big step forward because it allowed the DLVO theory were verified. In par-
to measure directly the force law in liq- ticular the electrostatic double-layer force
uids and vapors at ångström resolution was analyzed for different salts and under
Transmitted light
to spectroscope
Microscope
R objective Lid of outer box
B A drop of PZ
liquid
crystal
S Mica
surfaces
Base of
3 cm outer box
Heat-filtered
white light
Fig. 2 Apparatus to measure surface forces between two mica surfaces (SFA).
(The figure is reproduced with kind permission from J. Klein [33].) This
particular device was built to study the behavior of liquid crystals confined in the
narrow gap between the mica surfaces.
2.5 Analyzing Electric Double Layers with the Atomic Force Microscope 229
different conditions. Also, limits of the force between lipid bilayers and biological
DLVO theory became obvious. The DLVO macromolecules.
theory treats the intervening medium as
continuous, so it is not surprising that 2.5.2.5 The Bimorph Surface Force
the model breaks down when the liquid Apparatus
medium between two surfaces is only few Despite the new insights achieved with
molecular diameters in width. At high salt the SFA, there are also severe problems
concentrations and small separations, an and limits. It is difficult and slow to oper-
additional monotonic repulsive force of ate, only mica and few other materials [42,
about 1 to 3 nm range was detected in 43] provide suitable substrates, elastic de-
aqueous electrolyte [34]. It has been at- formation has to be taken into account.
tributed to the energy required to remove With the invention of a bimorph surface
the water of hydration from the sur- force apparatus, known as MASIF, some
face, or the surface-adsorbed species (sec- of these limitations could be overcome [44,
ondary hydration), presumably because of 45]. In this case the force between two glass
strong charge–dipole, dipole–dipole or or polymer spheres of typically few mil-
H-bonding interactions. Because of the limeter diameter is measured. One sphere
correlation with the low (or negative) en- is therefore attached to a piezoelectric bi-
ergy of wetting of these solids with water, morph that serves as a force sensor.
it was termed hydration force [35–38]). With the MASIF, the existence of an-
other force component, namely, the hy-
drophobic force [46], which is not de-
2.5.2.4 The Osmotic Stress Method scribed by the DLVO theory and which
In parallel, another important (although was already observed with the SFA [47],
less direct) technique for measuring forces was confirmed. The hydrophobic force is a
between macromolecules or lipid bilay- monotonic attractive force of longer range
ers was developed, namely, the osmotic (100 nm) and greater magnitude than van
stress method [39–41]. A dispersion of der Waals forces. These occur in aqueous
vesicles or macromolecules is equilibrated medium between surfaces that generally
with a reservoir solution containing wa- exhibit high water contact angles (e.g.
ter and other small solutes, which can hydrocarbon and fluorocarbon surfaces).
freely exchange with the dispersion phase. At present there is no clear understand-
The reservoir also contains a polymer ing of the mechanism of these forces. It
that cannot diffuse into the dispersion. is easy to conceive of a negative hydra-
The polymer concentration determines tion force, for example, the disruption of
the osmotic stress acting on the disper- water H-bonding near a noninteracting
sion. The spacing between the macro- surface, with a similar range as the hy-
molecules or vesicles is measured by dration force. However, it is difficult to
X-ray diffraction (XRD). In this way, one conceive of one with the long measured
obtains pressure-versus-distance curves. range. At present the focus of understand-
The osmotic stress method is used to ing these long-ranged forces is on the role
measure interactions between lipid bilay- of dissolved gas.
ers, DNA, polysaccharides, proteins, and Even with the MASIF, a severe limit re-
other macromolecules [36]. It was particu- mained: that of surface roughness. Glass
larly successful in studying the hydration and, in particular, polymer surfaces are not
230 2 Electrochemical Double Layers
atomically smooth over areas of millimeter to the piezoelectric translator, onto which
or even 10 µm diameter. This limit could the sample is mounted, while measuring
be drastically reduced with the AFM. In the cantilever deflection. The direct result
the AFM the interacting areas are typically of such a force measurement is the
10 to 50 nm in diameter. Hence, all sam- cantilever deflection, zc , versus position
ples whose surface is smooth, clean, and of the piezo, zp , normal to the surface
homogeneous over such small areas can (Fig. 3). To obtain a force-versus-distance
be quantitatively analyzed. This increased curve, zc and zp have to be converted
the number of accessible materials enor- into force and distance. The force, F , is
mously. obtained by multiplying the deflection of
Small interacting areas provide another the cantilever with its spring constant K:
principal advantage. When two solid sur- F = K · zc . The tip–sample separation
faces come into contact, the surfaces de- D is calculated by adding the deflection to
form. Even under zero load, the adhesion the position.
between two surfaces leads to the forma- The deflection of the cantilever is
tion of a contact area rather than a contact usually measured using the optical lever
point. It also leads to an indentation of technique. Therefore a beam from a laser
the two interacting surfaces. For a sphere diode is focused onto the end of the
and a flat planar surface (or two crossed cantilever and the position of the reflected
cylinders), this central indentation
δ at neg- beam is monitored by a position sensitive
ligible load is proportional to δ ∝ 3 R/E 2 , detector (PSD). Often the back of the
cantilever is covered with a thin gold
where R is the radius of the sphere and E
layer to enhance its reflectivity. When a
is Youngs modulus [48, 49]. Using typical
force is applied to the probe, the cantilever
values for the SFA (R = 1 cm, E of mica
bends and the reflected light beam moves
≈60 GPa [50]), the MASIF (R = 1 mm, E
through an angle equal to twice the change
of silica ≈60 GPa), and the AFM (R =
of the end-slope tan α. For a cantilever
50 nm, E of silicon nitride ≈150 GPa),
with a rectangular cross section of width
the estimated central indentations at zero
w, length L, and thickness h, the change
loads are in the order of 14 nm, 6 nm, and
of the end-slope (Fig. 4) is given by
0.1 nm, respectively. This was calculated
with the theory of Johnson, Kendall and 6F L2
Roberts [48, 49] using a surface energy of tan α = (1)
Ewh3
γ = 0.05 N m−1 in all the cases. Hence,
E is the elastic modulus of the cantilever
in a usual AFM measurement, the elas-
material. F is the force applied to the
tic deformation can be neglected, while in
end of the cantilever in normal direction.
SFA and MASIF experiments, deforma-
The signal detected with the optical lever
tions have to be taken into account.
technique is proportional to the end-slope
2.5.3
of the cantilever. The deflection of the
Force Measurements with the AFM cantilever is given by
4F L3 2
2.5.3.1 The Practice of Force zc = 3
= L tan α (2)
Measurements Ewh 3
In a force measurement the sample is Hence, the deflection is proportional to
moved up and down by applying a voltage the signal. One should, however, keep in
2.5 Analyzing Electric Double Layers with the Atomic Force Microscope 231
h α ∆zc
F
mind that these relations only hold under with t being the time for measuring a pixel
equilibrium condition. If the movement of of the force curve in seconds. The result
the cantilever is significantly faster than is in meter. With typically t = 0.1 ms, the
that allowed by its resonance, frequency deflection resolution is 0.1 Å. In practice
Eqs. (1) and (2) are not valid anymore and the deflection sensitivity is often limited
the signal is not necessarily proportional by thermal cantilever vibrations, which are
√
to the deflection. 4kB T /3K [103]. Here, kB and T are the
The resolution of the optical lever
√ tech- Boltzmann constant and the absolute tem-
nique is approximately 10−13 / t [102], perature, respectively. With typical spring
232 2 Electrochemical Double Layers
of spring constants are given in Refs. [105, 2.5.3.3 The Probe: Tip or Colloidal Particle
110–113]. In the AFM, the two interacting surfaces
In principle, the spring constant can are the planar sample surface and the
be calculated. The thickness and modulus surface of the AFM probe. For microscopy
of elasticity of cantilevers are, however, the most commonly used probes are the
not easy to determine, so it is desir- sharp microfabricated silicon nitride or
able to measure spring constants. Several silicon tips that provide a high lateral
methods have been described [105, 114, resolution. These probes also provide high
115] but none appears to be simple, re- resolution when measuring surface forces,
liable, and precise at the same time. but introduce the problem that the surface
In the most common methods, tung- geometry is difficult to determine in
sten or gold beads are attached to the the 10-nm regime [120–122]. This would,
end of the cantilever spring. The spring however, be necessary since typical radii of
constant can be calculated by measur- curvature are from 5 to 60 nm. In addition,
ing a change in the resonant frequency the surface chemistry of silicon nitride tips,
due to the additional mass at the end which are most frequently used for force
of the spring [116]. Torii and coworkers measurements, is rather complex [51, 52,
determine the spring constant of micro- 123, 124].
fabricated cantilevers by pressing them Alternatively, spherical particles of typ-
against a calibrated macroscopic reference ically 2 to 20 µm diameter are attached
cantilever [117]. Sader and coworkers de- to the end of cantilevers. This can be
termined the spring constant of cantilevers achieved either by gluing them with a
with a rectangular cross section from the water-resistant polymer glue [9, 10] or by
resonance frequency and the quality fac- melting [97] or sintering them onto the
tor [118]. Another promising approach is to cantilever (Fig. 5). With particles of defined
use thermal cantilever noise to determine radius, the force can be analyzed more
the spring constant [119]. quantitatively. In particular, the surface
(a) 3 µm (b) 3 µm
Fig. 5 Scanning electron micrographs of a silica microsphere glued with polymer glue to
the end of a silicon nitride cantilever (a). The polystyrene particle shown to the right was
sintered at 120 ◦ C to the cantilever (b).
234 2 Electrochemical Double Layers
charge and den Hamaker constant can gold or other materials that can be sput-
be extracted. The number of materials tered or evaporated [125, 126]. First, the
that can be brought into the shape of material is deposited onto mica. Then, a
microspheres with sufficient smoothness steel plate is glued on top. Finally, the
is, however, limited. Most frequently, sil- steel plate with the deposited material is
ica microspheres are used. Silica spheres cleaved off the mica surface. Since mica
can be made with different diameters. is extremely inert, chances are high that
They have a low surface roughness of cleavage really happens at the mica sur-
less than 1 Å (as determined by AFM) face. The surface of the deposited material
and their surface chemistry is well known. now exposed can then be used in a force
Different kinds of polymer particles are experiment.
commercially available. They are, how-
ever, much rougher than silica. Their 2.5.3.4 An Instability: The Jump-in
roughness can be decreased by anneal- Implicit in the measurement of surface
ing roughly at the glass transition tem- forces with a ‘‘spring force’’ is the as-
perature. Still, the surface roughness is sumption that the two are equal and
significantly higher than silica. Zirconia opposite and thus the spring is not in
microspheres were produced by annealing motion. However, close above the sur-
zirconia powder [83]. face, the spring often cannot provide
While the preceding paragraph concen- a sufficient force to counterbalance the
trated on the importance of knowing the surface force. These regimes are charac-
geometry of the tip, it is equally impor- terized by rapid motion of the cantilever,
tant to know the geometry of the sample. and cause difficulty for determining sur-
Again, the most convenient geometry is a face forces. The onset of this unstable
smooth sphere or a planar surface. Prob- regime (‘‘jump-in’’) is characterized by
the point at which the gradient of the
lems due to sample roughness are greatly
attractive force exceeds the spring con-
reduced in the AFM compared to other
stant plus the gradient of the repulsive
surface force techniques since the sample
force [127]. Several techniques have been
needs only to be smooth on a scale com-
employed to access the unstable regime:
parable to the radius of curvature at the
(1) the application of a force feedback
end of the tip. Several materials such as (electric or magnetic) to balance the sur-
mica, oxidized silicon, graphite (HOPG), face force [33, 128–131]; (2) measurement
MoS2 , and so on are available with suffi- of the time-dependent force and dis-
cient smoothness over the required areas placement in the unstable regime, then
(Table 1). In addition, the surface of many calculation of the velocity and accelera-
materials can be smoothed by template tion to yield the surface force according
stripping. Therefore the material, say a to Newton’s equation of motion [132];
polymer, is melted on a smooth sur- (3) use a high approaching speed so that
face [76]. This can be a mica or a silicon the hydrodynamic drag FH , which for a
wafer surface. After cooling and right be- sphere approaching a flat with a veloc-
fore an experiment, the surface is peeled ity v increases with decreasing distance
off and the freshly exposed polymer sur- as FH = 6πηvR 2 /D (η is the viscosity
face is used for the force measurement. of the liquid) [133], compensates for the
The same technique can also be used for attraction [134].
2.5 Analyzing Electric Double Layers with the Atomic Force Microscope 235
Another instability occurs when tip or jump occurs for nonelastic deformation.
sample surfaces are not perfectly rigid but In this case the jump depends on the rate
are deformed by the tip. For that case of of the plastic deformation [136].
elastic deformation, Pethica and Sutton
showed that at sufficiently small separa- 2.5.3.5 AFM-related Techniques
tions, typically 1 to 2 Å, the tip and the Commercial AFMs are made for imaging.
sample would jump together, irrespective For force measurements, certain features,
of apparatus construction [135]. A similar such as the lateral scanning capability and
236 2 Electrochemical Double Layers
the feedback electronic that regulates the fluid cell can be made of one or few mate-
height, are not required. Hence, if one rials (such as Teflon and quartz) that can
plans to use the device solely for force mea- be cleaned thoroughly, for example, with
surements, it is significantly less expensive hot sulfuric or nitric acid.
to build homemade setups [137–139]. In
addition, some features of commercial 2.5.4
AFMs are not perfectly suited for measur- Theory of Measuring Surface Forces with
ing force curves. One such feature is the the AFM in Aqueous Electrolyte
split photodiode, which has an extremely
high sensitivity but a limited dynamic 2.5.4.1 Derjaguins Approximation
range. In addition, the sensitivity depends The force between two surfaces depends
on the precise shape of the reflex and the on both the material properties and the
AFM needs to be calibrated for each ex- geometry of the surfaces. Derjaguin [144]
periment [140]. These disadvantages were approximated the influence of arbitrary
overcome in some homebuilt devices by re- geometry on the interaction potential
placing the spilt photodiode with a linear V (D) by reducing it to the simple geometry
position-sensitive device [132, 141]. With of two flat surfaces:
∞
position-sensitive devices, the accessible
force range is highly increased. This is V (D) = VA (z) dA (6)
D
particularly important when measuring
long-range forces during the approach and V (D) is the interaction potential that
the adhesion force during the retraction depends on the distance between the in-
in one experiment. With microfabricated teracting surfaces D. VA (z) is the potential
tips, long-range forces in aqueous elec- energy per unit area between two flat sur-
trolyte are of the order of 0.1 to 1 nN. faces that are separated by a distance z.
Adhesion forces can be 100 times higher. dA is the increase of the cross-sectional
To have a sufficient sensitivity as well as area of the two surfaces with increasing
the capability to measure the strong ad- separation distance.
hesion force, a position-sensitive device is According to Eq. (6), the fundamental
necessary. property of the material is the interaction
The position of the sample is adjusted by energy, VA , (or force, FA = −dVA /dz) per
the piezoelectric translator. Piezoelectric unit area. The most useful measurement
crystals show creep and hysteresis, which is the one between two surfaces of known
affect the accuracy of the distance determi- geometry, so that VA can be determined
nation [142]. One possibility to overcome and thus the interaction energy of all
this problem is to use piezoelectric trans- geometries can be calculated.
lators with integrated capacitative position The problems of unknown surface
sensors, which are commercially avail- geometry and chemistry can partly be
able [143]. In the same setup, another solved by attaching smooth micrometer-
deficit of commercial AFMs was overcome. sized spheres to the cantilever. For a
Standard fluid cells of commercial AFMs sphere of radius R, and a flat, a simple
are small and manually difficult to access. equation relates the measured force, F , to
In addition, they consist of different ma- the interaction energy per unit area, VA :
terials (glass, steel, silicon, etc.) that are F (D)
difficult to clean. In self-made devices the VA (D) = (7)
2πR
2.5 Analyzing Electric Double Layers with the Atomic Force Microscope 237
From this relationship, it can also be seen calculations that are also valid at higher
that a larger radius results in a higher potentials and that explicitly take into
force, and thus greater sensitivity in VA . account the geometry, see Refs. [150–154].
Naturally, this increase in VA sensitivity The requirement of low potentials is
comes at the expense of reduced lateral not the only limit of Eq. (8). In ad-
resolution. Another advantage of using dition, Eq. (8) is restricted to distances
spheres as probes is the possibility to make roughly larger than the Debye length. For
a variety of probes by attaching particles D ≤ λD , boundary conditions influence
of different chemical composition to the the force. The force is calculated from
cantilevers. Relation (7) is derived from the potential distribution between the two
Eq. (6) by assuming a parabolic shape surfaces [18, 19]. In continuum theory the
of the probe. It can also be used for potential distribution is determined from
microfabricated tips since they often show the Poisson–Boltzmann equation, which
a roughly parabolic shape. In this case, R is is a second-order differential equation.
the radius of curvature of the tip at its end. To solve this equation, certain bound-
ary conditions have to be assumed. Two
2.5.4.2 The Electrostatic Double-layer boundary conditions are often used: ei-
Force ther it is assumed that upon approach
When two charged surfaces approach each the surface charges remain constant (con-
other at some point, the electric double stant charge) or that the surface poten-
layers start to overlap and a force, also tials remain constant (constant potential).
called disjoining pressure, begins to act [13, These boundary conditions have a strong
145]. This electrostatic double-layer force influence on the electrostatic force at dis-
decays roughly exponentially. When the tances smaller than λD . Two surfaces with
surface potentials of both surfaces are constant charge of equal sign always repel
below approximately 25 to 50 mV, the force each other for D → 0. Two surfaces with
between a spherical tip and a flat sample constant potential are attracted for D → 0,
can be approximated in an analytical even when the surface potentials have the
form [146–149]: same sign (except for the hypothetical case
that the potentials are precisely equal in
4πRσ1 σ2 λD −D/λD
Fel = e (8) magnitude and sign) [155].
εε0 To demonstrate the effect of boundary
with the Debye length conditions, Fig. 6 shows the force between
a microsphere of 3-µm radius interacting
εε0 kB T with a flat surface versus distance. The
λD = force was calculated with the so-called
2n∞ e2
nonlinear Poisson–Boltzmann equation,
Here, σ1 and σ2 are the surface charge which is also valid for potentials above
densities of probe and sample, ε0 is 50 mV. A salt concentration of 1 mM
the permittivity of free space, ε is the was chosen resulting in a Debye length
dielectric constant, e is the unit charge. λD = 9.6 nm. At close distances, the force
The Debye length λD is determined by calculated with constant charge bound-
the salt concentration; n∞ is the bulk ary conditions is much higher than the
concentration of a monovalent salt in force calculated assuming constant poten-
molecules per volume. For more detailed tial. For D λD , the force calculated with
238 2 Electrochemical Double Layers
10 000
σ 1 = 0.036 e nm−2, σ 2 = 0.023 e nm−2
1000 1 mM salt
[µN m−1]
λ D = 9.6 nm
F/R
Ψ1 = 80 mV
100
Ψ2 = 50 mV
10
0 10 20 30 40 50 60
Distance
[nm]
Fig. 6 Electrostatic double-layer force acting on a sphere of 3-µm
radius in water containing 1 mM monovalent salt. The force was
calculated with the nonlinear Poisson–Boltzmann equation and
the Derjaguin approximation for constant potentials (1 = 80 mV,
2 = 50 mV) and for constant surface charge (σ1 = 0.0058 C m−2
= 0.036 e nm−2 , σ2 = 0.0036 C m−2 = 0.023 e nm−2 ). The
surface charge was adjusted by σ1/2 = εε0 1/2 /λD , so that at large
distances both lead to the same potential. In addition, the force
expected with approximation (8) is plotted as a dotted line.
constant charge and constant potentials In this model the surface charge is caused
are equal and surface charges and po- by the dissociation of ions from surface
tentials can be converted according to groups. The dissociation constant and as a
σ1/2 = εε0 1/2 /λD . Also the force calcu- consequence, the surface charge depends
lated with approximation (8) is plotted. It on the potential. This also explains the
lies between the two extreme cases. often observed dependency of the surface
Which boundary condition is more real- charge on pH and the salt concentration.
istic depends on the materials used. In The charge regulation model is success-
addition, the electrolyte and the speed fully applied to many materials, especially
of the approach might have an influ- those in which the surface charge is mainly
ence. Prica and coworkers measured force determined by pH, such as oxides, sil-
curves between zirconia, which showed icon nitride [51, 52, 123, 124], mica [34,
a constant charge behavior [84]. Also the 157–159], and biological materials.
force between two surfaces coated with
densely packed carboxylic groups followed 2.5.4.3 Other Surface Forces in Aqueous
constant charge conditions [94]. In other Electrolyte
cases, constant potential conditions are With any surface force–measuring de-
more appropriate. Most cases, however, vice, the total force between two surfaces
lay between the two extremes. Then often, is determined. The origin of all surface
a charge regulation model is applied [156]. forces is the interaction between electric
2.5 Analyzing Electric Double Layers with the Atomic Force Microscope 239
charges. Under normal circumstances it progress has been made during the last
is practical, however, to separate the to- few years in calculating Hamaker con-
tal force into several components and take stants from spectroscopic data. A list of
the components as being additive (for a Hamaker constants for different combina-
critique see Ref. [160]). These components tions of the three media involved is given
are the van der forces, the electrostatic in Refs. [163, 164]. There are some cases
double-layer force, the hydration repulsion in which the van der Waals force is re-
(between hydrophilic surfaces), and the hy- pulsive [165]. A repulsive van der Waals
drophobic attraction (between hydropho- force was, for example, detected between
bic surfaces). To analyze the electrostatic a silicon nitride tip and a silicon oxide
double-layer force, one needs to separate it surface in diiodomethane [166]. Also the
from all other components. This is usually van der Waals interaction between many
done by assuming a certain distance de- solids and vapors in water is often repul-
pendency for the van der Waals force and sive and causes the adsorption of water to
the electrostatic force. In addition, it can solid surfaces.
be identified by the influence of salt. Hydration forces [35–38] are relatively
The van der Waals potential per unit area short-ranged, so that at salt concentra-
is given by tions below 0.1 M, they can easily be
distinguished from the longer range elec-
H trostatic and van der Waals forces. They
VA = − (9)
12πD 2 are repulsive and, except for the case
Using Eq. (7), the force between a sphere of molecularly smooth surfaces and low
interacting with a plane can be calculated: salt concentrations where force oscillations
were observed [167], decay exponentially
HR with distance:
Fv dW = − (10)
6D 2
V A = Ae−D/λ0 (11)
For D 10 nm, retardation effects have
to be considered, and the force decays Characteristic decay lengths λ0 deter-
with D −3 instead of D −2 . H is the so- mined with the SFA, the osmotic stress
called Hamaker constant that depends on method, or the AFM range from 0.2 to
the dielectric properties of the interacting 1.4 nm. Typical amplitudes are A = 10−3
materials and the medium between. De- to 10 J m−2 . In contrast to the electrostatic
tailed calculations showed that Derjaguins double-layer force, hydration forces tend
approximation leads to errors that are usu- to become stronger and longer ranged
ally less than 10% [161]. For large distances with increasing salt concentration, espe-
the precise shape of the probe has to be cially for divalent cations.
considered [19, 162]. Hydrophobic forces occur between sur-
The van der Waals force is usually at- faces that are hydrophobic and show a
tractive. It is always attractive between two contact angle around 90◦ or higher. They
identical materials immersed in any third are attractive and often lead to a jump-in
material. For hydrocarbons immersed in at distances between 5 and 100 nm. They
water, the Hamaker constant is about can easily be avoided by using hydrophilic
0.2 − 1 × 10−20 J, for many oxides in wa- probes. For contact angles below 70 to
ter, it is 0.8 − 5 × 10−20 J, and for metals 80◦ , they are usually much smaller then
in water, it is 30 − 40 × 10−20 J. Great DLVO force and can be neglected [168].
240 2 Electrochemical Double Layers
Microfabricated tips of silicon nitride or structures [54, 61, 62], adsorbed protein
silicon oxide are naturally hydrophilic. One layers [99], or even whole cells [63, 101]. In
should, however, keep in mind that they all cases the force could be well described
might contaminate and as a result become by DLVO theory; independently measured
hydrophobic. surface potentials (or charge densities) and
Hamaker constants agreed with AFM re-
2.5.5 sults. Only at short separation below 1
AFM Force Measurements of the Electric to 5 nm, differences occurred [10, 65, 79,
Double Layer 176], particularly at high ionic strength.
These are usually attributed to hydration
2.5.5.1 Between Solid Surfaces forces.
With the AFM, DLVO forces were mea- As one example, the force-versus-
sured between several materials that are of distance between a silica particle and a
interest especially in colloidal science, for titania flat is shown in Fig. 7. Like in
example, glass, silica, and silicon nitride [4, many publications, the force is scaled
9, 10, 51, 53, 64–69], gold [64, 89–91], by division through the radius of the
zinc, and lead sulfide [86–88], titanium silica microsphere. According to Eq. (7),
oxide [73, 74, 82], zirconia [83–85], alu- the scaled force, F /R, is equal to 2πVA .
mina [10, 71, 72], different polymers [56, Force curves were recorded at different
76, 77, 97, 98, 169–171], LB and pH values ranging from pH 8.8 for the
other thin organic layers [57, 58, 94, 95, top curve to pH 3.0 for the bottom curve.
172], surfactant adsorbed to solid sur- The surface charges of both materials are
faces [92, 173–175], biological membrane mainly determined by the pH. Silica has
1.5
1.0 pH
[mN m−1]
F/R
0.6
0.1
−0.3
0 10 20 30 40 50 60
Distance
[nm]
Fig. 7 Force between a silica (SiO2 ) microsphere of 2.5-µm radius
and a titania (TiO2 ) crystal versus distance. The force is scaled by the
radius of the sphere. The curves were recorded at pH values of 8.8,
7.2, 6.3, 5.3, and 3.0 from top to bottom with 1 mM KNO3
background electrolyte. (The figure is reproduced with kind
permission from C. J. Drummond [73].)
2.5 Analyzing Electric Double Layers with the Atomic Force Microscope 241
an isoelectric point around pH 3.0, while Campbell and Hillier determined the force
the isoelectric point of titania is pH 5.6. between a silica sphere as probe and
As a consequence, at high pH, where carbon [75]. Figure 8 shows the result of
both materials are negatively charged, an Hillier and coworkers [179] who measured
electrostatic repulsion is observed. The re- the force between a gold sample and a
pulsion decreases as the pH decreases, silica sphere attached to the end of a
and at pH 3.0, that is, below the isoelectric cantilever. Silica is negatively charged at
point of titania, there is an electrostatic the pH ≈ 5.5 used in these experiments.
attraction as well as a van der Waals force As expected, the silica microsphere is re-
resulting in an overall attraction between pelled at negative potentials of the gold.
the two surfaces. When changing the potential to more pos-
itive values, the repulsion decreased and
2.5.5.2 High Surface Potentials finally attractive forces were observed. For
With classical surface force–measuring separations larger than 10 nm, the force
techniques, it is impossible to analyze the curves could be described by DLVO theory.
electrostatic double layer at surface poten- For smaller separations, measured force
tials significantly higher than 0.1 V. The curves deviated significantly from calcu-
reason is that high surface potentials can lated force-versus-distance curves. Hillier
only be obtained by applying an external and coworkers attribute this discrepancy
potential to a metallic or conducting sam- to several factors: an overestimation of
ple. All materials that are accessible to the Hamaker constant, the roughness of
the SFA or MASIF are not conducting. the two interacting surfaces, and hydration
Therefore it was a particular challenge to effects.
measure electrostatic double-layer forces Although the results of Hillier and
at high surface potentials with the AFM. coworkers qualitatively agree with results
In first experiments, Ishino and cowork- of Raiteri and coworkers [178] obtained
ers [177] coated a microfabricated AFM tip with silicon nitride tips, there are also
with gold and used it as working elec- some discrepancies. First, the force ob-
trode. They indeed observed a variation served between silica and gold changed
of the force with the applied potential almost linearly with the applied poten-
when measuring the interaction between tial in a range from −0.7 to +0.4 VSCE .
the gold-coated tip and a stearic acid Raiteri and coworkers observed a change
monolayer. An electrostatic repulsion was of the force only in a narrow potential
observed at negative applied potentials range of approximately 0.4 V. Above and
and attractive forces were measured at below this voltage range, the force sat-
positive potentials. Since the stearic acid urated. This saturation behavior is also
monolayer was negatively charged, such a predicted by DLVO theory. Second, Hillier
behavior is expected. and coworkers [179] always measured de-
A more convenient setup consists of cay lengths that agreed with the calculated
a conducting sample, which serves as Debye length. Raiteri and coworkers ob-
a working electrode, and an insulating served the same for potentials that lead to
probe. Raiteri and coworkers and Döppen- a repulsive force. In the attractive regime,
schmidt and coworkers [55, 178] measured however, the decay length was significantly
the force between a gold, platinum, or larger than the Debye length. These two
graphite sample and a silicon nitride tip. questions are still open.
242 2 Electrochemical Double Layers
Cantilever with
silica microsphere
RE
Potentiostat CE
Gold
sample
WE
Sample stage
1
−700 1
−500 0.8 −700
0.8 V −400
−400
−300 0.6
[mV] −200 V −200
[mN m−1]
0.6 0 0
+100 0.2 +100
0
[mN m−1]
0.4 −0.2
F/R
−0.4
0.2 0 2 4 6 8 10 12 14 16 18 20
Distance
[nm]
0
−0.2
−0.4
0 10 20 30 40 50 60
Distance
[nm]
Fig. 8 Force between a silica sphere and gold electrode in an aqueous
electrolyte solution of 1 mM KCl and pH ≈ 5.5 as a function of the applied
potential at the gold electrode. The curves correspond to, from top to bottom,
electrode potentials of −0.7, −0.5, −0.4, −0.3, −0.2, −0.1, 0, and +0.1 V
(versus SCE). Inset: Force curves recorded in a 10 mM KCl solution. (The figure
is reproduced with kind permission from A. C. Hillier [179].)
gas bubbles play a central role in many experiments were not very precise and
industrial processes such as flotation or the the distance resolution was below 10 nm.
deinking of paper. With the AFM or with The first precise experiments were done by
AFM-related setups, the electric double Ducker and coworkers [182] who observed
layers at such deformable interfaces can an electrostatic double-layer force that de-
be analyzed. It is possible to measure cayed with the expected Debye length. This
the force between a solid microsphere observation that the electric double layer
and an oil drop (or another immiscible decays exponentially with the Debye length
liquid) or a bubble in aqueous medium. even at air–water interfaces was later con-
These experiments are, however, more firmed with devices that were equipped
difficult to perform and to interpret. First, with capacitatively calibrated piezo trans-
since the interface is deformable, it is lators (Fig. 9). Ducker and coworkers de-
difficult to determine zero distance. In tected, however, an unexpected attractive
principle, the shape of the interface can be force between a presumably hydrophilic
calculated using the Laplace equation [180, silica particle and a bubble. Fielden and
181]. This is, however, not trivial and in coworkers confirmed that between hy-
many cases practically impossible. Second, drophilic particles and air bubbles only
the tip or particle can even penetrate into repulsive forces are acting [183] and their
the bubble or oil drop. In this case, a experiments can be considered as the first
three-phase contact line is formed and accurate measurements. As expected from
the capillary force completely dominates adsorption experiments, a hydrophilic par-
the interaction. Electrostatic double-layer ticle is repelled by an air bubble and a
forces can only be detected before a three- stable water film remains on the particle
phase contact is formed. surface. The previously observed attrac-
First experiments were done between tion was probably due to contamination.
differently treated silica particles and an In contrast, a hydrophobic particle imme-
air bubble [132]. As expected, they showed diately forms a three-phase contact and
that a hydrophilic particle is repelled by snaps into the bubble (or vice versa: the
an air bubble, while a hydrophobic par- bubble engulfs the particle). Before con-
ticle jumps into the bubble. These early tact, the electrostatic double-layer force
100
Decay length
decaying with the Debye length could be reverses its surface charge from nega-
observed in both cases. tive to positive and it makes the surface
Ducker and coworkers [182] and Fielden hydrophobic.
and coworkers [183] performed their ex- When increasing the DTAB concen-
periments with commercial AFMs. For tration to 5.4 mM, a small electrostatic
routine measurements this is not conve- repulsion was observed before the jump-
nient because the liquid cells of commer- in. The repulsion was at least partly due
cial instruments are small and difficult to to electrostatic repulsion since it decayed
access manually. In addition, a thorough exponentially with the Debye length. The
cleaning procedure is hampered by the reason is probably that at DTAB concen-
fact that the cell consists of many different trations above 5 mM also, the air–water
materials. Therefore, Preuss and cowork- interface becomes positively charged be-
ers built their own device to measure the cause of adsorbed DTAB. This leads to an
force between particles and bubbles [143, electrostatic repulsion between the posi-
184–186]. With this device, the influence tively charged air–water interface and the
of different surfactants on the interaction positively charged silica surface. Above
of hydrophobic and hydrophilic particles approximately 6 mM DTAB, the jump-in
with bubbles was studied. disappeared and only repulsive force were
As one example, the force between observed.
a hydrophilic silica particle and an air The interaction of a particle with an oil
bubble at different concentrations of dode- drop is relatively similar to that with an
cyltrimethylammonium bromide (DTAB) air bubble [192–194]. Both interfaces, the
is shown in Fig. 10. Without surfactant, oil–water and the water–gas, are nega-
the particle is repelled by the air bubble. tively charged around neutral pH. There-
At distances above 5 nm, the electro- fore an electrostatic repulsion is observed
static repulsion dominates. The reason when interacting with a silica particle. As
being the negative surface charges on the expected from DLVO theory, this repulsion
silica surface and at the water–air inter- decays roughly exponentially with a decay
face [187–190]. Even at close distance, a length equal to the Debye length [195].
stable water remains on the particle sur-
face and no three-phase contact is formed. 2.5.5.4 Imaging Charge Distributions
Adding even small amounts of the cationic Besides measuring the distance depen-
surfactant DTAB changes the interac- dency of a force, it is sometimes interesting
tion drastically. At concentrations between to know how the force changes from
0.1 mM and typically 5 mM DTAB (criti- one place on the sample to another. To
cal micellar concentration is ≈16 mM), no obtain this lateral information, Senden,
repulsion was observed. When the parti- Drummond, and Kékicheff [196] imaged a
cle comes into contact with the air–water silicon nitride surface at low force. When
interface, it jumps into the bubble and a imaging at very low forces (about 0.1 nN),
three-phase contact is formed. Such a be- the tip did not touch the surface because
havior can be explained with the strong electrostatic repulsion kept the tip sev-
adsorption of long-chain alkyltrimethy- eral nanometers above the surface. The tip
lammonium ions to silica [191]. At a ‘‘rode’’ on top of the double layer. In this
concentration of 0.1 mM, DTA+ forms way a constant electrostatic force image is
a monolayer on the silica surface. This obtained. When increasing the force, the
2.5 Analyzing Electric Double Layers with the Atomic Force Microscope 245
1.5
0 mM DTAB
1.0
0.5
0.0
0.0 0.1 0.2 0.3 0.4 0.5
5
0.1 mM DTAB
0
−5
−10
−15
[mN m−1]
F/R
0 2 4 6 8
0.4
15
0.2 5.4 mM
10
0.0
5
0,00 0,05 0,10
0
−5
−10
−15
−20
0 2 4 6 8
12
10 13.2 mM DTAB
8
6
4
2
0
tip came into contact with the sample and covered with micells or hemimicells of two
a topographic image was obtained. gemini surfactants. AFM images taken in
Using the electrostatic repulsion mode, the electrostatic repulsion mode allowed
Manne and coworkers studied the struc- to analyze the supramolecular structure of
ture of surfactants adsorbed to solid surfactants on solid surfaces and greatly
surfaces. They imaged the cationic sur- advanced our knowledge about surfactant
factant cetyltrimethylammonium bromide adsorption. The advantage over contact
(CTAB) adsorbed to a hydrophobic grap- mode is the significantly less destruc-
hite surface [197]. When imaging with tive interaction between tip and sample.
a very low force, they observed paral- Burgess and coworkers [201] could even
lel stripes spaced about 4.2 nm apart, visualize changes in the structure of SDS
that is, about twice the length of the adsorbed to gold(111) upon variation of
CTA+ ion. The adsorbed stripes were the potential. At negative and neutral sur-
generally observed in three orientations. face charge, SDS forms a hemimicellar
On the basis of observations like these, structure. At positive surface charge, a
the authors proposed that CTAB forms condensed monolayer is formed.
hemicylindrical hemimicells on graphite. In the described experiments the images
Manne and coworkers and others extended are not solely images of the charge
this work to other solids such as sil- distribution but they are a mixture of
ica [198], mica [175], or gold [199] and to topography and charge density. Heinz
other surfactants such as sodium dodecyl and Hoh [203] developed a protocol, called
sulfate (SDS) or dodecyldimethylammo- D minus D mapping, on how the two
niopropanesulfonate (DDAPS) [200]. As contributions can be separated. Therefore
an example, Fig. 11 shows mica surfaces they take isoforce images at different salt
(a) (b)
Fig. 11 To study the structure surfactants on surface, these AFM images (150 × 150 nm)
were recorded in electrostatic repulsion mode. They show two different gemini surfactants,
(Cm H2m+1 )[N+ (CH3 )2 ](CH2 )n [N+ (CH3 )2 ](CH2 )j with m-n-j of 18-3-1 (a) and 12-4-12 (b), in
aqueous medium on mica. The concentrations were 3.0 and 2.2 mM, respectively. Both are
above the critical micellar concentration. (The figure is reproduced with kind permission
from S. Manne [202].)
2.5 Analyzing Electric Double Layers with the Atomic Force Microscope 247
concentrations to remove topography and to calibrate the tip and then the charge
isolate electrostatic contributions to the density of the deposited material can eas-
tip–sample interaction. ily be calculated. In this way the author
To get more quantitative information on determined the charge density of pur-
the surface charge density, whole force ple membranes deposited on alumina [54].
curves must be taken at each individ- Rotsch and coworkers verified the positive
ual point of the sample. In this way, surface charge of dimethyldioctadecylam-
differences between the charge densities monium bromide (DODAB) patches on
of different regions on a sample can be negatively charged mica (Fig. 12). At neu-
obtained. This not only allows to distin- tral pH they observed a repulsive force
guish regions on the basis of their electric between the negatively charged mica and
properties but also opens the possibility to the negatively charged silicon nitride tip.
measure quantitative charge densities with On a DODAB patch an electrostatic at-
a standard AFM tip. Usually, it is practi- traction was observed, indicating that the
cally impossible to measure local charge patch is positively charged.
densities with standard tips directly. The Taking many force curves can nowadays
surface chemistry of silicon nitride and the be done automatically with commercial
size of the tip are not known accurately instruments. Usually it requires 0.1 to 1 s
enough to allow quantitative measure- to take one force curve. To obtain say
ments. To circumvent this problem, the 64 × 64 force curves lasts 7 to 70 min.
sample is adsorbed on a substrate with To increase the speed, Miyatani and
known charge density. Then, force curves coworkers developed a procedure in which
are taken on the substrate and on the de- the cantilever is moved in a sinusoidal
posited material. This substrate is used form rather than a triangular pulse [205].
30 1 µm
20
Cantilever deflection
10
[nm]
−10
2 µm
−20
−20 0 20 40 60
Translator position
(a) (b) [nm]
Fig. 12 AFM contact mode image showing a typical assembly of DODAB bilayer patches (a).
The patches were self-assembled from a vesicle suspension and imaged in pure water. To the
right, a series of approaching force curves are shown that were taken along a line across one of
the patches in 5 mM LiNO3 (b). The different electrostatic interaction – repulsive on the mica
substrate at the beginning and end of the line and attractive on the DODAB patch – is visible.
(The figure is reproduced with kind permission from M. Radmacher [204].)
248 2 Electrochemical Double Layers
This method is, however, limited by issues, the AFM has become a routine tool
the increasing viscous drag when the to measure interaction force for specific
cantilever approaches a sample [134]. particles or surfaces relevant to practical
applications.
2.5.6
Conclusion and Outlook References
Since the first measurements of the elec- 1. G. Binnig, C. F. Quate, C. Gerber, Phys.
trostatic double-layer force with the AFM Rev. Lett. 1986, 56, 930–933.
2. A. L. Weisenhorn, P. K. Hansma, T. R.
not even 10 years ago, the instrument has Albrecht et al., Appl. Phys. Lett. 1989, 54,
become a versatile tool to measure sur- 2651–2653.
face forces in aqueous electrolyte. Force 3. E. Meyer, H. Heinzelmann, P. Grütter
measurements with the AFM confirmed et al., Thin Solid Films 1989, 181, 527–544.
that with continuum theory based on the 4. A. L. Weisenhorn, P. Maivald, H.-J. Butt
et al., Phys. Rev. B 1992, 45, 11 226–11 232.
Poisson–Boltzmann equation and applied 5. J. H. Hoh, J. P. Revel, P. K. Hansma, Nan-
by Debye, Hückel, Gouy, and Chapman, otechnology 1991, 2, 119–122.
the electrostatic double layer can be ade- 6. S. Manne, H. E. Gaub, Curr. Opin. Colloid
quately described for distances larger than Interface Sci. 1997, 2, 145–152.
7. B. Cappella, G. Dietler, Surf. Sci. Rep. 1999,
1 to 5 nm. It is valid for all materials in-
34, 1–104.
vestigated so far without exception. It also 8. H.-J. Butt, M. Jaschke, W. Ducker, Bioelec-
holds for deformable interfaces such as trochem. Bioenerg. 1995, 38, 191–201.
the air–water interface and the interface 9. W. A. Ducker, T. J. Senden, R. M. Pashley,
between two immiscible liquids. Even the Nature 1991, 353, 239–241.
10. H.-J. Butt, Biophys. J. 1991, 60, 1438–1444.
behavior at high surface potentials seems
11. H. Freundlich, Z. Phys. Chem. 1910, 73,
to be described by continuum theory, al- 385–423.
though some questions still have to be 12. B. Derjaguin, L. Landau, Acta Physicochim.
clarified. For close distances, often the URSS 1941, 14, 633–662.
hydration force between hydrophilic sur- 13. E. J. W. Verwey, J. T. G. Overbeek, Theory of
the Stability of Lyophobic Colloids, Elsevier
faces influences the interaction. Between Publishing, New York, 1948.
hydrophobic surfaces with contact angles 14. G. Gouy, J. Phys. (Paris) 1910, 9, 457–468.
above 80◦ , often the hydrophobic attraction 15. D. L. Chapman, Philos. Mag. (London) 1913,
dominates the total force. 25, 475–481.
16. P. Debye, E. Hückel, Phys. Z. 1923, 24,
In the future, more fundamental investi-
185–206.
gations are probably going to focus on the 17. J. Mahanty, B. W. Ninham, Dispersion
interaction at close distances where only Forces, Academic Press, New York, 1976.
few molecular layers are present between 18. J. N. Israelachvili, Intermolecular and Sur-
the interacting surfaces. Then, however, face Forces, 2nd ed., Academic Press, Lon-
don, 1992.
the separation into electrostatic, van der
19. R. J. Hunter, Foundations of Colloid Science
Waals, and hydration forces might not be I+II, Clarendon Press, Oxford, 1995.
useful anymore. Also the investigation of 20. I. E. Dzyaloshinskii, E. M. Lifshitz, L. P.
shear forces (friction) or the change of Pitaevskii, Adv. Phys. 1961, 10, 165–209.
interaction under shear is a promising 21. P. M. Claesson, T. Ederth, V. Bergeron
et al., Adv. Colloid Interface Sci. 1996, 67,
area of research and many important fun- 119–183.
damental and applied questions can be 22. N. V. Churaev, Adv. Colloid Interface Sci.
addressed. Apart from these fundamental 1999, 83, 19–32.
2.5 Analyzing Electric Double Layers with the Atomic Force Microscope 249
179. A. C. Hillier, S. Kim, A. J. Bard, J. Phys. 193. B. A. Snyder, D. E. Aston, J. C. Berg, Lang-
Chem. 1996, 100, 18 808–18 817. muir 1997, 13, 590–593.
180. S. J. Miklavcic, R. G. Horn, D. J. Bachmann, 194. G. E. Yakubov, O. I. Vinogradova, H.-J.
J. Phys. Chem. 1995, 99, 16 357–16 364. Butt, J. Adhes. 2000, 14, 1783–1799.
181. D. J. Bachmann, S. J. Miklavcic, Langmuir 195. P. G. Hartley, F. Grieser, P. Mulvaney et al.,
1996, 12, 4197–4204. Langmuir 1999, 15, 7282–7289.
182. W. A. Ducker, Z. Xu, J. N. Israelachvili, 196. T. J. Senden, C. J. Drummond, P. Kékicheff,
Langmuir 1994, 10, 3279–3289. Langmuir 1994, 358–362.
183. M. L. Fielden, R. A. Hayes, J. Ralston, Lang- 197. S. Manne, J. P. Cleveland, H. E. Gaub et al.,
muir 1996, 12, 3721–3727. Langmuir 1994, 10, 4409–4413.
184. M. Preuss, H.-J. Butt, Int. J. Miner. Process 198. S. Manne, H. E. Gaub, Science 1995, 270,
1999, 56, 99–115. 1480–1482.
185. M. Preuss, H.-J. Butt, J. Colloid Interface Sci. 199. M. Jaschke, H.-J. Butt, H. E. Gaub et al.,
1998, 208, 468–477. Langmuir 1997, 13, 1381–1384.
186. S. Ecke, M. Preuss, H.-J. Butt, J. Adhes. Sci. 200. W. A. Ducker, L. M. Grant, J. Phys. Chem.
Technol. 1999, 13, 1181–1191. 1996, 100, 11 507–11 511.
187. S. Usui, H. Sasaki, H. Matsukawa, J. Colloid 201. I. Burgess, C. A. Jeffrey, X. Cai et al., Lang-
Interface Sci. 1981, 81, 80–84. muir 1999, 15, 2607–2616.
188. R. H. Yoon, J. L. Yordan, J. Colloid Interface 202. S. Manne, T. E. Schäffer, Q. Huo et al.,
Sci. 1986, 113, 430–438. Langmuir 1997, 13, 6382–6387.
189. P. Saulnier, J. Lachaise, G. Morel et al., J. 203. W. F. Heinz, J. H. Hoh, Biophys. J. 1999, 76,
Colloid Interface Sci. 1996, 182, 395–399. 528–538.
190. H. J. Schulze, C. Cichos, Z. Phys. Chem. 204. C. Rotsch, M. Radmacher, Langmuir 1997,
(Leipzig) 1972, 251, 252–268. 13, 2825–2832.
191. L. G. T. Eriksson, P. M. Claesson, J. C. Erik- 205. T. Miyatani, M. Horii, A. Rosa et al., Appl.
sson et al., J. Colloid Interface Sci. 1996, 181, Phys. Lett. 1997, 71, 2632–2634.
476–489.
192. P. Mulvaney, J. M. Perera, S. Biggs et al., J.
Colloid Interface Sci. 1996, 183, 614–616.
2.6 Comparison Between Liquid State and Solid State Electrochemistry 253
driving forces are similar. The reactions are then chemically inert, serving only
can be driven by both chemical potential for the supply and removal of electrons.
differences and electrochemical potential Furthermore, the LE tolerates the addition
differences. A recent book on solid-state of other materials that contribute mobile
reaction is by Schmalzried [13]. Properties ions which, however, do not participate in
of many relevant ionic materials as well the electrical current under steady state
as an excellent summary of the theory of conditions. The main use of this feature
point defect in solids can be found in the is in adding a supporting electrolyte, the
book by Kröger [14]. An outstanding con- importance of which will be discussed in
tribution to the field up to the year 1977 the theoretical Sect. 4.
was made by Wagner, whose results are In other arrangements, the electrodes
quoted in the reviews and books written provide material or serve as a sink.
later and mentioned in the preceding text. An electrode that serves as a source of
material is, for example, a metal anode that
2.6.2 dissolves and liberates cations into the LE.
Electrochemical Cell Geometry A metal electrode serves as a sink when it
is oxidized to form, for instance, an oxide
2.6.2.1 Cells Based on LEs layer or when electroplating occurs. This
An EC based on a LE is shown schemati- is also the case when insertion takes place
cally in Fig.1. It is composed of a liquid that as hydrogen into a palladium electrode.
conducts ions and two current-carrying Material can be exchanged also with a gas
electrodes, an anode and a cathode (E1 , phase. This is the case when bubbling
E2 ). An electrode can be composed of more hydrogen on a Pt electrode.
than one piece of one kind of metal as A key limitation of the existing LE ECs
when using a sacrificial anode. Additional is that the operation temperature must
electrodes (Eref ) that do not carry currents be kept low, compatible with the vapor
are added in most cases to serve as refer- pressure of the liquid. (An exception is the
ence electrodes in voltage measurements use of a few high-temperature, molten salt
and analysis of cell operation. LEs, for example, molten carbonates) [1,
The LE serves not only as a medium for 15, 16]. While low-operating temperature
selective transport of ionic charge but also, has obvious advantages, it may also
in many cases, as the source and sink for introduce limitations because of two very
the reactants and products involved in the different reasons. (1) high temperatures
electrochemical reaction. The electrodes allow to overcome electrode overpotentials,
E1 E ref E2
Fig. 1 Schematics of a
liquid-state electrochemical cell.
LE – liquid electrolyte; E1 ,
E2 – current-carrying electrodes;
Eref – reference electrode.
2.6 Comparison Between Liquid State and Solid State Electrochemistry 255
and (2) the heat generated as a by-product and the electrodes, E1 , E2 , being solids.
of the electrochemical reaction in hot These electrodes serve both as source and
cells can be used for heating much sink for electrons and for material. An ex-
more efficiently. This is the case in high- ample is the battery Lix C|SE(Li+ )|Liy TiS2 ,
temperature solid-oxide fuel cells in which where the anode is graphite intercalated
the hot exhaust gases can be used for with Li, and the cathode is TiS2 , a lay-
cogeneration of electricity by gas turbine ered compound that allows intercalation
driven electrical generators. A second key of Li into it, serving both as a conduc-
limitation of LE-based ECs is that they tor and a source/sink of Li+ ions and
cannot be used in thin-film devices and electrons. The SE conducts Li+ ions, and
miniature cells in microelectronics, while can be, for example, a glass made of
solid-state cells are being developed for Li2 O−Li2 SO4 −B2 O3 [8]. In this case, the
these applications. electrodes materials serve for the supply
and removal of both charge and matter.
2.6.2.2 Cells Based on SEs They must allow the diffusion (intercala-
Solid electrochemical cells are more re- tion) of the material into them and they
stricted in the structure because of the solid must also be electronic (electron/hole)
nature of the ionic conductor. Schematic conductors. The SE serves solely as a
configurations of solid-state cells are pre- membrane that facilitates the selective con-
sented in Fig. 2(a–c). Figure 2(a) shows duction of certain ions (e.g. Li+ ), but not
an all-solid cell with the SE, SE(M+ ), that of electron/holes and of other ions
E1 SE(M +) E2
(a)
P(X2)low P(X2)high
(−) SE(X−) (+)
TPB
(b)
(c)
MIEC anode MIEC cathode
P(X2)low P(X2)high
(−) (+)
SE(X−)
(d)
Fig. 2 Schematics of solid-state electrochemical cells. (a) Cell with two, solid, current-carrying
electrodes E1 and E2 , SE(M+ ) – solid electrolyte conducting the cation M+ . (b) Cell with
electrodes composed of a porous metal layer for electronic charge exchange and the gas phase,
X2 , for material exchange. PX2 –partial pressure of the gas X2 . SE(X− ) – solid electrolyte
conducting the anion X− . (c) The triple phase boundary (TPB) enlarged. (d) Cell with
mixed – ionic – electronic conducting electrodes for supply of electronic charge permitting
diffusion of ions through them, enabling material exchange with the gas phase.
256 2 Electrochemical Double Layers
that may come from the electrodes. SEs the rates of the different and competing
usually do not tolerate the addition of a elementary steps in the electrode reaction.
second type of mobile ions to serve as The geometry of this cell is again different
supporting electrolytes. All this is quite from that of a LE-based cell.
different from the case of LSE. There are A modification of the cell shown in
additional differences. Ions have to cross Fig. 2(b) is presented in Fig. 2(d) in
the SE/electrode interface. The mobility of which nonporous MIECs are used as
ions in SEs is, for most cases, lower than electrodes. In a fuel cell, oxygen has then
in LE, and is realized usually only at el- to diffuse from the gas phase through
evated temperatures. Because of the low the MIEC cathode to reach the SE and,
mobility, SEs contain high concentrations correspondingly, the fuel and the exhaust
(>1 atomic%) of mobile ions. The highest gas have to diffuse through the anode. (In
conductivity obtained at elevated T reaches reality, the MIEC electrodes are porous,
∼1S/cm for some materials (e.g. for Ag+ which increases the electrode active area).
in AgI at T > 149 ◦ C and for O2− in Bi2 O3 The MIEC can be either a semiconductor
at ∼800 ◦ C [11]). However, for most SEs or a metal. An example of a continuous
the conductivity is at least one order of cathode is: YSZ|Ag|air, with the Ag layer
magnitude lower. as the MIEC electrode. This is possible
Figure 2(b) exhibits a solid electrochem- because silver is permeable to oxygen at
ical cell with porous, inert electrodes in elevated temperature [17, 18].
contact with a gas phase at each elec-
trode that interacts with the electrode (by 2.6.2.3 Comparing the Two Groups of
electron charge transfer) and with the Cells
SE. An example is the fuel cell arrange- In LSE salt bridges are used to connect elec-
ment, H2 ,H2 O,Pt|YSZ|Pt,O2 . The SE yt- trically (by selective ion motion) but to sep-
tria stabilized zirconia (YSZ) (the oxide: arate physically and to a good approxima-
(Y2 O3 )x (ZrO2 )1−x , x ∼ 0.1) is an oxygen- tion also chemically, two compartments.
ion conductor. One electrode is exposed Owing to ion leaks, these are useful only
to a high concentration of oxygen (pure for low-current measurements, in partic-
oxygen or air) and the other to a fuel ular, open circuit voltage measurements.
such as H2 . (In modern fuel cells, Pt for In SSE, salt bridges are not needed. SEs
the electrodes is replaced by other metals, usually conduct only one kind of ions. Let
mixed ionic electronic conductor (MIECs) us consider the Daniel cell Cu|LE(Cu2+ ,
or cermets [9].) In the cell mentioned here, SO4 2− )|Bridge|LE(Zn2+ ,SO4 2− )|Zn. The
each electrode is a mixture of phases, one bridge is required to stop mixing of the
(Pt) supplying the electronic charge and two solutions. It allows a slow passage of
the second one (the gas phase) serving as ions, thus facilitating a continuous electri-
the source or sink of material. The active cal circuit. If the bride would be absent,
electrode area is around the triple phase this would destroy the cell as mixing
boundary (TPB) in which the three phases: would occur followed by electroplating of
gas, metal, and SE involved in the electrode Cu onto the Zn electrode. On the other
reaction, meet (see Fig. 2c). The length of hand, in SSE, if one would build a Cu,
the TPB is the length of the pore edges in Zn cell, one would use a SE that con-
the porous metal. The width of this area ducts only one kind of ions, say Cl− .
varies from case to case and depends on This could be SrCl2 [19]. The cell would
2.6 Comparison Between Liquid State and Solid State Electrochemistry 257
distance of closest approach due to the the disordered system is very large and
cloud of solvating molecules. We can there- the electron states highly localized with
fore safely consider LEs as ionic conduc- negligible mobility.
tors with negligible electronic conduction. The ions and electrons in SEs and in
The motion of ions in a LE is impeded. MIECs polarize the ionic neighborhood.
The ion or the ion and its solvating sphere Therefore, the ions, electrons, and the
move in a viscous liquid. The force of polarized surrounding form the corre-
retardation can be calculated according sponding small polarons [22]. This would
to Stokes equation and the corresponding be analogous to polarization of an ionic
diffusion coefficient or mobility is related liquid. However, the details are different
to the viscosity. and in the sphere of solvating molecules
Protons can propagate by two mecha- that exists in LSE, neutral molecules with
nisms. One is by the viscous flow of a an electric dipole are attracted to the
complex H+ (H2 O)n where n = 1, 2, 3, 4, ion and also change orientation. For SEs
the second is by the Grotthuss mechanism, with a high concentration of mobile ions
hopping of a proton from one hydronium (>10 ion%), for example, Ag+ ion in AgI
H3 O+ to a nearby water molecule. The at T > 149 ◦ C, the interaction between the
latter mechanism has some similarity to mobile ions forces local relaxation within
hopping conduction of ions in solids as the sea of these ions, which affects the con-
discussed in the following text. ductivity and its dependence on frequency
for ac signals [23].
2.6.3.2 Charge Transport in SEs The high concentration of ions both in
SEs are those solid ionic conductors LEs and in SEs affect the width of the
for which ionic conductivity, σion , is space charge that exists at the boundary
significant and the electronic conduction, of the electrolyte near the contact interface
σel , is small. There is no exact definition of with the electrode. Any abrupt change in
what is a significant ionic conductivity, but the concentration of charged species must
usually σion > 10−4 S/cm is considered result in a space charge [24, 25]. The width
significant with the lower limit applying of that space charge is roughly the Debye
to cells with thin-layer SEs. Electronic length and for typical LEs, it is only 0.3
conduction in solids is, usually, present to 10 nm; while for typical SEs, it is 0.3
to a higher degree than in LEs. SEs to 1 nm. The detailed characteristics of the
are those MIECs for which σel σion . space charge are not identical for LEs and
The reasons for low electronic conduction SEs. One difference arises from the fact
in ionic solids are: (1) The energy gap that in SEs usually only one type of ions
is large and intrinsic electron and hole is mobile. There is therefore a difference
concentrations are very low. (2) The between enriching and depleting that ion,
occupied electron states in the conduction and the profile of the space charge when
band and hole states in the valence it is enriched or depleted is different [12].
band are localized rather than delocalized. The other differences are due to the or-
Therefore, the electron and hole mobilities dered structure of the SE and the absence
are thermally activated and low. The of a solvating cloud in SEs, as discussed in
lack of electron conduction in LE can Sect. 6.
be discussed in the same terms. In In SEs, ionic and electronic charge
these terms, the quasi–energy gap of transport is associated with deviations
2.6 Comparison Between Liquid State and Solid State Electrochemistry 259
from the normal state. This can be three ways, similar to those for electronic
best understood by considering crystalline defects:
solids. Then conduction is associated with
1. By thermal excitation. The result, in
point defects [14]. The perfect crystal is
an elementary solid, is a vacancy
defined as that with perfect lattice pe-
and an interstitial. In a binary-ionic
riodicity, at zero temperature. In semi-
compounds, say MX (a compound
conductors, electrons in the conduction
composed of cations M and anions X),
band are thus defects formed by ex-
thermal excitation may result in one
citation into states that are empty at
of five combinations of a vacancy, an
T = 0. Conduction in LEs can also be
interstitial and/or a misplaced ion [14].
formulated as a deviation from a stan-
The most widely discussed defects are
dard liquid state [26]. However, this is
Frenkel pairs, an interstitial of, for
uncommon and is not discussed fur-
instance, a cation and the vacancy
ther here.
left behind, Mi + VM , and Schottky
The electronic charge carriers can
pairs of vacancies of the cation M
be generated, under equilibrium, in
and anion X: VM + VX . We use here,
three ways:
the Kröger–Vink notation of point
1. By thermal excitation across the band defects. Charge is measured relative
gap. to the perfect crystal. A dot denotes
2. By extrinsic doping with donors or a positive relative charge, a prime
acceptors. denotes a negative relative charge and
3. By change of stoichiometry. The ionic x denotes a zero relative charge [14].
defects formed in this way act as native Thus, the Frenkel pairs in AgBr,
donors or acceptors. interstitial silver ions and silver ion
vacancies, are, in the Kröger–Vink
Of course, any combination of these three notation: Agi • + VAg \ .
causes may be realized simultaneously in 2. By doping. Charged ionic defects can be
any given material. In addition to elec- generated also by doping with charged
trons/holes generated under thermal equi- impurities. For substitutional doping,
librium, they may also be injected from they must be aliovalent. For example,
electrodes and excited by optical pump- doping substitutionally ZrO2 with CaO
ing under nonequilibrium conditions. As forming the defect CaZr \\ generates
mentioned before, the mobility of the elec- also ionic defects in the form of charged
trons and holes is thermally activated as oxygen vacancies, VO •• . The later are
they form small polarons. mobile at elevated temperatures. We
For ionic conduction to exist in a crystal, have mentioned before that doping may
an ion has to move out of its normal po- result also in generating electrons or
sition either into an interstitial site or find holes. Whether doping will result in
an existing defect in the form of a vacancy. ionic defects (called self compensation)
The conductivity increases therefore lin- or electronic ones depends on the
early with the defects concentration, in the materials involved, and can be predicted
dilute limit (as long as interaction between if the nature of the point defects and
the defect can be neglected). The defects their concentrations in the host are
can be generated, under equilibrium, in known in detail [27].
260 2 Electrochemical Double Layers
n = 2[VO •• ]. In this case, the electron con- σion ), whenever ambipolar motion of
ductivity is higher than the ionic one due different charge carriers occurs, as is the
to a significantly higher electron mobility. case in chemical diffusion and under
In amorphous condensed phases, the conditions in which the driving force
interpretation of the conduction is rather for ionic motion is small. The existence
similar to that in crystalline materials if one of electronic conduction in addition to
views the amorphous phase as a crystalline an ionic one yields a very important
solid that is highly disordered. In this view, class of materials: MIECs.
the solid contains a high concentration of 3. The concentration of ionic defects in
defects with at least one kind being mobile. SEs must be high compared to that in
This explains why glasses exhibit, in many LEs in order to obtain comparable ionic
cases, a higher ionic conductivity than the conductivity because the mobility of the
corresponding crystalline solid with the ions in SEs is usually lower than in
same composition. LEs. Even so, in most cases, the SEs
Polymer ion and electron conductors are will exhibit a high conductivity only at
of much current interest in SSE because elevated temperatures.
of their ease of production, low cost and 4. The path for ion motion and the temper-
relative high conductivity at room tem- ature dependence of ionic conductivity
perature. The conduction mechanism for in SEs and LEs are very different.
ions is different from that in inorganic
solids and is based either on segmental 2.6.4
motion in the polymer and hopping of the Theoretical Current Voltage Relations for
ion (in a remote similarity to the Grot- Ideal ECs (No Electrode Polarization)
thuss mechanism for proton conduction
in water) or on a LE embedded in the poly- 2.6.4.1 General Current Density Equations
mer. Electron conduction in polymers can The current density equation for the ions
be discussed in terms used for conduction in the bulk of the LE or SE are the same:
in inorganic semiconductors. Single-phase σion
ji = − ∇ µ̃ion (1)
MIECs polymer with significant conduc- zion q
tivities do not exist yet. Quasi-MIECs are
formed by blending two polymers, one where µ̃ion is the electrochemical poten-
conducting ions and the other conducting tial of the ions, zion the charge of the ion
electrons/holes [34]. and q is the elementary charge of a pro-
ton. For SEs the charge considered is the
2.6.3.3 Summarizing the Differences
effective one in the Kröger–Vink notation
Between Transport in LEs and SEs (though absolute charges could also be dis-
The key differences between conduction cussed [35]). The convention used here is
in LEs and in SEs are the one used in SSE referring the various
quantities to a single particle rather than
1. There is usually only one kind of mobile to a mole as done in LSE. Thus, the chem-
ions in SEs, as compared to at least two ical potential is per particle, the charge is
in LEs. counted in elementary charges, q, rather
2. There is an electronic conductivity, σel , than in Faraday units and one uses the
present in SEs that has to be taken Boltzmann constant rather than the gas
into consideration, even if small (σel constant.
262 2 Electrochemical Double Layers
For LEs one has to write at least reaction, and we shall use this approach
two current-density equations, one for here also. Reference electrodes are also
each mobile ionic specie. If a supporting used in SSE, but they are not universal
electrolyte is added to the LE, two more standard ones, but rather chosen ad hoc
current-density equations have to be added for each particular system.
for the additional ions. The gradient in the
latter two equations vanishes under steady 2.6.4.3 Supporting Electrolyte
state conditions because the electrodes are Let us first examine the effect of a
blocking for the ions of the supporting supporting electrolyte in a LEs and its
electrolyte. On the other hand, for SEs, absence in SEs. The supporting electrolyte
one has to consider the current density of does not participate in the electrochemical
electrons and/or holes. For electrons: reaction, by definition. Thus, jion = 0
σe (where j is the current density) for the
je = ∇ µ̃e (2)
q ions of the supporting electrolyte. Hence,
by Eq. (1)
where µ̃e is the electrochemical potential
of the electrons. µ̃e is equal to the redox ∇ µ̃ion = 0, jion = 0 (3)
potential of a redox couple, and it is equal
to the Fermi level in solids. An electrochemical potential, µ̃, can be
Quasi-coupling between the ionic and expressed in terms of a chemical potential,
electronic currents is sometimes intro- µ, and the inner (Galvani) electrical
duced by adding a term Lie ∇ µ̃e to Eq. (1) potential ϕ,
and a term Lei ∇ µ̃i to Eq. (2). However, it
µ̃ = µ + zqϕ (4)
was shown that these corrections are not
really necessary (except perhaps in met- where zq is the charge of the relevant
als [33]) if the current density of all mobile specie. In the limit of relative high con-
species, as well as interactions between de- centrations of the supporting electrolyte,
fects are all taken into consideration [36]. gradients in the chemical potential of its
ions can be neglected even when a voltage
2.6.4.2 Half Cells is applied to the electrodes, and/or there
In LSE it is common to discuss an EC as exist a difference in the composition of the
if separated into halves, and to consider electrodes,
the potential at each electrode with respect ∇µion = 0 (5)
to a third standard reference electrode.
Combining Eqs. (3–5) yields:
Furthermore, many of the ECs discussed
in LSE contain supporting electrolytes. ∇ϕ = 0 (6)
This is quite different from SSE, in which
supporting electrolytes are absent. Thus, that is, the Galvani potential is uniform,
in LEs the voltage as well as the Nernst the inner electric field vanishes in the LE
theoretical voltage can be expressed in due to the supporting electrolyte. In this
terms of ion concentration in the LE. case, the other (minority) ions flow under
In SSE the analysis is done in terms of a chemical potential gradient only.
the chemical potentials of the materials In SSE, on the other hand, the mobile
and electrochemical potentials of the ions ions are those participating in the elec-
and electrons/holes involved in the cell trochemical reaction. The concentrations
2.6 Comparison Between Liquid State and Solid State Electrochemistry 263
can be fixed by CO2 /Na2 CO3 /O2 under a with fixed ionic charges and with negligible
PO2 value for which CO2 is stable [38]. µX electronic conductivity.
may also be fixed by an auxiliary EC [33]. The way the Nernst voltage is calculated
The open-circuit voltage is the voltage in LSE is by considering the electrode
when the total current through the EC reaction under open-circuit condition,
vanishes. In SSE the total current, It , rather as in SSE from µX (Eq. 9).
may contain a nonnegligible electronic Let us consider, for example, the cell,
contribution, Iel and Ag|AgCl|LE(H2 O,H+ ,Cl− )|Pt,H2 , and cal-
culate the Nernst voltage using first the
It = Iion + Iel (15) LSE approach and then the SSE approach.
The voltage of the cell is, according to
Under open circuit conditions, It = 0 and
Eq. (8),
the partial currents Iion = −Iel need not
vanish. Therefore, the voltage for It = 0 −qV = µ̃e (Pt) − µ̃e (Ag) (17)
is different from the voltage for Iion = 0.
It can be shown that when a single ionic µ̃e (Ag) is fixed by the electrode reac-
species is mobile then [37], tion: Ag + Cl− ↔ AgCl + e− , which un-
der equilibrium yields,
Voc < V (jion = 0) = Vth ,
µ̃e (Ag) = µ0 (Ag) + µ̃(Cl− )Ag,AgCl
Voc −−−→ V (jion = 0) only when
− µ0 (AgCl) (18)
Iel −−−→ 0. (16a)
where µ̃(Cl− )Ag,AgCl is the value of the
Eq. (16a) holds whether the electron/hole
electrochemical potential of Cl− near the
transport is via the conduction/valence
interface Ag/AgCl and µ0 denotes the
band or via defect band hopping.
standard chemical potential at the given
When variable charge, mobile ions exist,
temperature, T , which for solids is equal to
for example, Cu+ and Cu++ , then there
the chemical potential. The measurement
are different Nernst voltages, one for each
of the open-circuit voltage should be done
zion value (see Eq. 9). We neglect now any
under steady state. Then the Cl− current
electronic current. Then, the open-circuit
must vanish in the two, purely ionic
voltage is not equal to either of the Nernst
conductors AgCl and LE(H2 O,H+ ,Cl− )
voltages. For the example with zion = 1
placed in series. Thus, µ̃(Cl− ) is uniform
and 2 [33, 39],
throughout the cell and we may remove
Vth (zion = 2) < Voc < Vth (zion = 1) the subscript Ag/AgCl and we may add
the subscript LE.
(16b)
µ̃e (Pt) is fixed by the electrode reaction
In LSE the Nernst voltage is, in princi-
H2 ↔ 2H+ + 2e− , which, under equilib-
ple, defined similarly (with respect to a
rium, yields,
standard-reference electrode) and called
the equilibrium potential. However, in µ̃e (Pt) = 12 µ(H2 ) − µ̃(H+ )Pt,LE (19)
practice one usually refers to the open-
circuit voltage of a half cell with respect where µ̃(H+ )Pt,LE is the electrochemical
to a standard electrode. The open-circuit potential of the protons near the H2 ,Pt/LE
voltage is equal to the Nernst voltage when interface. µ̃(H+ ) turns out to be uniform
it is measured on purely ionic conductors, in the LE, since both µ̃(Cl− ) and µ(HCl)
2.6 Comparison Between Liquid State and Solid State Electrochemistry 265
are uniform in the LE, thus the subscript µ(Cl2 )Pt,LE , the chemical potential of Cl2
Pt,LE can be replaced by LE. Combining at the LE/Pt,H2 interface, is fixed by the
Eqs. (17), (18) and (19) yields, reaction of H2 and HCl in the LE: H2 +
Cl2 ↔ 2HCl ↔ 2H+ + 2Cl− . Under equi-
−qVoc = 12 µ(H2 ) − µ0 (Ag) + µ0 (AgCl)
librium,
− µ̃(H+ )LE − µ̃(Cl− )LE (20)
1
2 µ(Cl2 )Pt,LE = µ(H+ )Pt,LE + µ(Cl− )Pt,LE
Under equilibrium one can make the
following substitution: − 12 µ(H2 ) (26)
µ̃(H+ )LE + µ̃(Cl− )LE = µ(HCl)LE Combining Eqs. (25) and (26) and noticing
= µ(H+ )LE + µ(Cl− )LE (21) that µ(H+ ) and µ(Cl− ) are uniform in the
LE, as mentioned before, yields,
in Eq. (20), then,
−qVoc = 12 µ(H2 ) − µ0 (Ag) + µ0 (AgCl) −qVth = 12 µ(H2 ) − µ0 (Ag) + µ0 (AgCl)
− µ(H+ )LE − µ(Cl− )LE (22) − µ(H+ )LE − µ(Cl− )LE (27)
or when expressed in terms of activities, Vth equals Voc of Eq. (22) since the elec-
tronic component in the combined elec-
−qVoc = 12 µ0 (H2 ) − µ0 (Ag) + µ0 (AgCl)
trolyte AgCl/LE vanishes (see Eq. (16a)).
− µ0 (H+ )LE − µ0 (Cl− )LE This demonstrates that the two definitions
a(H2 )1/2 of the Nernst voltage are indeed equal
+ kB T ln (23) when the membrane is a pure ionic con-
a(H+ )LE a(Cl− ) LE
ductor. This would not be so if electronic
For low ion concentrations a(H+ )LE and conduction would be present (as can be the
a(Cl− )LE may be replaced by the corre- case in SSE), as then Vth = Voc (Eq. 16a).
sponding (uniform) concentration [H+ ] Different values for the Nernst voltage
and [Cl− ] in the LE. defined as Voc rather than Vth would be
The argumentation in the SSE approach obtained also when the LE contains vari-
would be as follows: There is a selective able charge-mobile ions, for example, Cu+
membrane for Cl− ions, that is, an and Cu++ , as then Vth = Voc as can be
electrolyte, made of two parts connected in seen in Eq. (16b).
series: AgCl and the LE. The Nernst voltage
is determined by µ(Cl2 ) according to 2.6.4.6 Chemical Diffusion Coefficient
Eq. (9), Chemical diffusion of a single component
2qVth = µ(Cl2 )Pt,LE − µ(Cl2 )Ag,AgCl occurs, in ionic solids, by the motion of
(24) ions. This must be accompanied by an
µ(Cl2 )Ag,AgCl , the chemical potential of Cl2 equal current of electronic charge to avoid
at the Ag/AgCl interface is governed by charge accumulation. Thus diffusion in
the reaction: 2Ag + Cl2 ↔ 2AgCl. Under ionic solids with a single type of mobile
equilibrium, ions is an ambipolar motion of ions
and electrons (or holes). This motion is
1
2 µ(Cl2 )Ag,AgCl = µ0 (AgCl) − µ0 (Ag) governed by Eqs. (1) and (2) and the
(25) condition It = 0. The chemical diffusion
266 2 Electrochemical Double Layers
the ions there. At the AgI/Ag2 S interface, The distribution ϕ(
r ) and the concentra-
Ag+ ions are transferred from the AgI tion of the charges that fix ϕ(
r ) therefore
lattice into the Ag2 S lattice. deserve an over view. In LSE the LE is
For porous gas electrodes, shown terminated abruptly at the contact with
schematically in Fig. 2(b), there can be the electrode (and at the upper side in
many different steps involved. For exam- contact with the atmosphere). Charge is
ple, the overall cathodic reaction on a Pt transferred, also under equilibrium, from
electrode on YSZ is: O2 (gas) + 4e− (Pt) ↔ the metal electrode into the LE (or vice
2O2− (YSZ). One can suggest that the over- versa) leaving behind the opposite charge
all reaction consists of the following series and thus forming a double layer. In the
of elementary steps: diffusion of O2 in the metal the excess or missing charge (in the
gas phase toward the TPB, O2 adsorption ‘‘sea’’ of electrons) concentrates near the
on the Pt, O2 dissociation, electron transfer M/LE interface, in a very thin layer, less
to Oad to form Oad 2− , diffusion of Oad 2− than a mono atomic–layer wide. In the LE
along the Pt electrode surface toward the the charge is spread out a little more over
SE (YSZ), and then entering of the ion into a length determined by the Debye length,
the YSZ. There can be other series of steps λD (which is the inverse of the Debye re-
that yield the same overall reaction [52]. ciprocal length, κ commonly used in LSE).
A different electrode is one that is not λD is about 0.3 to 10 nm wide for most
porous but a continuous MIEC and the LEs due to the high concentration of the
source or sink for material is not the MIEC mobile ions.
but the gas phase, as shown schematically In the LE the local dielectric constant at
in Fig. 2(d), Then diffusion of material in the interface may be significantly modified
the form of ions through the MIEC, as well as a result of orientation of the water-
as charge transfer of ions at the MIEC/SE molecule dipole in the electric field of the
double layer. Furthermore, ions can flow
interface has also to be considered.
to the interface and be adsorbed there, in
It is evident from those examples
the double-layer region.
that electrode processes in SSE involve
The explanation for the charge transfer
elementary steps that may be limited by
on contact of the metal and the LE, un-
diffusion, others by slow-charge transfer
der equilibrium, depends on the approach
of electrons in an electronation or de-
taken. Of course, the different explana-
electronation step, and yet others by a slow
tions are equivalent. From the SSE point
transfer of ions across an interface. The
of view, the charge in the double layer
latter step is not common in LSE. On the
(for zero applied voltage) originates from
other hand convection exits in LSE but not
a difference in two properties between the
in SSE.
electrode material and the LE, the work
function and the chemical potential of the
2.6.6.2 Inner (Galvani) Electrical Potential mobile component. For inert metals it is
only the difference in the electron work
2.6.6.2.1 Inner Electrical Potential in LSE functions that matters. Electrons are trans-
The analysis of elementary steps that in- ferred until the electrochemical potential
volve charge transfer, whether of electrons of the electrons in the metal and the nearby
or of ions, require a detailed discussion of LE is the same. Then the potential of zero
the inner (Galvani) electrical potential, ϕ. charge, used in LSE, equals the difference
270 2 Electrochemical Double Layers
in the work functions (divided by the el- translational symmetry at the surface. Elec-
ementary charge, q). When the electrode trons or holes trapped at these states cause
can be dissolved (e.g. a Zn electrode) then a double layer between the trapped charges
material and electrons are transferred, un- and the opposite charge left behind in the
til the chemical potential of the material bulk of the semiconductor and distributed
as well as the electrochemical potentials of nearby [53]. Due to quantum effects, elec-
the corresponding ions and of the electrons trons can be found outside the free surface.
all are uniform across the interface. This also contributes to a double layer.
At the free surface of the LE facing In SEs and MIECs of interest the
the gas phase a double layer can also be contributions to surface charges come
formed, in analogy to the free surface of a mainly from mobile ions that exist in
solid, discussed in the following text. Our large concentrations rather than from
main concern is the interface between the electrons or holes, when they are the
LE and the metal electrons. minority charge carriers. These ions can
accumulate or be depleted near the surface
2.6.6.2.2 Inner Electrical Potential in SSE forming a charged layer there (the ‘‘core’’
In SSE the situation is somewhat more in Fig. 3a) [54, 55]. The position of the
complex because of the rigid structure of ‘‘core’’ may vary to be in between the first
the solid. The Galvani potential distribu- and second atomic layer, within the first
tion is shown in Fig. 3. Let us start with atomic layer or on the first atomic layer.
the free surface of a solid. A semicon- The opposite ionic charge left behind in
ductor may have surface states for the the bulk is distributed nearby. For a core
electrons because of the breaking of the centered at xc (in Fig. 3), the opposite
ϕ ϕ
ϕ
X X X
+ Metal
+ Metal +
+ + +
+
+ − + − −
+
+ − − + + − − − −
+
+ − − − + + − − − + − − −
0 Xc X1 X 0 Xc X1 X 0 X1 X
(a) (b) (c)
Fig. 3 Distribution of charge and Galvani potential, ϕ, at a solid interface. xc
- the position of the core. In (b) and (c) the metal is left of x = 0. (a) Free
surface; (b) metal/MIEC interface that contain a core and (c) metal/MIEC
interface without a core.
2.6 Comparison Between Liquid State and Solid State Electrochemistry 271
Potential energy
X X X
(a) (b) (c)
Potential barriers of type (1) may have a thermal activation but not necessarily
contribution from the electrical potential the same energy. Under equilibrium
energy as shown in Figs. 5(b) and (c). This the two transfers (which depend on
then leads to expressions of the form of the barrier heights, local concentrations
Eq. (33a) with 0 < β < 1. The parameter of reactants and products and reaction
β is associated with the symmetry of the constants) exhibit the same rate. The
potential maximum, therefore β is denoted electrical potential contribution is, under
as the symmetry factor. equilibrium, ϕeq (Fig. 5b). Under current
Let us first see why a potential barrier a deviation δ(ϕ) from the equilibrium
of the form of Fig. 4(a) arises in LSE. value ϕeq is generated. Part −βδ(ϕ)
For ion transfer between the solution modifies one side barrier height and a
and the metal electrode, say a solvated part (1-β)δ(ϕ) modifies the other side
proton, the position, sufficiently removed barrier height (Fig. 5c).
from the metal electrode, where it is fully
When the charge transferred is an
surrounded with water molecules, is of
electron, Eq. (33a) still holds. An example
relative low energy. The position in which
for such a reaction is: HLE + + eM − ↔
the proton is attached to the metallic elec-
Had,M , where: HLE + is a proton in the
trode with most of the water molecule
LE, eM − an electron in the metal electrode,
surrounding it removed, is also of relative
and Had,M an hydrogen atom adsorbed
low energy due to the interaction of the
proton with the metal. The positions in on the metal electrode. It is not obvious
between these two ones are of higher en- that this equation holds for an electron
ergy forming a maximum in the potential transfer as an electron can easily tunnel
energy. The barrier may have a contribu- through a potential barrier that is a few Å
tion of electrical potential energy, which wide and only ∼1 eV high, as is typically
is linear in position x. The subdivision the case at the Metal/LE interface [57, 58].
into different energy contributions looks However, the barrier that the electron
qualitatively then, as shown in Figs. 5(b) has to cross is not the only factor to be
and (c). considered. For tunneling to take place
The transfer of a proton from left to the initial and final states of the electron
right over the barrier in Figs. 4(b), 5(b) have to be degenerate, that is, have the
and (c) requires thermal activation and the same energy. Furthermore, the state the
transfer from right to left also requires electron is transferred to must be empty
2.6 Comparison Between Liquid State and Solid State Electrochemistry 273
Potential
~
µ
µ0
µ0
µ0 + zqϕ eq
zqϕeq
X X
(a) (b)
~
µ
µ0 + zqϕ
µ0
zqϕ µ0 + zqϕ eq
X
(c)
Fig. 5 Potential energies (µ0 , µ̃, zqϕ) in SSE. (a) Inside the SE (or
MIEC); (b) at the SE/electrode interface under equilibrium and (c)
at the SE/electrode interface under current.
to be able to receive the electron. It turns elementary steps some transferring charge
out that the position of the proton and and some not. The products of one step are
hydrogen atom with the right energy for the reactants for the following step. When
electron tunneling lies at a relative high the charge transfer occurs across exactly
energy and the motion toward that position the same potential difference, ϕ then the
forms a potential barrier similar to the one I –V relations can be related through I0
shown in Figs. 5(b) and (c). This results in to the concentrations of the species that
Eq. (33a) also for electron transfer [56]. exist in the first and the last step in the
Charge transfer to a neutral species series. Taking one elementary step to be
that approaches the metal electrode from rate determining (rds) [56],
beyond the outer Helmholtz plane, for
I = I0 (eαa qδ(ϕ)/kB T − e−αc qδ(ϕ)/kB T )
example, O2 dissolved in the LE, may
also follow the Butler–Volmer Eq. (33a). (34a)
This is expected to be different from the where,
I –V relation for an electron transfer to O2 γc
αc = + rβ,
already adsorbed onto the metal (see in the ν
following text). nt − γc
αa = + rβ (34b)
ν
2.6.6.3.2 The Butler–Volmer Equation for where, nt is the total number of elementary
Multistep Reactions An over all process at charges transferred, γc is the number
an interface consists usually of a series of of elementary charges transferred before
274 2 Electrochemical Double Layers
the rds, r is the number of elementary by the average over a volume of atomic
charges transferred in each rds, ν is size. However, for charge transfer at an
the number of times the rds has to be interface the detailed distribution of the
repeated to complete one reaction and β energies on an atomic scale must be con-
is the symmetry factor of the rds. Thus, sidered, as these exhibit potential barriers
for example, r = 1 for an electron being that the ion has to overcome [66].
transferred; r = 2 for an O2− ion being To be specific let us consider the case of
transferred; and r = 0 if the rds is a dilute defect concentration, [i], so that the
chemical reaction without charge transfer. chemical potential is:
ε ε
Ep Ep
εa
εF εF
εa
Metal Metal
(a) (b)
Fig. 6 Tunneling of an electron through a potential energy
barrier into a metal. (a) Thermally assisted tunneling and (b)
tunneling not requiring thermal excitation.
276 2 Electrochemical Double Layers
where f (δ(ϕ)) yields the dependence of is true whether it is the free metal/gas
the tunneling current on the distortion interface or the metal/SE interface. Thus
in the potential barrier by δ(ϕ), pr is the tunneling distance for an electron is
the probability that the acceptor level is significantly smaller in SSE (except when
occupied by an electron and pr,eq is pr un- in LSE the electron transfer is also to a
der equilibrium. The relevant expression species directly adsorbed onto the metal
depends on the transfer direction being electrode. However, the concentration of
thermally activated relative to the polarity uncharged adsorbed acceptors is expected
of the applied voltage. Eq. (37a) also holds to be relatively low). The energy barrier
when the electron is transferred from the heights are expected to be similar in SSE
metal to the atom/ion. The temperature de- and LSE; however, ε, the energy required
pendence is given by the exponential term to achieve degeneracy is lower than the
exp(±qδ(ϕ)/kB T ). Neglecting the effect energy barrier height. We therefore expect
of δ(ϕ) on f and pr , Eq. (37a) reduces to, that electron tunneling at the electrode in
SSE is faster than in LSE.
I = I0 (1 − e−qδ(ϕ)/kB T ) or For comparing the transfer rate of ions
I = I0 (eqδ(ϕ)/kB T − 1) (37b) and electrons in SSE across a potential
barrier, let us consider an example of
Eq. (37a) is different from Eq. (33a) as one oxygen transfer from the gas phase into
term in the brackets does not depend on YSZ, first adsorbing onto a metal electrode
δ(ϕ). with electron transfer and diffusion to the
TPB occurring there:
2.6.6.4.4 Further Discussion of Ion and O2 −−−→ 2Oad −−−→ 2O− ad
Electron Transfer It is questionable if the
rate of an electron transfer in SSE, from −−−→ 2O2− ad −−−→ 2O2− ad,TPB
a metal to an atom/ion (or vice versa), −−−→ 2O2− (YSZ) (38)
can be slower than an ion transfer at
the electrode. The electron transfer is by We notice (see Fig. 6) that electron transfer
tunneling. For a given atom/ion as the in one direction does not require thermal
acceptor (or donor), four factors govern excitation at all and in the opposite
the tunneling rate. First, availability of direction it needs only to be excited by
the electrons (or of an empty site for ε, a fraction of the energy potential
the reverse transfer). Electrons are ample barrier Ep , to reach a degenerate state.
in a metal. Then comes the distance ε is expected to be also lower than the
for tunneling, the energy barrier height, potential barrier that ions have to cross,
and the energy ε required to achieve since the later should be similar to the
degeneracy. Let us first compare the total barrier height in Ep of the electrons.
electron tunneling process in SSE with The high concentration of electrons in
the one in LSE, in which the electron the metal, the short tunneling distance of
tunneling process is different and can be ∼0.1 nm, the barrier height of ∼1 eV, and
rate determining. In LSE the ion is solvated the relative low value of ε should result
and the cloud of solvating molecules keeps in a high value of the exchange current
the ion quite far from the metal electrode. I0 for electron transfer and I0 should be
In SSE, both the ion and the atom are higher than that for ion transfer. Thus if
adsorbed onto the metal electrode. This charge transfer is found experimentally to
2.6 Comparison Between Liquid State and Solid State Electrochemistry 277
be the rate-determining step in SSE then it and that a charge-transfer rds is thus due
is, most likely, an ion transfer that is rate- to ion transfer. Then there is a negligible
determining not an electron transfer [60]. difference µ̃e at the interface. An overpo-
The electron as well as ion-transfer tential is still measured because of the way
currents depend not only on the rate overpotentials, η, at the electrodes are mea-
per adsorbed atom/ion but also on the sured [59, 60, 62, 68]. η is not measured by
atom/ion coverage. If this is low it lowers inserting an inert metallic electrode into
both exchange currents. There can be two the SE and measuring η with respect to
reasons for low coverage: either another the working electrode, that is, at the two
elementary step, which governs the supply sides of the interface. Such a measure-
of the atom/ion, is the rate determining ment would yield η = −µ̃e /q̃. Rather,
step, or the equilibrium coverage is low. the measurement is done between a work-
The latter has been suggested for gold ing electrode and a reference electrode put
cathodes on YSZ for oxygen transfer [67]. on top of the SE exposed to the same
In this case a following rate- determining material (for instance, gas) as the work-
charge-transfer step is, most likely, an ion- ing electrode, thus adopting a well-defined
transfer step. chemical potential µX . The overpotential
The expression for multistep reactions then yields changes in the electrochemical
assumes that all charge transfers occur potential of the ions between the working
at a single interface across the same ϕ. and reference electrodes.
In view of the complexity of the interface The way the measurement of the overpo-
in SSE the multistep reaction can take tential is arranged δ(ϕ) at the rds is equal
place spread over different sites with dif- to the measured potential between the
ferent Galvani potential differences. If so, working electrode and the reversible refer-
Eq. (34a) cannot be applied. For instance, ence electrode, δ(ϕ) = η. Then Eq. (33a)
let us consider the overall reaction in which can be written as,
a neutral adsorbed oxygen atom on a metal
electrode obtains two electrons and finally I = I0 (e(1−β)zqη/kB T − e−βzqη/kB T )
ends inside the SE (for instance, YSZ) at (39a)
an interstitial site: Oad + 2e− → Oi 2− (SE). Eq. (34a) as,
Let us assume that the series of elemen-
tary steps involved are: (1) electronation of I = I0 (eαa qη/kB T − e−αc qη/kB T ) (39b)
the adsorbed oxygen atom: Oad + e− M →
and Eq. (37b) as
Oad − , followed by (2) diffusion along the
metal, (3) transfer of the ion into the SE I = I0 (1 − e−qη/kB T ) or
to an interstitial site, Oad − → Oi − (SE),
and (4) the adding of an electron there: I = I0 (eqη/kB T − 1) (39c)
Oi − (SE) + e− (M) → Oi 2− (SE). It is obvi-
ous that charge transfer occurs at four 2.6.6.4.5 Relating the Overpotential to a
different locations (though all are associ- Difference in a Chemical or Electrochemical
ated with the same electrode reaction) with Potential If one wants to allocate to δ(ϕ)
different ϕ values and Eq. (34a) cannot a relation to a total driving force, this is not
be applied. trivial. A total driving force is associated
It was argued before that in SSE the with a gradient in the electrochemical
transfer of electrons is usually rather rapid potential of electrons, ions, or of the
278 2 Electrochemical Double Layers
chemical potential of neutral species. associated with each step and use for the
For example, for electrons, it is the charge transfer Eq. (33a) or Eq. (37b) and
difference in the electrochemical potential Eq. (39a) or Eq. (39c). δ(ϕ) can, in certain
of the electrons, −δ µ̃e /q = −δ(µe )/q + cases, be related to δ µ̃e , δ µ̃ion or δµX across
δ(ϕ), which is not necessarily δ(ϕ). the interface.
However, in some cases δ(µe ) = 0 and In LSE the specie adsorbed on the
−δ µ̃e /q = δ(ϕ), as is the case in LSE metal electrode, M, are in many cases
when the electron is exchanged between neutral ones. Ions are solvated in the
a metal and a redox couple in the LE, LE. For example, in the process Cl− (LE)
the redox potential of which is fixed by →Clad,M + eM − the adsorbed specie is
a high concentration of the redox couple, neutral Cl. In SSE charged specie are ex-
as then δ(µe ) = 0 also under current. In pected to be adsorbed in key reactions, for
SSE, on the other hand, δ(µe ) = 0, is not example, Oad − on Pt is obtained in the re-
necessarily the case for electrons injected action: Oad + eM − → Oad − . Furthermore,
into an SE and electron exchange between adsorbed species with different charge may
a metal and adsorbed species. coexist.
ϕ may represent δ µ̃ion . For a fixed We have mentioned that reference elec-
concentration of ionic defects at the trodes used for electrode characterization
interface δ µ̃ion = zqδ(ϕ). Furthermore, in SSE (and LSE) are chemically active
if electron transfer is fast and δ µ̃e = 0, ones. For determining the distribution of
one can express zqδ(ϕ) in terms of µ̃e in the SE or MIEC, inert metal elec-
δµX = δ µ̃ion . This yields [51, 60, 68], trodes are used [46, 69, 70].
In SSE it is customary to follow the distri- supply of ions through the SE is usually
bution of the electrochemical potential of negligible and does not contribute signifi-
the electrons, µ̃e , and of the ions, µ̃ion . This cantly the reaction. Instead the ions cover
has the advantage that under equilibrium, the electrode material and thus change the
µ̃e and µ̃ion are uniform. Furthermore, un- catalytic activity of the electrode. The ap-
der current for a reversible electrode with plied voltage governs that coverage. This
respect to electron transfer: µ̃e = 0 and is the so-called NEMCA effect developed
for a reversible electrode with respect to ion by Vayenas and coworkers [72–74]. For ex-
transfer: µ̃ion = 0 at the interface. µ̃e ample, a cell used for the oxidation of CH4
can be directly related to the measurable will be: (−)air,Pt|YSZ(O2− )|Pt,CH4 ,O2 (+)
voltage, V , and µ̃ion + zµ̃e to the chem- where CH4 reacts with O2 supplied to-
ical potential difference at the electrodes, gether with it. The rate of reaction is
µX , that can also be measured. governed by the voltage on the EC, which
controls the coverage of the working Pt
2.6.8 electrode by oxygen coming through the
Heterogeneous Catalysis in SSE YSZ SE.
47. I. Riess, Solid State Ionics 1991, 44, 207. 62. Da. Yu. Wang, A. S. Nowick, J. Electrochem.
48. K. D. Becker, H. Schmalzried, V. von Soc. 1979, 126, 1155.
Wurmb, Solid State Ionics 1983, 11, 213. 63. F. H. van Heuveln, H. J. M. Boumeester, J.
49. J. R. Macdonald, D. R. Franceschetti, J. Electrochem. Soc. 1997, 144, 134.
Chem. Phys. 1978, 68, 1614. 64. B. A. Boukamp, B. A. van Hassel, I. C. Vinke
50. J. R. Macdonald, (Ed.), Impedance Spec- et al., Electrochim. Acta 1993, 38, 1817.
troscopy, John Wiley & Sons, New York, 1987. 65. Y. Jaing, S. Wang, Y. Zhang et al., Solid State
51. S. B. Adler, J. A. Lane, B. C. H. Steele, J. Ionics 1998, 110, 111.
Electrochem. Soc. 1996, 143, 3554. 66. J. Maier, Angew. Chem. 1993, 105, 558.
52. I. Riess, J. Schoonman in CRC Hand- 67. M. Gödickemeier, K. Sasaki, L. J. Gauckler
book of Solid State Electrochemistry (Eds.: et al., J. Electrochem. Soc. 1997, 144, 1635.
P. J. Gellings, H. J. M. Boumeester), CRC 68. I. Riess, Electrode processes in solid oxide
Press, Boca Raton, 1997, pp. 269–294. fuel cells, in Oxygen Ion and Mixed Con-
53. N. W. Ashcroft, N. D. Mermin, Solid State ductors and their Technological Applications,
Physics, Holt-Saunders Intl., Philadelphia, NATO Advanced Study Institute Series E
1976, pp. 367–371. (Eds.: H. L. Tuller, J. Schoonman, I. Riess),
54. R. N. Ediriweera, P. R. Mason, Solid State Kluwer Academic Publishers, Norwell, 2000,
Ionics 1985, 17, 213. pp. 21–56, Vol. 338.
55. J. Maier in Recent Trends in Superionic 69. J. Mizusaki, K. Fueki, T. Mukaibo, Bull.
Solids and Solid Electrolytes (Eds.: S. Chandra, Chem. Soc. Jpn. 1975, 48, 428.
A. Laskar), Academic Press, New York, 1989, 70. C. Rosenkranz, J. Janek, Solid State Ionics
p. 137. 1995, 82, 95.
56. J.O’M. Bockris, A. K. N. Ready, Modern Elec- 71. T. M. Gür, R. A. Huggins in Fast Ion
trochemistry, Plenum/Rosetta, N.Y., 1970, Transport in Solids (Eds.: P. Vashishta,
Vol. 1, 2, Chaps. 8 & 9. J. N. Mundy, G. K. Shenoy), Elsevier, Ams-
57. R. H. Fowler, L. Nordheim, Proc. R. Soc. terdam, New York, 1979.
London, Ser. A 1928, 119, 173. 72. C. G. Vayenas, S. Bebelis, I. V. Yentekakis
58. S. M. Sze, The Physics of Semiconductor et al., Catal. Today 1992, 11(3) 303–442.
Devices, Wiley, N.Y., 1969. 73. D. Tsiplakides, S. G. Neophytides, O. Enea
59. J. Mizusaki, K. Amano, S. Yamauchi et al., et al., J. Electrochem. Soc. 1997, 144, 2072.
Solid State Ionics 1987, 22, 313. 74. S. G. Neophystides, D. Tsiplakides, P. Stone-
60. I. Riess, M. Gödickemeier, L. J. Gauckler, hart et al., Nature 1994, 370, 45.
Solid State Ionics 1996, 90, 91.
61. A.-M. Svensson, S. Sunde, K. Nisancioglu, J.
Electrochem. Soc. 1997, 144, 2719.
282 2 Electrochemical Double Layers
2.7.1
Abstract
Introduction
This chapter deals with charged poly-
PEs are charged macromolecules that are
mers (polyelectrolytes) in solution and at
extensively studied not only because of
surfaces. The behavior of polyelectrolytes
their numerous industrial applications but
(PEs) is markedly different from that of
also from a purely scientific interest [1–4].
neutral polymers. In bulk solutions, that
The most important property of PEs is
is, disregarding the surface effect, there are
their water solubility giving rise to a
two unique features to charged polymers:
wide range of nontoxic, environmentally
first, owing to the presence of long-ranged
friendly, and cheap formulations. On the
electrostatic repulsion between charged theoretical side, PEs combine the field of
monomers, the polymer conformations statistical mechanics of charged systems
are much more extended, giving rise to with the field of polymer science and
a very small overlap concentration and offer quite a number of surprises and
high solution viscosity. Second, the pres- challenges.
ence of a large number of counterions The polymers considered in this review
increases the osmotic pressure of PE solu- are taken as linear and long polymer
tions, making such polymers water soluble chains, containing a certain fraction of
as this is of great importance to many ap- electrically charged monomers. Chemi-
plications. At surfaces, the same interplay cally, this can be achieved, for example,
between monomer–monomer repulsion by substituting neutral monomers with
and counterion degrees of freedom leads acidic ones. Upon contact with water,
to a number of special properties. In par- the acidic groups dissociate into positively
ticular, the adsorption behavior depends charged protons, which bind immediately
on both the concentration of polymers and to water molecules, and negatively charged
the added salt in the bulk. We first de- monomers. Although this process effec-
scribe the adsorption behavior of single tively charges the polymer molecules, the
PE molecules and discuss the necessary counterions make the PE solution electro-
conditions to obtain an adsorbed layer and neutral on macroscopic length scales.
characterize its width. Depending on the The counterions are attracted to
stiffness of the PE, the layer can be flat and the charged polymers via long-ranged
compressed or coiled and extended. We Coulomb interactions, but this physical
2.7 Polyelectrolytes in Solution and at Surfaces 283
(a) (b)
(c)
(d)
the persistence length gives an estimate for monomer size. The mean-squared end-to-
the typical radius of curvature, while tak- end radius of a semiflexible chain is known
ing into account thermal fluctuations. For and reads [6]
synthetic polymers with trans-cis confor-
mational freedom of the chain backbone, R 2 = 20 L + 220 (e−L/0 − 1) (5)
the stiffness is due to fixed bond angles
where the persistence length is 0 and the
and hindered rotations around individual
total contour length of a chain is L. Two
backbone bonds. This effect is even more
limiting behaviors are obtained for R from
pronounced for polymers with bulky side
Eq. (5): for long chains, L 0 , the chains
chains, such as poly-DADMAC, because
behave as flexible ones, R 2 20 L; while
of steric constraints, and the persistence for rather short chains, L 0 , the chains
length is of the order of a few nanome- behave as rigid rods, R L. Comparison
ters. with the scaling of the freely jointed chain
Biopolymers with a more complex struc- model (Eq. 2) shows that a semiflexible
ture on the molecular level tend to be chain can, for L 0 , be described by a
stiffer than simple synthetic polymers. freely jointed chain model with an effective
Some typical persistence lengths encoun- Kuhn length of
tered in these systems are 0 ≈ 5 mm
for tubulin, 0 ≈ 20 µm for actin, and a = 20 (6)
0 ≈ 50 nm for double-stranded DNA. Be-
cause some of these biopolymers are and an effective number of segments or
charged, we will discuss at length the monomers
L
dependence of the persistence length on N= (7)
20
the electrostatic conditions. In some cases
the main contribution to the persistence In this case the Kuhn length takes
length comes from the repulsion between into account the chain stiffness and is
charged monomers. independent from the monomer length.
To describe the bending rigidity of This monomer size is denoted by b
neutral polymers, it is easier to use a whenever there is need to distinguish
continuum model [6], in which one ne- between the monomer size b and the
glects the discrete nature of monomers. persistence length 0 (or Kuhn length a).
The bending energy (rescaled by the ther- In Fig. 1 we show snapshots taken from
mal energy, kB T ) of a stiff or semiflexible Monte-Carlo simulations of a semiflexible
polymer of contour length L is given by chain consisting of 100 polymer beads
of diameter b. The persistence length
2
0 L d2 r(s) is varied from 0 = 2b (Fig. 1b), over
ds (4)
2 0 ds 2 0 = 10b (Fig. 1c), to 0 = 100b (Fig. 1d).
Comparison with the freely jointed chain
where d2 r(s)/ds 2 is the local curvature model (having no persistence length)
of the polymer. We assume here that is given in Fig. 1(a) (a = b, 0 = 0). It is
the polymer segments are nonextendable, seen that as the persistence length is
that is, the tangent vectors are always increased, the chain structure becomes
normalized, |dr(s)/ds| = 1. Clearly, this more expanded. The theoretical prediction
continuum description will only be good for the average end-to-end radius R (Eq. 5)
if the persistence length is larger than the is shown as the black bar on the figure and
2.7 Polyelectrolytes in Solution and at Surfaces 287
gives a good estimate on typical sizes of water is polarizable and reacts to the
semiflexible polymers. presence of a charge with polarization
charges. In addition, and this is by
2.7.3 far a more important effect, water
Properties of Polyelectrolytes in Solution molecules carry a permanent dipole
moment that partially orients in the
For PEs, electrostatic interactions provide vicinity of charged objects. Note that
the driving force for their salient features for water, ε ≈ 80, so that electrostatic
and have to be included in any theoreti- interactions and self-energies are much
cal description. The reduced electrostatic weaker in water than in air (where ε ≈
interaction between two pointlike charges 1) or some other low-dielectric solvents.
can be written as z1 z2 v(r) where Still, the electrostatic interactions are
especially important in polar solvents
e2 because in these solvents, charges
v(r) = (8)
kB T εr dissociate more easily than in unpolar
is the Coulomb interaction between two solvents.
elementary charges, z1 and z2 are the va- 3. In biological systems and most indus-
lences (or the reduced charges in units trial applications, the aqueous solution
of the elementary charge e), and ε is contains mobile salt ions. Salt ions
the medium dielectric constant. Through- of opposite charge are drawn to the
out this review, all energies are given in charged object and form a loosely
units of the thermal energy kB T . The in- bound counterion cloud around it. They
teraction depends only on the distance r effectively reduce or screen the charge
between the charges. The total electrostatic of the object. The effective (screened)
energy of a given distribution of charges electrostatic interaction between two
is obtained from adding up all pairwise charges z1 e and z2 e in the presence
interactions between charges according of salt ions and a polarizable sol-
to Eq. (8). In principle, the equilibrium vent can be written as z1 z2 vDH (r),
behavior of an ensemble of charged parti- with the Debye–Hückel (DH) poten-
cles (e.g. a salt solution) follows from the tial vDH (r) given on the linear-response
partition function, that is, the weighted level by
sum over all different microscopic con- B −κr
figurations, which – via the Boltzmann vDH (r) = e (9)
r
factor – depends on the electrostatic en-
ergy of each configuration. In practice, The Bjerrum length B is defined as
however, this route is very complicated for
several reasons: e2
B = (10)
εkB T
1. The Coulomb interaction (Eq. 8) is
long-ranged and couples many charged and denotes the distance at which the
particles. Electrostatic problems are Coulombic interaction between two unit
typically many-body problems, even for charges in a dielectric medium is equal
low densities. to thermal energy (kB T ). It is a measure
2. Charged objects in most cases are of the distance below which the Coulomb
dissolved in water. Like any material, energy is strong enough to compete with
288 2 Electrochemical Double Layers
the thermal fluctuations; in water at room over the two-body potential (Eq.9). How-
temperatures, one finds B ≈ 0.7 nm. The ever, we will at various points in this review
exponential decay is characterized by the also discuss how to go beyond the DH ap-
so-called screening length κ −1 , which is proximation, for example, in the form of
related to the salt concentration csalt by the nonlinear Poisson–Boltzmann theory
(see Sect. 2.7.5) or a box model for the
κ 2 = 8πz2 B csalt (11) counterion distribution (see Sect. 2.7.6).
strong to swell the polymer coil, we re- expected to have a conformation charac-
call that the electrostatic energy (rescaled terized by an effective persistence length
by the thermal energy kB T ) of a homoge- eff , larger than the bare persistence length
neously charged sphere of total charge Q 0 , that is, one expects eff > 0 .
and radius R is This effect is clearly seen in Fig. 2,
where we show snapshots of Monte-Carlo
3B Q2
Wel = (12) simulations of a charged chain of 100
5R monomers of size b each and bare persis-
The exact charge distribution inside the tence length 0 /b = 1 and several values
sphere only changes the prefactor of order of κ −1 and τ . The number of monomers in
unity and is not important for the scaling an electrostatic blob can be written accord-
arguments. For a polymer of length L ing to Eq. (14) as Lel /0 = (τ 2 B 0 )−2/3
and line charge density τ , the total charge and yields for Fig. 2(a) Lel /0 = 0.25,
is Q = τ L. The electrostatic energy of a for Fig. 2(b) Lel /0 = 0.63, for Fig. 2(c)
(roughly spherical) polymer coil is then Lel /0 = 1.6, and for Fig. 2(d) Lel /0 = 4.
−1/2
Accordingly, in Fig. 2(d) the electrostatic
Wel B τ 2 L3/2 0 (13) blobs consist of four monomers, and the
weakly charged chain crumples at small
The polymer length at which the electro-
length scales. A typical linear charge den-
static self-energy is of order kB T , that is,
sity reached with synthetic PEs such as
Wel 1, follows as
Polystyrenesulfonate (PSS) is one charge
Lel 0 (B 0 τ 2 )−2/3 (14) per two carbon bonds (or, equivalently, one
charge per monomer), and it corresponds
and defines the electrostatic blob size or to τ ≈ 4 nm−1 . Since for these highly flex-
electrostatic polymer length. We expect a ible synthetic PEs the bare persistence
locally crumpled polymer configuration if length is of the order of the monomer
Lel > 0 , that is, if size, 0 b, the typical charge parameter
for a fully charged PE therefore is roughly
τ B 0 < 1 (15) τ 2 B 0 ≈ 3 and is intermediate between
Fig. 2(a and b). Smaller linear charge den-
because the electrostatic repulsion be-
sities can always be obtained by replacing
tween two segments of length 0 is smaller
some of the charged monomers on the
than the thermal energy and is not suf-
polymer backbone with neutral ones, in
ficient to align the two segments. This
which case the crumpling observed in
is in accord with more detailed calcula-
Fig. 2(d) becomes relevant. Larger bare
tions by Joanny and Barrat [11]. A recent
persistence lengths can be achieved with
general Gaussian variational calculation
biopolymers or synthetic PEs with a con-
confirms this scaling result and in ad-
jugated carbon backbone.
dition yields logarithmic corrections [12].
The question now arises as to what
Conversely, for
are the typical chain conformations at
τ B 0 > 1 (16) much larger length scales. Clearly, they
will be influenced by the repulsions.
electrostatic chain–chain repulsion is al- Indeed, √ in the persistent regime, obtained
ready relevant on length scales comparable for τ (B 0 ) > 1, the polymer remains
to the persistence length. The chain is locally stiff even for contour lengths larger
290 2 Electrochemical Double Layers
(a)
(b)
(c)
(d)
than the bare persistence length 0 , and Skolnick and Fixman, reads [13–15]
the effective persistence length is given by
B τ 2
OSF = (18)
eff 0 + OSF (17) 4κ 2
and is calculated from the electrostatic
The electrostatic persistence length, first energy of a slightly bent polymer using
derived by Odijk and independently by the linearized DH approximation (Eq. 9).
2.7 Polyelectrolytes in Solution and at Surfaces 291
It is valid only for polymer conformations Gaussian √ regime is obtained in the opposite
that do not deviate too much from the limit of τ (B 0 ) < 0 κ.
rodlike reference state. The electrostatic The electrostatic persistence length is
persistence length gives a sizable contri- visualized in Fig. 3, in which we present
bution to the effective persistence length snapshots of Monte-Carlo simulations of a
only for OSF > 0 . This is equivalent to charged chain consisting of 100 monomers
the condition of size b. The bare persistence length was
fixed at 0 = b, and the charge-interaction
τ B 0 > 0 κ (19) parameter was chosen to be τ 2 B 0 = 2,
close to the typical charge density in
The persistent regime is obtained for experiments on fully charged synthetic
parameters satisfying both conditions (16) PEs. The snapshots correspond to vary- √
and (19). Another regime called the ing screening lengths of (1) κ −1 /b = 2,
(a)
(b)
(c)
Fig. 3 Snapshots of Monte-Carlo simulations of a PE chain consisting of N = 100
monomers of size b. In all simulations, the bare persistence length is fixed at
0 /b = 1 and the charge-interaction parameter is chosen to be τ 2 B√ 0 = 2. The
snapshots correspond to varying screening lengths of (a) κ −1 /b = 2, leading √ to an
electrostatic contribution to the persistence
√ length of OSF /b = 1; (b) κ −1 /b = 18,
leading to OSF /b = 9; and (c) κ −1 /b = 200, leading to OSF /b = 100. According
to the simple scaling principle (Eq. 17), the effective persistence length in the
snapshots (a–c) should be similar to the bare persistence length in Fig. 1(b–d).
292 2 Electrochemical Double Layers
This electrostatic stiffening is only relevant chain crumpling at small length scales
for the so-called Gaussian-persistent regime and chain stiffening at large length scales
valid for KK > Rel , or equivalently occur simultaneously [16–20]. However,
extremely long chains are needed in order
τ B 0 > (0 κ)3/2 (22) to obtain reliable results for the persistence
length, since the stiffening occurs only at
When this inequality is inverted, the
intermediate length scales and therefore
Gaussian-persistence regime crosses over
fitting of the tangent–tangent correlation
to the Gaussian one.
function is nontrivial. Nevertheless, sim-
The crossover boundaries (Eqs. 16, 19,
ulations point to a different scaling than
22) between the various scaling regimes
that in Eq. (21), with a dependence on the
are summarized in Fig. 5. We obtain three
screening length closer to a linear one, in
distinct
√ regimes. In the persistent√ regime,
qualitative agreement with experimental
for τ (B 0 ) > 0 κ and τ (B 0 ) > 1,
results [3]. The situation is complicated by
the polymer takes on a rodlike struc-
the fact that more recent theories for the
ture with an effective persistence length
single PE chain give different results, some
larger than the bare persistence length
confirming the simple scaling results de-
and given by the OSF expression (Eq. 18).
scribed in Eqs. (18) and (21) [12, 21, 22],
In√ the Gaussian-persistent
√ regime, for
some confirming Eq. (18) while criticiz-
τ (B 0 ) < 1 and τ (B 0 ) > (0 κ)3/2 ,
ing Eq. (21) [11, 23, 24]. This issue is not
the polymer consists of a linear array of
resolved and is under intense current in-
Gaussian electrostatic blobs, as shown in
vestigation. For multivalent counterions,
Fig. 4, with an effective persistence length
fluctuation effects can even give rise to a
KK larger than the electrostatic blob size
and given by Eq. (21). PE collapse purely due to electrostatic in-
√ Finally, in the3/2 Gaus-
teractions [25–27], which is accompanied
sian
√ regime, for τ ( B 0 ) < ( 0 κ) and
τ (B 0 ) < 0 κ, the electrostatic repul- by a negative contribution to the effective
sion does not lead to stiffening effects at persistence length [28].
any length scale.
The persistence length KK was also 2.7.3.1.1 Manning Condensation A pe-
obtained from Monte-Carlo simulations culiar phenomenon occurs for highly
with parameters shown in Fig. 2(d), where charged PEs and is known as the Manning
counterion entropy. This entropic term ex- will be discussed below. An important
plains also why charged polymers can be length scale, schematically indicated in
dissolved in water even when their back- Fig. 6, is the mesh size or correlation
bone is quite hydrophobic. Precipitation length ξ , which measures the length
of the PE chains will also mean that the below which entanglement effects between
counterions are confined within the pre- different chains are unimportant. The
cipitate. The entropy loss associated with mesh size can be viewed as the polymer
this confinement is too large and keeps the (blob) scale below which single-chain
polymers dispersed in solution. In con- statistics are valid. A semidilute solution
trast, for neutral polymers there are no can be thought of being composed of
counterions in solution. Only the second roughly a close-packed array of polymer
term in the osmotic pressure exists and blobs of size ξ .
contributes to the low osmotic pressure of The starting point for the present
these polymer solutions. In addition, this discussion is the screened interaction
can explain the trend toward precipitation between two charges immersed in a
even for very small attractive interactions semidilute PE solution containing charged
between neutral polymers. polymers, their counterions, and, possibly,
additional salt ions. Screening in this case
2.7.3.3 Semidilute Polyelectrolyte Solution is produced not only by the ions but also
In the semidilute concentration regime, by the charged chain segments that can
cm > cm ∗ , different polymer coils are be easily polarized and shield any free
strongly overlapping, but the polymer so- charges.
lution is still far from being concentrated. Using the random-phase approximation
This means that the volume fraction of the (RPA), the effective DH interaction can be
monomers in solution is much smaller written in Fourier space as [49, 50]
than unity, b3 cm 1. In this concentra-
tion range, the statistics of counterions and
+
−
+
−
+
− −
The structure factor S(q) shows a
+
+
+
−
of polymer chains and the electrostatic − −
+
− −
+
−
− −
+
−
+
−
−
+
+ + +
+ + + + + +
concentration range. The mesh size ξ is +
+
on whether the chain form factor of fraction f = 0.5 and with no added salt.
Eq. (32) exhibits flexible-like or rigid-like As the monomer density decreases (bot-
scaling. Concentrating now on the flexible tom to top in the figure), the peak moves
case, that is, the intermediate q-range in to smaller wave numbers and sharpens, in
Eq. (32), the peak of S(q) scales as agreement with previous implementations
1/4 of the RPA. In Fig. 7(b) we show the same
24πB f 2 cm
q∗ (34) data in a different representation. Here we
beff clearly demonstrate that the large-q region
in agreement with experimental results. A already is dominated by the 1/q behavior
peak is only obtained if the left-hand side of of the single-chain structure factor, S0 (q).
Eq. (33) is dominated by the q-dependent Since neutron-scattering data easily extend
part, that is, if (q ∗ )2 > κsalt
2 + 4πz f c .
B m
to wave numbers as high as q ∼ 5 nm−1 ,
In Fig. 7(a) we show density-normalized the stiff rodlike behavior in the high q-
scattering curves for a PE solution char- limit, exhibited on such a plot, will be
acterized by the persistence length eff = important in interpreting and fitting ex-
1 nm (taken to be constant and thus inde- perimental data even at lower q-values.
pendent of the monomer concentration), In a semidilute solution there are three
with monomer length b = 0.38 nm (as ap- different, and in principle, independent
propriate for poly-DADMAC) and charge length scales: the mesh size ξ , the
30
20
S(q) /cm
10
0
0 1 2 3 4 5
q
(a) [nm−1]
20
Fig. 7(a) RPA prediction for the
rescaled structure factor S(q)/cm of a
15 semidilute PE solution with persistence
length eff = 1 nm, monomer length
q S(q) /cm
screening length κ −1 , and the persistence semidilute PE solutions, all three length-
length eff . In the absence of added salt, scales scale in the same manner with cm ,
−1/2
the screening length scales as namely, as ∼ cm , as is known also from
experiments [43, 44, 54] and previous theo-
κ −1 ∼ (zB f cm )−1/2 (35) retical calculations [55, 56]. In simulations
of many PE chains, the reduction of the
Assuming that the persistence length is
chain size due to screening by PE chains
larger or of the same order of magnitude as
was clearly seen [57–60].
the mesh size, as is depicted in Fig. 6, the
polymer chains can be thought of straight
2.7.4
segments between different crossing links.
Adsorption of a Single Polyelectrolyte Chain
Denoting the number of monomers inside
a correlation blob as g, this means that ξ ∼
After reviewing bulk properties of PE
bg. The average monomer concentration
solutions, we elaborate on the adsorption
scales as cm ∼ g/ξ 3 , from which we
diagram of a single semiflexible PE on an
conclude that
oppositely charged substrate. In contrast
ξ ∼ (bcm )−1/2 (36) to the adsorption of neutral polymers, the
resulting phase diagram shows a large
Finally, the persistence length within a region where the adsorbed polymer is
semidilute PE solution can be calculated flattened out on the substrate and creates
by considering the electrostatic energy cost a dense adsorption layer.
for slightly bending a charged rod. In PE Experimentally, the adsorption of char-
solutions, it is important to include in ged polymers on charged or neutral sub-
addition to the screening by salt ions the strates has been characterized as a function
screening due to charged chain segments. of the polymer charge, chemical com-
This can be calculated by using the RPA position of the substrate, pH, and ionic
interaction (Eq. 29). Since the screening strength of the solution [61, 62], as well as
due to polymer chains is scale-dependent the substrate charge density [63–68]. Re-
and increases for large separations, a q- peated adsorption of anionic and cationic
dependent instability is encountered and PEs can lead to well-characterized mul-
leads to a persistence length [53] tilayers on planar [69–74] and spherical
substrates [75–77]. Theoretically, the ad-
−1/2
sd
OSF ∼ (bcm ) (37) sorption of PEs on charged surfaces
poses a much more complicated prob-
where the ‘‘sd’’ superscript stands for
lem than the corresponding adsorption
‘‘semidilute.’’ This result is a generaliza-
of neutral polymers. The adsorption pro-
tion of the OSF result for a single chain
cess results from a subtle balance between
and applies to semidilute solutions. Com-
electrostatic repulsion between charged
paring the three lengths, we see that
monomers, leading to chain stiffening,
and electrostatic attraction between the
sd zB f −1
ξ ∼ OSF ∼ κ (38) substrate and the polymer chain. The
b
√ adsorption problem has been treated the-
Since the prefactor (B f/b) for syn- oretically employing the uniform expan-
thetic fully charged polymers is roughly sion method [78] and various continuous
of order unity, one finds that for salt-free mean-field theories [79–83]. In all these
300 2 Electrochemical Double Layers
works, the polymer density is taken to by the Gouy-Chapman (GC) theory for
be constant in directions parallel to the counterion distribution close to a charged
surface. surface. Although these effects are clearly
The adsorption of a single semiflexible important, it is difficult to include them
and charged chain on an oppositely systematically, and we remain at the lin-
charged plane [84] can be treated as a earized DH level.
generalization of the adsorption of flexible Because of the confinement in the
polymers [85]. A PE characterized by the adsorbed layer, the polymer feels an
linear charge density τ is subject to entropic repulsion. If the layer thickness
an electrostatic potential created by σ , δ is much smaller than the effective
the homogeneous surface charge density persistence length of the polymer, eff ,
(per unit area). Because this potential as depicted in Fig. 8(a), a new length scale,
is attractive for an oppositely charged the so-called deflection length λ, enters the
substrate, we consider it as the driving description of the polymer statistics. The
force for the adsorption. More complex deflection length λ measures the average
interactions are neglected. They are due to distance between two contact points of
the dielectric discontinuity at the substrate the polymer chain with the substrate. As
surface and are due to the impenetrability shown by Odijk, the deflection length
of the substrate for salt ions. 1/3
scales as λ ∼ δ 2/3 eff and is larger than
An ion in solution has a repulsive the layer thickness δ but smaller than
interaction from the surface when the the persistence length eff [86, 87]. The
solution dielectric constant is higher than entropic repulsion follows in a simple
that of the substrate. This effect can lead to manner from the deflection length by
desorption for highly charged PE chains. assuming that the polymer loses roughly
On the contrary, when the substrate is a free energy of one kB T per deflection
a metal, there is a possibility to induce length.
PE adsorption on noncharged substrates On the other hand, if δ > eff , as shown
or on substrates bearing charges of the in Fig. 8(b), the polymer forms a random
same sign as the PE. See Ref. [84] for more coil with many loops within the adsorbed
details. layer. For a contour length smaller than
Within the linearized DH theory, the L ∼ δ 2 /eff , the polymer obeys Gaussian
electrostatic potential of a homogeneously statistics and decorrelates into blobs with
charged plane is an entropic cost of one kB T per blob. The
entropic repulsion force per monomer unit
Vplane (x) = 4πB σ κ −1 e−κx (39)
length is thus [86, 87]
Assuming that the polymer is adsorbed −5/3 −1/3
δ eff for δ eff
over a layer of width δ smaller than frep ∼
the screening length κ −1 , the electrostatic eff δ −3 for δ eff
(41)
attraction force per monomer unit length
where we neglected a logarithmic correc-
can be written as
tion factor that is not important for the
fatt = −4πB σ τ (40) scaling arguments. As shown in the pre-
ceding section, the effective persistence
We neglect nonlinear effects due to coun- length eff depends on the screening
terion condensation. They are described length and the line charge density; in
2.7 Polyelectrolytes in Solution and at Surfaces 301
κ−1ᐉ0−1
κ−1ᐉ 0−1 Adsorbed-
Adsorbed coiled Adsorbed-flat
coiled 5/21
Adsorbed Σ −2/3 1/3
1/3 flat
1 Desorbed 5/7
1
3/5
Σ −2/3 Desorbed 5/7
τ ᐉ1/2
Β ᐉ0
1/2 τ ᐉ1/2
Β ᐉ0
1/2
Fig. 9 Adsorption scaling diagram shown on a log–log plot for (a) strongly charged
3/2 1/2
surfaces,
= σ 0 B > 1, and for (b) weakly charged surfaces,
< 1. We find a
desorbed regime, an adsorbed phase in which the polymer is flat and dense, and an
adsorbed phase in which the polymer shows loops. It is seen that a fully charged PE is
expected to adsorb in a flat layer, whereas charge-diluted PEs can form coiled layers with
loops and dangling ends. The broken lines denote the scaling boundaries of PE chains in
the bulk as shown in Fig. 5. The numbers on the lines indicate the power-law exponents of
the crossover boundaries between the regimes.
2.7 Polyelectrolytes in Solution and at Surfaces 303
Z
Rp = 5 nm. Shown is the main
transition from the unwrapped 50
configuration (at the bottom) to the
wrapped configuration (at the top) as
a function of sphere charge Z and Unwrapped
inverse screening length κ. Wrapping 0
0 2 4 6 8 10
is favored at intermediate salt concen-
trations. The parameters are chosen κ
for the problem of DNA-histone com-
[nm−1]
plexation. (Adapted from Ref. [99].)
widely used to stabilize colloidal suspen- attraction between the sphere and the
sions [93] and to fabricate hollow poly- DNA, and electrostatic repulsion between
meric shells [75–77]. When the curvature the DNA segments with respect to the
of the small colloidal particles is large chain configuration [99]. Configurational
enough, it can lead to a much more fluctuations away from this ground state
pronounced effect for PE adsorption as are unimportant for such stiff polymers.
compared with neutral polymer. This is We show in Fig. 10 the main transition
mainly due to the fact that the electrostatic between an unwrapped state, at low sphere
energy of the adsorbed PE layer depends charge Z, and the wrapped state, at large
sensitively on curvature [94–98]. Bending sphere charge Z. It is seen that at values
a charged polymer around a small sphere of the sphere charge between Z = 10 and
costs a large amount of electrostatic en- Z = 130, the wrapping only occurs at in-
ergy, which will disfavor adsorption of termediate values of the inverse screening
1/2
long, strongly charged PE at very low salt length κ ∼ csalt . At low salt concentrations
concentrations. (lower left corner in the phase diagram),
In Fig. 10 we show the adsorption phase the self-repulsion between DNA segments
diagram of a single stiff PE of finite length prevents wrapping, while at large salt
that interacts with an oppositely charged concentrations (lower right corner in the
sphere of charge Z (in units of e). The diagram), the electrostatic attraction is not
specific parameters were chosen as ap- strong enough to overcome the mechani-
propriate for the complexation of DNA (a cal bending energy of the DNA molecule.
negatively charged, relatively stiff biopoly- These results are in good agreement
mer) with positively charged histone pro- with experiments on DNA/histone com-
teins, corresponding to a DNA length of plexes [100]. Interestingly, the optimal salt
L = 50 nm, a chain persistence length of concentration, where a minimal sphere
0 = 30 nm, and a sphere radius of Rp = charge is needed to wrap the DNA, oc-
5 nm. The phase diagram was obtained by curs at physiological salt concentration,
minimization of the total energy including for κ −1 ≈ 1 nm. For colloidal particles
bending energy of the DNA, electrostatic of larger size and for flexible synthetic
304 2 Electrochemical Double Layers
200
cm(x)/cbm
100
0
0 10 20 30 40 50
x
[Å]
Fig. 11 Adsorption profiles obtained by numerical solutions
of Eqs. (55) and (56) for several sets of physical parameters in
the low-salt limit. The polymer concentration scaled by its bulk
value cmb = φ 2 is plotted as a function of the distance from the
b
surface. The different curves correspond to: f = 1, a = 5 Å and
ys = eψs /kB T = −0.5 (solid curve); f = 0.1, a = 5 Å and
ys = −0.5 (dots); f = 1, a = 5 Å and ys = −1.0 (short dashes);
f = 1, a = 10 Å and ys = −0.5 (long dashes); and f = 0.1,
a = 5 Å and ys = 1.0 (dot–dash line). For all cases,
φb2 = 10−6 Å−3 , v2 = 50 Å3 , ε = 80, T = 300 K, and
csalt = 0.1 mM. (Adapted from Ref. [124].)
and x/D. We concentrate now on two are much stronger than the excluded
limiting regimes where Eq. (59) can be volume ones.
justified: (1) the low-salt regime D κ −1 Neglecting the latter interactions and
and (2) the high-salt regime D κ −1 . minimizing the free energy with respect
We first discuss the case of constant to D and cM gives
surface potential, which can be directly
compared to the numerical profiles. Then a2 1
D2 ∼ (61)
we note the differences with the constant f |ys | f |ψs |
surface charge boundary condition in and
which the interesting phenomenon of |ys |2
charge overcompensation is discussed in cM ∼ |ψs |2 (62)
4πlB a 2
detail.
recalling that ys = eψs /kB T . As discussed
above, these expressions are valid as
2.7.5.3.1 Low-salt Regime D κ −1 and
long as (1) D κ −1 and (2) the ex-
ψs = constant In the low-salt regime the
cluded volume term in Eq. (60) is negli-
effect of the small ions can be neglected
gible. Condition (1) translates into csalt
and the free energy (per unit surface area)
f |ys |/(8πlB a 2 ). For |ys | 1, a = 5 Å and
(Eqs. 51–54) is approximated by (see also
lB = 7 Å this limits the salt concentra-
Refs. [80, 124])
tion to csalt /f 0.4 M. Condition (2)
on the magnitude of the excluded vol-
a2 ume term can be shown to be equivalent
F cM − f |ys |cM D
6D to f v2 |ys |/ lB a 2 . These requirements
1 are consistent with the data presented in
+ 4πlB f 2 cM
2
D 3 + v2 cM
2
D (60)
2 Fig. 11.
We recall that the profiles presented
In the above equation and in what follows, in Fig. 11 were obtained from the nu-
we neglect additional prefactors of order merical solution of Eqs. (55) and (56),
unity in front of the various terms that arise including the effect of small ions and
from inserting the scaling profile (Eq.59) excluded volume. The scaling relations
into the free energy. The first term of are verified by plotting in Fig. 12 the
Eq. (60) is the elastic energy characterizing same sets of data as in Fig. 11, using
the response of the polymer to concentra- rescaled variables as defined in Eqs. (61)
tion inhomogeneities. The second term and (62). That is, the rescaled electro-
accounts for the electrostatic attraction of static potential ψ(x)/ψs and polymer con-
the polymers to the charged surface. The centration cm (x)/cM ∼ cm (x)a 2 /|ys |2 are
third term represents the Coulomb repul- plotted as functions of the rescaled dis-
sion between adsorbed monomers. The tance x/D ∼ xf 1/2 |ys |1/2 /a. The different
last term represents the excluded volume curves roughly collapse on the same curve.
repulsion between adsorbed monomers, In many experiments the total amount
where we assume that the monomer con- of adsorbed polymer per unit area
centration near the surface is much larger is measured as function of the physical
than the bulk concentration cM φb2 . characteristics of the system such as the
In the low-salt regime and for highly charge fraction f , the pH of the solution,
charged PEs, the electrostatic interactions or the salt concentration csalt [103–110].
2.7 Polyelectrolytes in Solution and at Surfaces 309
0.8
0
0.6
cm(x)/c M
ψ (x) / | ψs |
−0.5 0.4
0.2
−1
0
0 1 2 3 0 1 2 3
x/D x/D
(a) (b)
Fig. 12 Scaling behavior of PE adsorption in the low-salt regime (Eqs. 61 and 62). (a) The
rescaled electrostatic potential ψ(x)/|ψs | as a function of the rescaled distance x/D. (b) The
rescaled polymer concentration cm (x)/cM as a function of the same rescaled distance. The
profiles are taken from Fig. 11 (with the same notation). (Adapted from Ref. [124].)
This quantity can be easily obtained from 2.7.5.3.2 High-salt Regime D κ −1 and
our scaling expressions yielding ψs = constant Let us now consider the
opposite case of a high ionic strength
∞
solution. Here, D is much larger than
= [cm (x) − φb2 ] dx
0 κ −1 , and the electrostatic interactions are
|ys |3/2 |ψs |3/2 short ranged with a cutoff κ −1 . The free
DcM 1/2
∼ (63) energy of the adsorbing PE layer (per unit
lB af f 1/2
surface area) then reads
The adsorbed amount (f ) in the low-
a2
salt regime is plotted in the inset of F cM − f |ys |cM κ −1
Fig. 13(a). As a consequence of Eq. (63), 6D
decreases with increasing charge frac- 1
+ 4πlB f 2 κ −2 cM
2 2
D + v2 cM D (64)
tion f . Similar behavior was also re- 2
ported in experiments [106]. This effect
is at first glance quite puzzling be- The electrostatic cutoff enters in two
cause as the polymer charge increases, places. In the second term, only the
the chains are subject to a stronger at- first layer of width κ −1 interacts elec-
traction to the surface. On the other trostatically with the surface. In the
hand, the monomer–monomer repulsion third term, each charged layer situated
is stronger and, indeed, in this regime, at point x interacts only with layers at
the monomer–monomer Coulomb repul- x for which |x − x | < κ −1 . This term
sion scales as (f cM )2 and dominates can be also viewed as an additional
over the adsorption energy that scales electrostatic excluded volume with vel ∼
as f cM . lB (f/κ)2 .
310 2 Electrochemical Double Layers
0.002 0.003
0.01
csalt −> 0 csalt −> 0
0.002 0.001 0.005
0 0.002 0
0 0.5 1 −4 −2 0 2 4
[Å−2]
[Å−2]
csalt = 1 M
Γ
Γ
0.001
0.001 csalt = 1 M
csalt = 0.1 M
csalt = 0.1 M
0 0
0 0.2 0.4 0.6 0.8 1 −4 −2 0 2 4
(a) f (b) pH − pK0
Fig. 13 Typical adsorbed amount as a used for ε, T, and v2 are the same as in Fig. 11,
function of (a) the charge fraction f and (b) the while ys = eψs /kB T = −0.5 and a = 5Å. The bulk
pH − pK0 of the solution for different salt concentration cmb = φ 2 is assumed to be much
b
concentrations (Eq. 67). The insets correspond smaller than cM . (Adapted from Ref. [124].)
to the low-salt regime (Eq. 63). The parameters
Minimization of the free energy gives given by the pH of the solution through
the expression
1/2
κa 2 c
D ∼ salt (65) 10pH−pK0
f |ys | f |ψs | f = (68)
1 + 10pH−pK0
and
f 2 |ys |2 /(κa)2 where pK0 = − log10 K0 and K0 is the ap-
cM ∼ (66) parent dissociation constant. We note that
f 2 /csalt + αv2
this relation is only strictly valid for in-
yielding finitely dilute monomers and that distinct
deviations from it are observed for PEs
−1/2
f |ys |csalt f |ψs | −1/2 because of the electrostatic repulsion be-
∼ ∼ c tween neighboring dissociating sites [126].
salt + αv2
f 2 /c vel + αv2 salt
(67) Still, the results in Fig. 13(b) capture the
where α is a numerical constant of order main qualitative trends of pH-dependent
unity that depends on the profile details. PE adsorption.
The adsorption behavior is depicted in Another interesting observation that can
Figs. 13 and 14. Our scaling results are be deduced from Eq. (67) is that is only a
−1/2
in agreement with numerical solutions function of f csalt . Indeed, as can be seen
of discrete lattice models (the multi- in Fig. 13, csalt only affects the position of
Stern layer theory) [61, 62, 111, 117–120]. the peak and not its height.
In Fig. 13, is plotted as function of The effect of salt concentration is shown
f (Fig. 13a) and the pH (Fig. 13b) for in Fig. 14, where is plotted as function of
different salt concentrations. The behavior the salt concentration csalt for two charge
as seen in Fig. 13(b) represents annealed fractions f = 0.01 and 0.25. The curves
PEs where the nominal charge fraction is on the right-hand side of the graph are
2.7 Polyelectrolytes in Solution and at Surfaces 311
f = 0.01
0.02
[Å−2]
Γ
0.01
f = 0.25
0
10−6 10−5 10−4 10−3 10−2 10−1 1 10
csalt
[M]
Fig. 14 The adsorbed amount as a function of the salt
concentration csalt (Eq. 67) for f = 0.01 and 0.25. The solid curves
on the right-hand side correspond to the scaling relations in the
high-salt regime (Eq. 67). The horizontal lines on the left-hand side
mark the low salt values (Eq. 67). The dashed lines serve as guides
to the eye. The parameters used are ε = 80, T = 300 K, v2 = 50Å3 ,
a = 5Å, ys = −2.0. (Adapted from Ref. [124].)
calculated from the high-salt expression For strongly charged PEs, on the other
for (Eq. 67). The horizontal lines on the hand, the balance between these two elec-
left-hand side of the graph indicate the trostatic terms depends on the amount of
low-salt values of (Eq. 63). The dashed salt. At low salt concentrations, the domi-
lines in the intermediate salt regime serve nant interaction is the monomer–surface
only as guides to the eye since our scaling Coulomb repulsion. Consequently, addi-
approach is not valid when D and κ −1 are tion of salt screens this interaction and
of the same order. increases the adsorbed amount. When the
Emphasis should be drawn to the dis- salt concentration is high enough, this
tinction between weakly and strongly Coulomb repulsion is screened out and the
charged PEs. For weak PEs, the adsorbed effect of salt is to weaken the surface at-
amount is a monotonously decreas- traction. At this point the adsorbed amount
ing function of the salt concentration starts to decrease. As a result, the behav-
csalt in the whole range of salt con- ior over the whole concentration range is
centrations. The reason being that the nonmonotonic with a maximum at some
monomer–monomer Coulomb repulsion, ∗ , as seen in Fig. 14.
optimal value csalt
proportional to f 2 , is weaker than the From this analysis and from Figs. 13,
monomer–surface interaction, which is 14 and Eq. (67), it is now natural to divide
linear in f . the high-salt regime into two subregimes
312 2 Electrochemical Double Layers
according to the PE charge. At low diagram. For weak PEs, this cut goes
charge fractions (subregime HS I), f through the left-hand side of the diagram
f ∗ = (csalt v2 )1/2 , the excluded volume starting from the low-salt regime and,
term dominates the denominator of upon addition of salt, into the HS I regime.
Eq. (67) and Such a path describes the monotonous be-
−1/2
havior inferred from Fig. 14 for the weak
∼ f |ψs |csalt (69) PE (f = 0.01). For strong PEs, the cut
goes through the right-hand side of the di-
whereas at high f (subregime HS II), f
agram, starting from the low-salt regime,
f ∗ , the monomer–monomer electrostatic
passing through the HS II regime, and
repulsion dominates and decreases with
ending in the HS I. The passage through
f and increases with csalt :
the HS II regime is responsible for the
∼ csalt |ψs |f −1
1/2 nonmonotonous behavior inferred from
(70)
Fig. 14 for the strong PE (f = 0.25).
The various regimes with their crossover Similarly, Fig. 13(a,b) correspond to hor-
lines are shown schematically in Fig. 15. izontal scans through the top half of the
Keeping the charge fraction f constant diagram. As long as the system is in the
while changing the amount of salt cor- HS I regime, the adsorbed amount in-
responds to a vertical scan through the creases when the polymer charge fraction
1
High salt HS II
(HS I)
0.1
[M]
csalt
0.01
0.001
increases. As the polymer charge fur- 2.7.5.4.1 Low Salt Limit: D κ −1 As-
ther increases, the system enters the suming that there is only one length scale
HS II regime and the adsorbed amount characterizing the potential behavior in the
decreases. Thus, the nonmonotonous be- vicinity of the surface, as demonstrated in
havior of Fig. 13. We finally note that Fig. 12(a), we find that the surface poten-
the single-chain desorption transition for tial ψs and the surface charge σ are related
a flexible chain, which was discussed in by ψs ∼ σ eD. In the low salt limit we find
Sect. 2.7.4, is also valid for adsorption from from Eq. (61)
solutions and will lead to a desorption tran-
sition at very high salt concentrations in D ∼ (f σ lB )−1/3 (71)
Fig. 15 (which is not shown for clarity).
in agreement with Eq. (47).
2.7.5.4 Overcompensation of Surface Let us define two related concepts via the
Charges: Constant σ effective surface charge density defined
We turn now to a different electrostatic as σ = f − σ , which is the sum of
boundary condition of constant surface the adsorbed polymer charge density and
charge density and look at the interesting the charge density of the bare substrate.
phenomenon of charge compensation by For σ = 0, the adsorbed polymer charge
the PE chains in relation to experiments for exactly compensates the substrate charge.
PE adsorption on flat surfaces, as well as on If σ is positive, the PE overcompensates
charged colloidal particles [72, 73, 75–77]. the substrate charge and more polymer
What was observed in experiments is that adsorbs than is needed to exactly cancel
PEs adsorbing on an oppositely charged the substrate charge. If σ is positive and
surface can overcompensate the original reaches the value σ = σ , it means that
surface charge. Because PEs create a thin the PE charge is f = 2σ and leads to
layer close to the surface, they can act as a charge inversion of the substrate charge.
an effective absorbing surface to a second The effective surface charge consisting of
layer of PEs having an opposite charge the substrate charge and the PE layer has
compared to the first layer. Repeating the a charge density that is exactly opposite to
adsorption of alternating positively and the original substrate charge density σ .
negatively charged PEs, it is possible to Do we obtain overcompensation or
create a multilayer structure of PEs at the charge inversion in the low salt limit
surface. Although many experiments and within mean-field theory? Using scaling
potential applications for PE multilayers arguments, this is not clear since we find
exist, the theory of PE overcompensation that σ ∼ f ∼ σ . That is, each of the two
is only starting to be developed [83, 84, terms in σ scales linearly with σ , and
123–125, 127, 128]. the occurrence of overcompensation or
The scaling laws presented for constant charge inversion will depend on numerical
ψs can be used also for the case of constant prefactors determining the relative sign
surface charge. A surface held at a constant of the two opposing terms. However, if
potential ψs will induce a surface charge we look at the numerical solution for the
density σ . The two quantities are related by mean-field electrostatic potential (Fig. 12),
dψ/ dx = 4πσ e/ε at x = 0. We will now we see indeed that all plotted profiles have
consider separately the two limits: low salt a maximum of ψ(x) as function of x. An
D κ −1 and high salt D κ −1 . extremum in ψ means a zero local electric
314 2 Electrochemical Double Layers
field. Or equivalently, using Gauss law, this relative strength of the surface charge den-
means that the integrated charge density sity σ and the nonelectrostatic interaction,
from the wall to this special extremum it is also possible to cause a charge over-
point (including surface charges) is exactly compensation and even an exact charge
zero. At this point the charges in solution inversion in a special case.
exactly compensate the surface charges. Finally, we note that the dependence of
the charge parameter σ on the amount of
salt, csalt , is different for constant surface
2.7.5.4.2 High Salt Limit: D κ −1
charge and constant surface potential
When we include salt in the solution
cases. While for the former, σ is
and look at the high salt limit, the
nonmonotonous and has a maximum (as
situation is more complex. The only length
mentioned above) as function of the salt
characterizing the exponential decay of ψ
concentration, in the latter case, σ/σ is
close to the surface is the DH screening
a monotonic decreasing function of csalt
length. Hence, using dψ/ dx|s ∼ σ e yields
(Eq. 73). This can be explained by the extra
ψs ∼ σ eκ −1 , and therefore from Eq. (65),
powers of csalt in the latter case coming
−1/2
from the relation ψs ∼ σ csalt ∼ σ κ −1 .
κ 2a2
D∼ ∼ κ 2 f −1 σ −1 (72) Let us remark that in other theories the
f σ lB overcharging is due to lateral correlations
between adsorbed PEs, which in conjunc-
The estimation of the PE layer charge can
tion with screening by salt ions leads to
be obtained by using the expression for D
strongly overcharged surfaces [84, 128].
and cM in this high salt limit (Eqs. 63–65),
yielding
2.7.5.5 Final Remarks on Adsorption from
κ −2 )
βσ (8πlB csalt βσ Semidilute Solutions
f = The results presented earlier (Sect. 2.7.5)
1 + αv2 csalt f −2 1 + v2 /vel
for adsorption from solutions have been
(73) derived using mean-field theory. Hence,
where vel = f 2 /αcsalt is the electrostatic lateral fluctuations in the polymer and
contribution to the second-virial coefficient ionic concentrations are neglected. In ad-
v2 and α > 0 and β > 1 are positive dition, we neglect the delicate influence of
numerical factors. the charges on the PE persistence length
We see that σ = f − σ is a decreas- and any deviations from ground state dom-
ing function of v2 . Charge overcompensa- inance [9]. The region of validity of the
tion can occur when v2 is smaller than vel theory is for long and weakly charged
(up to a prefactor of order unity). When polymer chains in contact with a mod-
v2 can be neglected in the vicinity of the erately charged surface. The PE solution is
surface, or when v2 = 0 (theta solvents), placed in contact with a single and ideal
there is always charge overcompensation, surface (infinite, flat, and homogeneous).
σ = (β − 1)σ > 0. This is the case of The problem reduces then to an effec-
strongly charged PEs. Similar conclusions tive one-dimensional problem depending
have been mentioned in Refs. [83, 127] only on the distance from the charged
where v2 was taken as zero but the sur- surface. We take very simple boundary
face has a nonelectrostatic short range conditions for the surface assuming that
interaction with the PE. By tuning the the polymer concentration is zero on the
2.7 Polyelectrolytes in Solution and at Surfaces 315
surface and keeping the surface in con- The scaling arguments are then repeated
stant potential or constant surface charge for constant σ boundary conditions. It
conditions. is found that the PE can possibly cause
We find numerical solutions for the charge overcompensation and even in-
PE profile equations in various cases. version of the nominal substrate charge,
These numerical solutions agree well with leading the way to multilayer formation of
simple scaling assumptions describing the positively and negatively charged PEs. The
adsorption of PEs. Scaling expressions for scaling approach can serve as a starting
the amount of adsorbed polymer and point for further investigations. The ana-
the width D of the adsorbed layer, as a lytical and approximated expressions are
function of the fractional charge f and the valid only in specific limits. Special atten-
salt concentration csalt , are obtained for tion should be directed to the crossover
two cases: constant ψs and constant σ . regime where D and κ −1 are of compara-
For constant ψs and in the low-salt ble size.
regime, a f −1/2 dependence of is The problem of charge inversion is not
found. It is supported by our numerical well understood at present. Alternative
solutions of the profile Eqs. (55) and (56) approaches rely on lateral correlations be-
and is in agreement with experiment [106]. tween semiflexible adsorbed PE chains,
This behavior is due to strong Coulomb which also can lead to strong overcompen-
repulsion between adsorbed monomers sation of surface charges [84, 128].
in the absence of salt. As f decreases,
the adsorbed amount increases until the 2.7.6
electrostatic attraction becomes weaker Polyelectrolyte Brushes
than the excluded volume repulsion, at
which point, starts to decrease rapidly. Charged polymers that are densely end-
At high salt concentrations it is not grafted to a surface are called polyelectrolyte
possible to neglect the excluded vol- brushes or charged brushes. They have
ume interaction of the monomers since been the focus of numerous theoreti-
the electrostatic interactions are screened cal [129–138] and experimental [139–142]
by the salt. We obtain two limit- studies. In addition to the basic inter-
ing behaviors: (1) For weakly charged est, charged brushes are considered for
PEs, f f ∗ = (csalt v2 )1/2 , the adsorbed their applications as efficient means for
amount increases with the fractional preventing colloids in polar media (such
charge and decreases with the salt con- as aqueous solutions) from flocculating
centration, ∼ f/ (csalt ), owing to the and precipitating out of solution [93]. This
monomer–surface electrostatic attraction. stabilization arises from steric (entropic)
(2) For strong PEs, f f ∗ , the ad- and electrostatic (energetic) repulsion. A
sorbed amount decreases with the frac- strongly charged brush is able to trap
tional charge and increases
with the salt its own counterions and generates a
concentration, ∼ (csalt )/f , owing to layer of locally enhanced salt concen-
the dominance of monomer–monomer tration [131]. It is thus less sensitive to
electrostatic repulsion. Between these two the salinity of the surrounding aqueous
regimes, we find that the adsorbed amount medium than a stabilization mechanism
reaches a maximum in agreement with based on pure electrostatics (i.e. without
experiments [107, 110]. polymers).
316 2 Electrochemical Double Layers
(a) L
+ + +
+ +
−
− − − − − −
+ + + +
+
− − − −
−
+ − + +
+ +
+
− −
− − −
−
+ + − + +
−
+
− − − +
− −
− −
+ + +
+ +
+
− − −
+ − −
+ + +
+ +
L≈H
(b)
The last contribution is the stretching In this case, minimization of Fos + Fel
energy of the chains, which is with respect to the counterion height H
leads to
3L2 1
Fst = ρ (77) H ∼ (78)
2N a 2 zB Nfρ
which is the Gouy-Chapman length for z-
Here, a is the monomer size or Kuhn
valent counterions at a surface of surface
length of the polymer, implying that
charge density σ = Nfρ. Balancing now
we neglect any chain stiffness for the
the polymer stretching energy Fst and the
brush problem. The different free en-
electrostatic energy Fel , one obtains the
ergy contributions lead, upon minimiza-
so-called Pincus brush
tion with respect to the two length
scales H and L, to different behaviors. L N 3 ρa 2 B f 2 (79)
Let us first consider the weak charg-
ing limit, that is, the situation in which In the limit of H ≈ L, the PE brush
the counterions leave the brush, H > L. can be considered as neutral and the
318 2 Electrochemical Double Layers
repulsions, their chain statistics is re- will be depleted from the surface. We
lated to that of semiflexible polymers. also briefly review the phenomenon of
Neutral and charged semiflexible polymers charge overcompensation and inversion,
are controlled by their bending rigidity, when the adsorbed PE layer effectively
which is usually expressed in terms of inverses the sign of the surface charge
a persistence length (see Sect. 2.7.2.2). leading the way to formation of PE multi-
For PEs, the electrostatic interaction layers.
considerably influences this persistence Important topics that we have left out
length. are the dynamics of PE solutions, which
In solution, we have considered the scal- is reviewed in Ref. [4], and the behavior of
ing behavior of a single PE (Sect. 2.7.3.1). PEs under bad-solvent conditions [39–41].
The importance of the electrostatic persis- In the future we expect that studies of
tence length was stressed. The Manning PEs in solutions and at surfaces will be
condensation of counterions leads to a directed more toward biological systems.
reduction of the effective linear charge We mentioned in this review the complex-
density (Sect. 2.7.3.1.1). Excluded volume ation of DNA and histones (Sect. 2.7.4.1).
effects are typically less important than This is only one of many examples of
for neutral polymers (Sect. 2.7.3.1.2). Di- interest in which charged biopolymers,
lute PE solutions are typically dominated receptors, proteins, and DNA molecules
by the behavior of the counterions. So interact with each other or with other
is the large osmotic pressure of dilute cellular components. The challenge for
PE solutions due to the entropic contri- future fundamental research will be to
bution of the counterions (Sect. 2.7.3.2). try to understand the role of electrostatic
Semidilute PE solutions can be described interactions combined with specific bi-
by the RPA, which in particular yields ological (lock–key) mechanisms and to
the characteristic peak of the structure
infer on biological functionality of such
factor.
interactions.
At surfaces, we discussed in detail
the adsorption of single PEs (Sect. 2.7.4),
the adsorption from semidilute solutions Acknowledgment
(Sect. 2.7.5), and the behavior of end-
grafted PE chains (Sect. 2.7.6). We tried It is a pleasure to thank our collabora-
to express the PE behavior in terms of tors I. Borukhov, J.F. Joanny, K. Kunze,
a few physical parameters such as the L. Leibler, R. Lipowsky, H. Orland, M.
chain characteristics (persistence length), Schick, and C. Seidel with whom we have
ionic strength of the solutions, and surface been working on PE problems. We ben-
characteristics. The shape and size of the efitted from discussions with G. Ariel, Y.
adsorbing layer is, in many instances, gov- Burak, and M. Ullner. One of us (DA)
erned by a delicate balance of competing would like to acknowledge partial sup-
mechanisms of electrostatic and nonelec- port from the Israel Science Foundation
trostatic origin. In some cases, it is found funded by the Israel Academy of Sciences
that the adsorbing PE layer is flat and and Humanities – Centers of Excellence
compressed, while in other cases, it is Program and the Israel–US Binational
coiled and extended. Yet, in other situa- Science Foundation (BSF) under grant
tions, the PEs will not adsorb at all and no. 98-00429.
320 2 Electrochemical Double Layers
55. T. A. Witten, P. Pincus, Europhys. Lett. 1987, 82. I. Borukhov, D. Andelman, H. Orland in
3, 315. Short and Long Chains at Interfaces (Eds.:
56. J.-L. Barrat, J.-F. Joanny, J. Phys. II (France) J. Daillant, P. Guenoun, C. Marques et al.),
1994, 4, 1089. Edition Frontieres, Gif-sur-Yvette, 1995,
57. M. J. Stevens, K. Kremer, J. Chem. Phys. pp. 13–20.
1995, 103, 1669. 83. J. F. Joanny, Eur. Phys. J. B 1999, 9, 117.
58. M. J. Stevens, K. Kremer, J. Phys. II 84. R. R. Netz, J. F. Joanny, Macromolecules
(France) 1996, 6, 1607. 1999, 32, 9013.
59. H. Schäfer, C. Seidel, Macromolecules 1997, 85. O. V. Borisov, E. B. Zhulina, T. M. Bir-
30, 6658. shtein, J. Phys. II (France) 1994, 4, 913.
60. M. J. Stevens, S. J. Plimpton, Eur. Phys. J. B 86. T. Odijk, Macromolecules 1983, 16, 1340.
1998, 2, 341.
87. T. Odijk, Macromolecules 1984, 17, 502.
61. M. A. Cohen Stuart, J. Phys. (France) 1988,
88. A. C. Maggs, D. A. Huse, S. Leibler, Euro-
49, 1001.
phys. Lett. 1989, 8, 615.
62. M. A. Cohen Stuart, G. J. Fleer, J. Lyklema,
89. G. Gompper, T. W. Burkhardt, Phys. Rev. A
et al., Adv. Colloid Interface Sci. 1991, 34,
477. 1989, 40, 6124.
63. Y. Fang, J. Yang, J. Phys. Chem. B 1997, 101, 90. G. Gompper, U. Seifert, J. Phys. A 1990, 23,
441. L1161.
64. J. O. Rädler, I. Koltover, T. Salditt et al., 91. R. Bundschuh, M. Lässig, R. Lipowsky, Eur.
Science 1997, 275, 810. Phys. J. E 2000, 3, 295.
65. T. Salditt, I. Koltover, J. O. Rädler et al., 92. V. Yamakov, A. Milchev, O. Borisov et al., J.
Phys. Rev. Lett. 1997, 79, 2582. Phys.: Condens. Matter 1999, 11, 9907.
66. B. Maier, J. O. Rädler, Phys. Rev. Lett. 1999, 93. D. H. Napper, Polymeric Stabilization of
82, 1911. Colloidal Dispersions, Academic Press, New
67. H. von Berlepsch, C. Burger, H. Dautzen- York, 1983.
berg, Phys. Rev. E 1998, 58, 7549. 94. T. Wallin, P. Linse, Langmuir 1996, 12, 305.
68. K. de Meijere, G. Brezesinski, H. Möhwald, 95. T. Wallin, P. Linse, J. Phys. Chem. 1996, 100,
Macromolecules 1997, 30, 2337. 17 873.
69. G. Decher, J. D. Hong, J. Schmitt, Thin 96. T. Wallin, P. Linse, J. Phys. Chem. B 1997,
Solid Films 1992, 210/211, 831. 101, 5506.
70. R. von Klitzing, H. Möhwald, Langmuir 97. E. M. Mateescu, C. Jeppesen, P. Pincus, Eu-
1995, 11, 3554. rophys. Lett. 1999, 46, 493.
71. R. von Klitzing, H. Möhwald, Macro- 98. R. R. Netz, J. F. Joanny, Macromolecules
molecules 1996, 29, 6901. 1999, 32, 9026.
72. G. Decher, Science 1997, 277, 1232. 99. K. K. Kunze, R. R. Netz, Phys. Rev. Lett.
73. M. Lösche, J. Schmitt, G. Decher et al., 2000, 85, 4389.
Macromolecules 1998, 31, 8893. 100. T. D. Yager, C. T. McMurray, K. E. van
74. F. Caruso, K. Niikura, D. N. Furlong et al.,
Holde, Biochemistry 1989, 28, 2271.
Langmuir 1997, 13, 3422.
101. F. von Goeler, M. Muthukumar, J. Chem.
75. E. Donath, G. B. Sukhorukov, F. Caruso
Phys. 1994, 100, 7796.
et al., Angew. Chem., Int. Ed. 1998, 110, 2323.
102. E. Gurovitch, P. Sens, Phys. Rev. Lett. 1999,
76. G. B. Sukhorukov, E. Donath, S. A. Davis
et al., Polym. Adv. Technol. 1998, 9, 759. 82, 339.
77. F. Caruso, R. A. Caruso, H. Möhwald, Sci- 103. P. Peyser, R. Ullman, J. Polym. Sci. A 1965,
ence 1998, 282, 1111. 3, 3165.
78. M. Muthukumar, J. Chem. Phys. 1987, 86, 104. M. Kawaguchi, H. Kawaguchi, A. Taka-
7230. hashi, J. Colloid Interface Sci. 1988, 124, 57.
79. X. Chatellier, J.-F. Joanny, J. Phys. II 105. J. Meadows, P. A. Williams, M. J. Garvey
(France) 1996, 6, 1669. et al., J. Colloid Interface Sci. 1989, 132, 319.
80. R. Varoqui, J. Phys. (France) II 1993, 3, 106. R. Denoyel, G. Durand, F. Lafuma et al., J.
1097. Colloid Interface Sci. 1990, 139, 281.
81. I. Borukhov, D. Andelman, H. Orland, Eu- 107. J. Blaakmeer, M. R. Böhmer, M. A. Cohen
rophys. Lett. 1995, 32, 499. Stuart et al., Macromolecules 1990, 23, 2301.
322 2 Electrochemical Double Layers
The adsorption of the ith solution com- However, the values µA and µS are
ponent is defined as a phenomenon that not independent, they are interrelated by
manifests itself in the difference of this Gibbs-Duhem equation [2, 3]:
component concentration in the vicinity
of the interface (ci ) and in the solu- NA dµA + NS dµS = 0 (3)
(b)
tion bulk (ci ). If the concentration of where NA and NS = 1 − NA are the mole
the ith component becomes higher than fractions of A and S. It follows from (2)
(b)
ci when the distance x decreases (as we and (3) that
approach the interface), its adsorption is
positive; in the opposite case, it is negative S N A
dγ = − A − dµA (4)
(curves 1 and 2, respectively, in Fig. 1). NS
The quantity characterizing the re-
versible adsorption of the ith component is The combination in parentheses defines
its surface excess i defined by the Gibbs the so-called relative surface excess:
adsorption isotherm [1] at constant tem- (S) S N A
A = A − (5)
perature (T = const) and pressure (p = NS
const):
which can be determined from experimen-
dγ = − i dµi (1) tal data. Particularly, for the system under
i
consideration
The quantity γ is the work consumed in (S) ∂γ
increasing the interface by unit surface A = − (6)
∂µA T ,p
area at T = const and p = const under re-
versible conditions, and µi is the chemical In order to determine separately the val-
potential of the ith component. ues of A and S , the second equation
In order to clarify the physical meaning interrelating these values should be de-
of i , let us consider a binary system rived. For this purpose, i is defined as the
328 3 Specific Adsorption
1
Concentration
[mol]
S1
(b)
ci
S2
2
0
Coordinate x
difference between the amounts of compo- where MA and MS are the molar masses
nent i in two equal solution cylinders with of components. A simultaneous solution
unit cross section, the first cylinder being of Eqs. (5) and (9) results in the following
normal and adjacent to the interface, and expressions:
the second one being completely in the MS NS
(M) (S)
solution bulk. This definition remains am- A = A
MA NA + MS NS
biguous if we do not clarify the meaning
of ‘‘equal.’’ If both cylinders contain equal (M) (S) MA NS
S = −A (10)
numbers of solvent moles, then S = 0. MA NA + MS NS
As follows from Eq. (5), the surface excess Finally, if both cylinders are of equal
of A defined by imposing this condition volumes, and the molar volumes of the
(S)
represents A (relative surface excess). If components VA and VS do not change in
both cylinders contain equal numbers of the course of adsorption, Eq. (5) should be
moles of A and S, the equation supplemented by the equation
(V ) (V )
A
(N )
+ S
(N )
=0 (7) V A A + VS S =0 (11)
A
(N ) (S)
= A (1 − NA ) (V ) (S) VS N S
A = A
VA N A + VS N S
(N ) (S)
S = −A (1 − NA ) (8) VA N S
(V ) (S)
S = −A (12)
VA N A + VS N S
Next, for two cylinders of equal masses,
the Eq. (5) should be supplemented by As follows from Eqs. (8), (10), and
equation (12), for adsorption from diluted so-
(S) (N ) (M)
lutions (NA 1), A ≈ A ≈ A ≈
(M) (M) (V )
M A A + M S S =0 (9) A . Under these conditions, the physical
3.1 Introduction 329
meaning of the surface excess i can be specific properties, the difference appears
represented by the areas S1 and S2 en- to be not so dramatic as that which we
closed by the curve corresponding to ci found when going to typical d-metals
versus x dependence in the vicinity of (platinum group metals are the most
(b)
interface and the line ci = const (see thoroughly studied representatives).
shaded areas in Fig. 1). For positive ad- In aqueous solutions, platinum-group
sorption, i > 0 whereas, for the negative metals adsorb hydrogen and oxygen and
adsorption, i < 0. In agreement with this can be classified as the so-called ‘‘per-
interpretation of surface excess, its dimen- fectly polarizable electrodes’’ in terms of
sionality is mol m−2 or mol cm−2 , which Planck [4]. For perfectly polarizable elec-
coincides with the dimensionality of the trode, one portion of electric charge de-
surface concentration. The latter can be livered to its surface is expended in the
defined as the number of moles of the change of electrode free charge, when an-
ith component that find themselves in a other portion is expended in the oxidation
monomolecular layer of substance closest of hydrogen adatoms or in the generation
to the surface. In contrast to the surface of oxygen adatoms. The complete quantity
excess, the surface concentration is ei- of electric charge delivered to the surface
ther positive or zero. For calculating the is said to be expended in the change of
surface concentration from experimental total electrode charge. It means that the
i values, it is essential to know how ci state of electrode is uniquely determined
changes with x. The problem may be sub- by its initial state and the change of its total
stantially simplified, if the surface excess charge.
of the ith component is located inside In the text that follows, we start with
an adsorption monolayer, and its bulk the adsorption phenomena on ideally
concentration is sufficiently low (Ni 1). polarizable electrodes, and then pass to
Under these conditions, typical for ad- perfectly polarizable electrodes.
sorption of many organic substances on Adsorption of various ions and mole-
electrodes, the surface concentration prac- cules on electrodes is favored by their
tically coincides with the surface excess. interaction with the electrode surface, and
Originally, the phenomenon of specific also by the squeezing of adsorbate species
adsorption was considered in connection out of the solution bulk. The latter phe-
with studies on mercury electrodes, the nomenon is induced by the difference in
latter being ideally polarizable in a wide free solvation energies G between the
potential range. It should be mentioned surface layer and in the bulk of the so-
that for ideally polarizable electrode, the lution. For positive adsorption, G < 0,
total quantity of electric charge delivered which, in its turn, can be caused by a de-
to its surface is expended in the change of crease in enthalpy (H < 0) and by an
the free electrode charge, that is, the charge increase of entropy (S > 0) as the adsor-
defined by excess or lack of electrons. Quite bate species pass from the solution bulk
a number of s,p-metals (Bi, Ga, Cd, Pb, to the surface. In principle, electrostatic
Sn, Sb, In, Tl) can also exhibit similar forces can solely induce the interaction
behavior in a certain potential interval, and of ions with the electrode surface: the
are sometimes integrated into the group of coulombic attraction and repulsion should
mercury-like metals. Although any metal result in positive and negative adsorption,
of this group undoubtedly exhibits certain respectively. In addition, at a direct contact
330 3 Specific Adsorption
of an ion with the metal surface (in the Gouy was the first to mention the
absence of separating solvent molecules), specific behavior of anions on mercury [5];
the ion–metal interaction can include however, the term ‘‘specific adsorption’’
other contributions: image forces (inter- was introduced by Stern [6] who made an
action with the charge of the opposite attempt to treat specific adsorption quan-
sign induced in the metal phase) and a titatively. Generally, any ion can be specif-
covalent bonding, which is sometimes con- ically adsorbed under definite conditions,
sidered in the terms of donor–acceptor and only under certain circumstances can
interaction (which can result in a partial specific adsorption be ignored in the first
charge transfer between adsorbed ions and approximation.
metal). It is conventionally supposed that the
If the ion–metal interaction is com- cations like Li+ , Na+ , and K+ , and anions
pletely electrostatic and the squeezing out like F− , having small sizes and surrounded
is absent, the electrolyte is surface-inactive. by dense solvation shells (also complex
In the opposite case, specific adsorp- anions like HF2 − , PF6 − , BF4 − ), are not
tion takes place, and the corresponding adsorbed specifically. Also, it is assumed
ion is surface-active. Figure 2 shows the that specific adsorption of anions is more
electrocapillary curves (dependences of γ pronounced than that of cations. These
on electrode potential E) for a mercury suppositions stemmed from the studies on
electrode in 0.01, 0.1, and 1 M aqueous a mercury electrode in aqueous solutions
solutions of a surface-inactive electrolyte and cannot be generalized a priori to other
(curves 1, 2, and 3) and the corresponding systems.
γ versus E-curves in solutions containing Historically, the term ‘‘specific adsorp-
surface-active anions (1 , 2 and 3 ). tion’’ goes back to the studies of adsorption
γ 0max
400
1
Surface tension
2
375 3
[mN m−1]
350
1
2 1′
325 3 3′
2′
of inorganic and organic ions, but it is interface, which is equal to the pzc
worthwhile to apply it also to adsorp- shift and increases with − . The spe-
tion of neutral organic molecules. The cific adsorption of cations shifts the pzc
reason lies not only in the squeezing towards positive values. For simultane-
out from solution, which occurs in both ous specific adsorption of anions and
cases, but also in the existence of spe- cations, the magnitude and the sign of
cific interactions with the metal. That is, the pzc shift can change with relative ionic
the molecules of aromatic compounds are concentrations.
bound to the electrode via the so-called The examples of pzc shifts induced by
π-electron interaction [7–10]; functionally specific adsorption of anions on mercury
substituted molecules (thiolates, amines, in aqueous solutions are presented in
and so forth) assume certain interfacial Table 1. As demonstrated by Frumkin [13],
orientations owing to specific interaction the specific features of metal–ion interac-
of substituting groups with the metal, par- tion can be elucidated by comparing the
ticularly mercury [11]. pzc shifts and the changes of surface po-
According to the Lippman equation [12] tential at the solution/air interface. The
adsorption of ion at the latter interface re-
∂γ
σM = − (13) sults only from squeezing-out effect. For
∂E T ,p,µ example, the change of surface potential
when going from 0.01 M KCl to 2 M KI
the charge density on the metal surface
solution reaches only 52 mV [14], while
σM is determined by the slope of the
the pzc shift for Hg/1 M KI (in compari-
electrocapillary curve. The maximum of
son with pzc in surface-inactive electrolyte
γ , E-curve corresponds to σM = 0; hence,
solution) is about 350 mV.
the position of this maximum determines
The pzc shift resulting from the ad-
the potential of zero charge (pzc). In
sorption of polar organic molecules is
surface-inactive electrolyte solutions, the
determined by the orientation of their
value of the pzc does not depend on
dipoles on the electrode surface.
concentration (Fig. 2). On the other hand,
with an increase in the concentration
of surface-active anion, the pzc shifts Tab. 1 The pzc and the pzc shift values for a
towards negative potentials, because of the mercury electrode in aqueous unimolar
solutions as given by Frumkin [13]
increasing specific adsorption of anions.
Indeed, the condition of electrical neu-
Electrolyte pzc (MCE) [V] pzc shifts [V]
trality in the 1,1-electrolyte solution is
The specific adsorption reduces the obtain the following expression from (17):
work of surface formation because of the
∂E0 ∂(+ + − )
difference in the energetic states of solvent = −RT
molecules on the surface and in the bulk ∂ ln a± σM ∂σM a±
of the solution, and also because of the (18)
lateral interactions in the adsorbed layer. At the pzc, according to Eq. (14), − = + ,
For neutral organic molecules, as a rule, and from Eq. (18) one obtains
the decrease in γ is observed only in the
vicinity of the pzc, because at high |σM | the ∂E0 ∂−
= −2RT
organic molecules are squeezed out by the ∂ ln a± σM =0 ∂σM a±
solvent of a higher permittivity. (19)
The electrocapillary theory developed by Equation (19) relates the dependence
Frumkin [15, 16] makes it possible to de- of the pzc on the concentration of
termine not only the electrode charge and the electrolyte solution in the presence
pzc but also the relative surface excesses of specific adsorption (− = 0 when
of various ions by treating the γ , E-curves. σM = 0) and its variation with the electrode
Namely, in a binary 1,1-electrolyte solu- charge. The dependence of the pzc of
tion, when we use a reference electrode a mercury electrode on the logarithm
reversible with respect to anion (see above of KI concentration was used for the
in Electrode Potentials), first time for studying the iodide specific
adsorption in [17] and later was named
dγ = −σM dE− − 2RT + d ln a± (15) the Esin-Markov effect. As follows from
the model theories of the electric double
whereas, in the same solution with the
layer (see Sect. 3.2), the limiting slope of
reference electrode reversible in respect to
the aforementioned dependence should
cation,
tend to the value −RT /λF , where the
dγ = −σM dE+ − 2RT − d ln a± (16) coefficient λ(0 < λ ≤ 1) characterizes the
discrete nature of the charge of specifically
In the formulae given above, R is the adsorbed anions.
universal gas constant, and a± is the The combination of Eqs. (13), (14), and
mean activity of the electrolyte solution. (17) make it possible to calculate the de-
By combining Eqs. (15) and (16) and pendencies of − and + on E0 from the
introducing the notation dE0 = d(E+ + experimental γ , E-curves. Figure 3 shows
E− )/2, one obtains the corresponding dependencies of F −
and F + on E0 for 0.1 M solutions of
dγ = −σM dE0 − RT (+ + − ) d ln a± surface-inactive and surface-active elec-
(17) trolytes. The positive adsorption of cations
On the combined potential, scale intro- at E0 < 0 and of anions at E0 > 0 in
duced above the pzc value does not depend a former solution are induced only by
on the concentration of the surface- the coulombic attraction of ions to the
inactive electrolyte. Hence, it is convenient oppositely charged electrode surface. In
to refer the potential to this zero point (see contrast, the negative adsorption of cations
Fig. 2). at E0 > 0 and of anions at E0 < 0 is caused
The properties of the total differential by the repulsion from the surface of the
of a function of two variables, allow us to same sign. In 0.1 M KF solution at E0 = 0,
3.1 Introduction 333
30
1′
20
2′
[µC cm–2]
F Γi
10
1
0
Fig. 3 Components of electrode charge
2
associated with the surface excesses of
anions (1,1 ) and cations (2,2 ), as −0.6 –0.3 0.0 0.3
function of the potential of a mercury
electrode in 0.1 M aqueous solutions of Potential
KF (1,2); KI (1 2 ). [V]
334 3 Specific Adsorption
Certain model assumptions are neces- the diffuse layer, then the value of +
sary in order to reveal the surface con- determined from the experimental data
(2)
centration of specifically adsorbed ions in should be equal to + . Hence, we can
the total surface excess i . Usually, the calculate the value of ϕ2 from Eq. (21)
ionic component of the electrical double and, then, find the contribution of anions
(2)
layer (EDL) is assumed to consist of the residing in the diffuse part, − from
dense part and the diffuse layer separated Eq. (22). Finally, we determine the specific
by the so-called outer Helmholtz plane. (1)
adsorption of anions, − from the
Only specifically adsorbing ions can pen- (1)
experimental value of − = − + − .
(2)
etrate into the dense layer close to the This technique of calculating the surface
surface (e.g. iodide ions), with their elec- concentration of anions was pioneered
tric centers located on the inner Helmholtz by Grahame and Soderberg [21] using
plane. The charge density of these specif- the experimental dependencies of the
ically adsorbed ions σ1 is determined by differential capacitance C on the potential,
(1)
their surface concentration i . Namely, where C is given by
for single-charged anions: 2
∂σM ∂ γ
(1) C= =− (24)
σ1 = −F − (20) ∂E a± ∂E 2 a±
On the other hand, the surface excesses of In contrast to γ , the value of C can be mea-
(2)
all ions in the diffuse layer i are defined sured not only for liquid electrodes but also
by their coulombic interaction with the for solid ones. Usually, the specific adsorp-
net charge (σM + σ1 ). According to the tion of ions increases the capacitance of the
classical theory of the diffuse layer [18–20], EDL, whereas the adsorption of organic
the behavior of 1,1-electrolyte solution with molecules decreases it. In the framework
concentration c is described by: of a simplified model of the double layer as
a capacitor, this corresponds to varying the
(2) √ F ϕ2
F + = A c exp − − 1 (21) distance between the capacitor plates and
2RT decreasing the permittivity. In dilute solu-
(2) √ F ϕ2 tions of a surface-inactive 1,1-electrolyte,
F − = A c exp −1 (22)
2RT a capacitance minimum appears in the
vicinity of pzc (see curve 1 in Fig. 4). The
where A = (2RT ε0 ε)1/2 , ε is the permittiv- position of this minimum is determined
ity of solvent, ε0 = 8.854 × 10−12 F m−1 , by the potential E0 = 0. In solutions with
and ϕ2 is the potential of the outer relatively high (≥0.1 M) concentrations
Helmholtz plane referred to the solution of surface-inactive electrolytes, this min-
bulk and related to the value of (σM + σ1 ) imum of C, E0 -curve disappears, but the
by the formula pzc position still corresponds to E0 = 0
(curve 2 in Fig. 4). Numerical integration
2RT σM + σ1
ϕ2 = sinh−1 √ (23) of these C, E0 -curves
F 2A c E0
If we suppose that cations in solutions σM = C dE0 (25)
0
of lithium, sodium, and potassium salts
are surface-inactive, that is, at any given results in the corresponding σM , E0 -curves
E0 , their surface excesses are located in (see curves 1 and 2 in Fig. 5).
3.1 Introduction 335
Differential capacitance
0.1 M KI (3). 40
[µF cm–2]
2
20
1
0
−1.0 –0.5 0.0 0.5
Potential
[V]
0
Charge
−10
−20
−1.0 −0.5 0.0 0.5
Potential
[V]
0.5
0
–4 0 4 8
ln(βci(b))
(b)
bulk concentration ci [43]. According to charge σM , and the potential drop in the in-
Boltzmann formula, ner part of the EDL. The model applied for
describing the EDL determines the choice
(2) (b) −zi F ϕ2 of the independent variable. Namely, in
ci = ci exp (37)
RT the framework of Frumkin model [33] for
adsorption of neutral organic molecules,
According to Eq. (23), the value of ϕ2 , is ln β is a square function of E0 ; in terms
a function of the charge of specifically of Vorotyntsev’s model [44] for ionic ad-
adsorbed ions σ1 = zi F max θ, where max sorption, ln β is also a function of E0 ,
is the limiting value of their surface but substantially more complicated [45]; in
concentration corresponding to θ = 1. Grahame-Parsons model [46, 47] for ionic
Thus, the adsorption isotherm equation adsorption, the value of ln β is considered
for specifically adsorbed ions appears to be as a function of σM ; in contrast, in Kolo-
very complex. tyrkin–Alekseev–Popov model [48], put
At low surface coverage, when ln(1 − forward later in Refs. [49, 50], and also
θ) ≈ −θ, the Frumkin isotherm (34) is in a similar Nikitas’ model [40, 41], ln β is
rearranged into a virial isotherm: a linear or square function of the potential
(b) (1) (1) drop in the inner layer: ϕ = E0 − ϕ2 .
ln βν + ln ci = ln i + 2Bν i (38) As demonstrated in Refs. [51, 52], in
the case of reversible specific adsorption
where
of Cl− , Br− , and I− anions on various
ln βν = ln β + ln max ;
(1)
i = θmax mercury-like electrodes from different
solvents, the values of ln β at σM = 0 and
1 − 2a (b)
EσM =0 at ci = const can be correlated
Bν = (39)
2max with a linear combination of donor (DN)
As mentioned in the preceding text, free and acceptor (AN) numbers of solvents:
energy of adsorption of ions or molecules, y = k0 − k1 DN − k2 AN (40)
and, correspondingly, the value of ln β
(see Eq. 35) depends on one of electrical The quantity of y in Eq. (40), which is
variables: electrode potential E, electrode equal to ln β at σM = 0 [51] or to EσM =0
3.1 Introduction 339
(b)
at ci = const [52], is the measure of value (namely, for gallium in aprotic sol-
specific adsorption energy of the ith vents), or, finally, remain independent of
ion, and k0 , k1 , and k2 are constants the anion nature (e.g. on In-Ga/acetonitrile
independent of the nature of solvent interface).
for the given metal and ion. In its Let us consider the specific adsorption
physical meaning, the value of k0 should phenomena on the d-metals.
represent the metal-anion bond energy in As was mentioned in the preceding text,
the absence of solvent. the main difference between ‘‘perfectly
In the physical sense, Eq. (40) is based on polarizable’’ and ‘‘ideally polarizable’’ elec-
the supposition that the greater the specific trodes consists in the fact that the potential
adsorption energy of anions should be, the of the former is determined by the pres-
smaller their solvation energy (presented ence of a certain redox system,
by the value of AN) and the weaker
the bond of solvent molecules with the Ox + nē ⇐⇒ Red (41)
uncharged electrode surface (presented
by the value of DN). This is induced by and the state of the system is unam-
the donor–acceptor nature of ion–dipole biguously determined by its initial state
interactions, and also by the interaction of and a quantity of charge passed across
metal with anions and solvent molecules in the interface. For platinum metals in
the course of their competitive adsorption aqueous solutions of surface-inactive elec-
on the uncharged electrode surface. trolytes, Ox1 = H+ , Red1 = Hads (n = 1);
In Ref. [52], the Eq. (40) was com- Ox2 = Oads + 2H+ , Red2 = H2 O (n = 2).
pared with the experimental values of A consistent thermodynamic theory of per-
(b)
y = EσM =0 at ci = 0.1 M, which re- fectly polarizable electrodes was developed
sulted from the specific adsorption of Cl− , in [53–60] on the basis of the initial theory
Br− , and I− anions on the electrodes made of amalgam [13, 61, 62] and platinum hy-
of mercury, gallium, and eutectic alloys In- drogen [63] electrodes. In this case, Gibbs
Ga (16.4 at % In) and Tl-Ga (0.02 at % Tl) equation at T , p = const takes the follow-
in water, acetonitrile, dimethylformamide, ing form:
N -methylformamide, and dimethyl sulfox-
ide. The correlation analysis demonstrated dγ = −Ox dµOx − Red dµRed
that the bond energy in the absence of sol-
vent (k0 ) increases in the sequence I− < − j dµj (42)
Br− < Cl− , in agreement with the row j
for dissociation energy of gaseous mono-
halogenides of mercury and gallium. At in which µOx , µRed , Ox , Red are the
the same time, for various donor–acceptor chemical potentials and surface excesses
properties of solvent, the energy of specific of Ox and Red components, µj , j are the
adsorption on the metal/solution interface chemical potentials and surface excesses
at σM = 0 can either increase in a sequence of other solution components.
Cl− < Br− < I− (this ‘‘usual’’ sequence It follows from Eq. (42) that, when
was observed in aqueous solutions on mer- µj = const, the function γ is a 3D surface,
cury and mercury-like metals), or decrease its intersects with the planes µOx = const
in the same sequence in parallel with k0 and µRed = const being the characteristic
340 3 Specific Adsorption
electrocapillary curves described by Lipp- the Nernst Equation for the corresponding
man equations: redox system.
In order to divide the values of F Ox
∂γ and F Red into bound charges (AOx
= +F Red (43)
∂E T ,p,µj ,µOx and ARed ) and free charge σM , certain
assumptions going beyond the frames of
∂γ
= −F Ox (44) thermodynamics are necessary.
∂E T ,p,µj ,µRed In a certain case of a platinum hydro-
gen electrode, on which the equilibrium
The quantities of F Ox and F Red are Hads ⇔ H+ + ē is established, at µj =
called the total charges of the surface at the const,
corresponding choice of values remaining
constant, because these values characterize dγ = −H dµH − H+ dµH+ (45)
the quantities of electricity required for
increasing the surface by a unit area where
without introducing any changes in the
composition of both contacting phases. F H = −Q1 = AH − σM (46)
It is evident that for redox-systems F H+ = Q2 = σM + AH+ (47)
more complex than (41), Gibbs adsorp-
tion equation should also become more that is, total charges F H and F H+ are
complicated. The number of characteris- the algebraic sum of bound (AH , AH+ )
tic Lippman Equations should be equal to and free (σM ) charges. At an excess of
the number of independent variables in surface-inactive electrolyte, AH+ can be
1
0
2
1
Surface tension
3
−200 2
[mN m–1]
4
−400
3 5
equated to zero, and F H+ = σM , that is, allows us to describe the effect of specif-
under certain conditions, F H+ can be ically adsorbed ions on the hydrogen
considered as a free surface charge. adsorption. Inasmuch as ionic adsorp-
The electrocapillary curves correspond- tion affects the hydrogen and oxygen
ing to Eq. (43) are called the electrocap- adsorption, an attempt should be made
illary curves of the first kind, and those in using these effects for determination
corresponding to the Eq. (44) are called of ionic adsorption on platinum group
the electrocapillary curves of the second metals [67].
kind [60, 64]. Figure 7 shows the examples In contrast to mercury-like metals,
of curves of these kinds for Pt electrode. the adsorption of ions on platinum
This figure demonstrates also the effects metals appears to be relatively slow [68],
the solution pH and the anion adsorp- the adsorption rate νads exponentially
tion on electrocapillary curves. It should decreasing with the surface coverage θ
be stressed that pH-dependent adsorption in accordance with Roginskii-Zel’dovich
behavior of Pt hydrogen electrode is a typ- equation
ical feature of hydrogen-adsorbing metals,
(b)
it is governed by Eq. (45), for the latter νads = Kads ci exp(−αf θ) (49)
no assumptions are made on the specific
adsorption of OH− ions. where Kads and α are constants (0 < α <
Therefore, for Pt hydrogen electrode, two 1) and f is the inhomogeneity factor [67].
pzc values should be introduced, which For the middle coverage of platinum
correspond to H = 0 and H+ = 0 (and metals by ions, a logarithmic dependence
to the maxima of the electrocapillary curves of θ on adsorption time tads is observed:
of the 1st and 2nd kinds), respectively. 1
When the concentration of a specifically θ = const1 + ln tads (50)
αf
adsorbed ion i is sufficiently low as
compared with the supporting-electrolyte These data confirm that the adsorption
concentration, so that it can be changed of ions on platinum metals follows the
without changing the chemical potential general tendencies of adsorption on the en-
of supporting ions, the Gibbs Equation ergetically inhomogeneous surfaces. This
takes the following form: conclusion agrees also with the results of
analyzing the isotherms found for ionic
dγ = −H dµH − i dµi (48) adsorption on platinum [68], which follow
where µi is a chemical potential of ad- the logarithmic law (Temkin isotherm):
sorbed ion, and i –its surface excess. 1
From this equation, the quantitative re- θ = const2 + ln c(b) (51)
f
lationships for determination of i versus
electrode potential dependencies by using and also with the data on the isotherms of
the electrode charging curve and the shifts displacement of one ion by another. The
of electrode potential with the concentra- latter proved to be helpful when comparing
tion of adsorbed ions (so-called adsorption the surface activities of ions. Particularly,
shift of potential) [65]. The measurements it was found (Fig. 8) that Cs+ ions
of this sort gave the results that agree with displace Na+ ions adsorbed on platinum
direct data on ionic adsorption obtained by at substantially lower bulk concentration
radiotracer techniques [66]. Equation (48) than that corresponding to the opposite
342 3 Specific Adsorption
0
−5 −4 −3 −2 −1 0
log c Na2SO4, log c Cs2SO4
[mol l−1]
6
[µC cm−2]
4 3
F ΓNa+
2
Fig. 9 Dependencies of Na+ 2
adsorption on the potential of a 4
platinized platinum electrode in the
1
following solutions: 0.0005 M 0
H2 SO4 + 0.0015 M Na2 SO4 (1); 0 0.2 0.4 0.6 0.8 1.0
0.001 M HC1 + 0.003 M NaC1 (2);
0.001 M HBr + 0.003 M NaBr (3); Potential
0.0005 M H2 SO4 + 0.003 M NaI [66]. [V, RHE]
344 3 Specific Adsorption
Br− < I− . Thus, the sequences for specific solution of this problem, the conclusion
adsorbability and superequivalent adsorp- on the pzfc (potential of zero free charge)
tion disagree. This phenomenon can be shift induced by specific adsorption can
explained under a supposition that the change. As demonstrated in Ref. [84],
charge-transfer degree increases with the the pzfc of a platinum electrode can
adsorbability. The approximately constant shift substantially towards more positive
or even slightly decreasing adsorption of and more negative potential values when
Na+ observed with the growth of potential I− and Tl+ , respectively, are adsorbed.
indicates that the charge-transfer degree Thus, the direction of the pzfc shift is
increases as the potential shifts towards opposite as compared with that induced
more positive values. by the adsorption of the same ions on
The charge transfer, which accompanies mercury-like metals. Hence, under the
the adsorption processes, can affect the conditions of charge transfer, we can
latter because the ionic charges are discuss only conventional zero free charge
spread over the surface in a peculiar potentials, because, actually, the nature of
manner. Consequently, the Esin-Marov the electrode surface changes.
effect can be substantially weakened. Complete charge transfer (upd) is possi-
A direct determination of Esin-Markov ble when the support-adsorbate interaction
coefficient on platinum group metals [54] is stronger than the lateral interaction in
demonstrates no effect altogether, which the adlayer. Usually, upd induces mono-
can be explained by a pronounced charge layer coverage; however, for certain sys-
transfer from adsorbed anions. According tems the formation of second and even
to the thermodynamic theory of the third layer turns out to be possible, and
platinum hydrogen electrode, there are two these layers do not resemble the corre-
Esin-Markov coefficients corresponding to sponding bulk deposits.
the constant values of free and total charges Consideration of perfectly polarizable
respectively. When comparing the data for electrodes presented in the preceding text
Pt and Hg electrodes, one should use the is related to equilibrium conditions (E =
former coefficient. const). When dealing with nonpotentio-
Charge transfer manifests itself also static modes (linear, stepwise, sinusoidal,
by a slow establishment of adsorption etc.), one always observes not only double-
equilibria mentioned in the preceding layer capacity but also pseudocapacity
text, a slow exchange between adsorbed induced by the changes of surface cov-
ions and solution, temperature effects, erage with time in the course of reactions
and the adsorption isotherms of strongly (52, 53). Conway, Gileadi and coworkers
surface-active ions. The latter isotherms intensively studied this problem Refs. [85,
demonstrate a slope close to that known 86].
for the adsorption of neutral and low- An application of the thermodynamic
polar species, namely, Hads , for which the theory of Pt hydrogen electrode to the re-
Temkin isotherm is applicable [67]. versible adsorption of organic substances
For charge transfer from adsorbed ions leads to a conclusion [87] that the max-
to the metal surface, a problem arises as to imum adsorption of neutral molecules
whether this charge should be attributed should be observed in the potential region,
to the ‘‘electrode’’ part of the double-layer where the adsorption of both hydrogen
or to its ionic part. Depending on the and oxygen is low (so-called double-layer
3.1 Introduction 345
region). However, this conclusion was ions, makes it possible to clarify the roles
never checked, because for the majority the ion solvation and the metallic surface
of adsorbates, the destruction of organic solvation play in the interfacial structure,
molecules in the course of adsorption and, particularly, understand the mecha-
was found [88], which was accompanied nism of solvophilicity of metals.
by their hydridization, dehydridization, as- In the potential region where forma-
sociation of adsorption products, and so tion of chemisorbed oxygen layers and
forth. Hence, their adsorption appears to phase oxides takes place, the adsorption
be irreversible, and the composition of phenomena become irreversible and ex-
adsorbed species differs from that in the tremely complicated, because almost all
(ads) (b)
solution bulk. In this case, µ̄i = µ̄i ; solution components can take part in
no exchange between adsorbed layer and electrode reactions and undergo electro-
solution occurs, the species cannot be chemical transformations. The adsorption
desorbed in pure solvent, and no unam- peculiarities under these conditions differ
biguous relationship exists between the strongly from those known for d-metals
bulk and surface concentrations of adsor- in the absence of oxygen adsorption. For
bates. Certain inorganic ions (like I− in example, the adsorbability sequence for
acid media) are also adsorbed irreversibly halide ions on Pt at E > 1.5 V (RHE) trans-
on platinum. forms into I− < Br− < Cl− < F− [93]. It
In parallel with complications induced is assumed that anions penetrate into
by irreversible adsorption, new approaches the oxide layer with parallel dehydration,
to the experimental studies of surface and their susceptibility to injection corre-
excesses are developed. Namely, the adsor- lates with the reciprocal crystallographic
bate’s state and coverage usually remain radia. The adsorption versus potential de-
unchanged when the potential is switched pendence appears to be nonmonotonous
off and the electrode withdrawn from so- and demonstrates a number of maxima;
lution. Moreover, some adsorbed layers moreover, in a number of cases, the max-
remain stable even under UHV conditions, ima of anion adsorption correspond to the
and a number of highly informative ex situ maxima of cation adsorption [94–98]. It
spectroscopic techniques were success- was also found that organic substances
fully applied to the studies of adsorption could adsorb at high anodic potentials on
on d-metals [89–91]. platinum group metals [99]. The mecha-
Modeling of electrochemical interfaces nisms of these phenomena still remain
under UHV conditions supplements the unclear.
aforementioned approaches. For example, In principle, the theory of reversible elec-
it exploits the controlled modification of trodes can serve as the basis for developing
the metal/vacuum interface by the various the thermodynamic approach to the sur-
components typical of electrochemical in- face electrochemistry of oxide materials
terfaces (ions, solvent molecules) and reg- with metallic conductivity (iridium, ruthe-
istration of infrared reflection-absorption nium, tin dioxides, and so on) [100]. In
spectra along with work function mea- addition, the properties of these interfaces
surements [92]. A gradual increase in the can be considered within the framework of
surface coverage by solvent molecules, three models: the classical model of bound
which changes from a submonolayer up sites [101], which can be modified by con-
to multilayers at various doses of adsorbed sidering several types of surface groups
346 3 Specific Adsorption
(MUSIC model) [102, 103]; various modi- and nanoheterogeneity should be stressed
fications of Grahame-Parsons model [104]; because rather unexpected features of
and the model of nonlocal electrostat- adlayers result from comparable sizes of
ics [105]. On oxide surfaces, functional adsorbate molecule and surface fragment.
groups are present, which can be ionized One of the aspects of this problem
reversibly upon their contact with the elec- is the effect of long-living metastable
trolyte solution: surface defectiveness that is responsible
for ‘‘adsorption traps,’’ the surface centers
MOH2 + ⇐⇒ MOH + H+ (56) with anomalously high adsorption ability.
− + Specific adsorption is rated among the
MOH ⇐⇒ MO + H (57)
most important phenomena in surface
MOH2 + . . . A− ⇐⇒ MOH + H+ + A− science. Its unique importance for electro-
(58) chemistry is dictated by the role of specific
adsorption of reactants and products, and
MOH + C+ ⇐⇒ MO− . . . C+ + H+ also of inert solution components in the
(59) kinetics of electrode reactions. Adsorption
where C+ and A− are cations and anions can change not only reaction rates but also
of the electrolyte. In this case, the specific the mechanisms.
adsorption of certain components can be
considered as an increase in the binding References
degree of this component with the surface
functional groups [106]. 1. J. W. Gibbs, The Collected Works of J. W.
For oxide/solution interfaces, a pro- Gibbs, Longmans, New York, 1931, p. 219,
nounced correlation was revealed be- Vol. 1.
2. E. A. Guggenheim, N. K. Adam, Proc. R.
tween the type of organic compound and Soc. London 1933, A139, 218–236.
their adsorption [107]. Such selectivity was 3. A. W. Adamson, Physical Chemistry of Sur-
noted for adsorption on metals and can faces, John Wiley & Sons, New York, 1976,
point to the chemisorption nature of the p. 68.
4. M. Planck, Ann. Phys. 1891, 44, 385–428.
metal–adsorbate interaction [108].
5. G. Gouy, Compt. Rend. 1908, 146,
A modern stage in the development of 1374–1376.
notions of specific adsorption is masked 6. O. Stern, Z. Elektrochem. 1924, 30, 508–516.
by a changeover to describing this phe- 7. M. A. Gerovich, Doklady AN SSSR 1954, 96,
nomenon on a microscopical level, using 543–546.
8. E. Blomgen, J. O’M. Bockris, J. Phys. Chem.
computational approaches to simulation 1959, 63, 1475–1484.
of adsorbed layers. We can expect a rapid 9. M. A. Gerovich, Doklady Akad. Nauk SSSR
progress in these directions, because the 1955, 105, 1278–1281.
information on the microscopic structure 10. E. Blomgen, J. O’M. Bockris, C. Jesch,
J. Phys. Chem. 1961, 65, 2000–2010.
of adlayers becomes progressively more
11. H. O. Finklea in Electroanalytical Chemistry
available because of studies on single- (Eds.: A. J. Bard, I. Rubinstein), Marcel
crystal electrodes and the results of physi- Dekker, New York, 1996, pp. 109–335.
cal surface science techniques. 12. G. Lippmann, Ann. Chim. Phys. 1875, 5,
There is a number of novel problems in 494–549.
13. A. N. Frumkin, Elektrokapillyarnye yavleniya
the field of specific adsorption induced i elektrodnye potentsialy (Electrocapillary
by fast progress of material science. Phenomena and Electrode Potentials), Ti-
Particularly, the role of nanostructure pographiya Sapozhnikova, Odessa, 1919.
3.1 Introduction 347
14. A. Frumkin, Z. Phys. Chem. 1924, 109, 41. P. Nikitas, Electrochim. Acta 1996, 41,
34–48. 2159–2170.
15. A. Frumkin, Z. Phys. Chem. 1923, 103, 42. R. Fowler, E. A. Guggenheim, Statistical
43–70. Mechanics, Cambridge University Press,
16. A. Frumkin, Zh. Fiz. Khim. 1956, 30, Cambridge, 1956.
2066–2069. 43. R. de Levie, J. Electrochem. Soc. 1971, 118,
17. O. A. Esin, B. F. Markov, Acta Physicochim. 185C–192C.
URSS 1939, 10, 353–364. 44. M. A. Vorotyntsev, J. Res. Inst. Catalysis,
18. G. Gouy, J. Phys. Radium 1910, 9, 457–466. Hokkaido Univ. 1982, 30, 167–177.
19. D. L. Chapman, Philos. Mag. 1913, 25, 45. B. Damaskin, U. Palm, M. Salve, J. Elec-
475–483. troanal. Chem. 1987, 218, 65–76.
20. D. C. Grahame, Chem. Rev. 1947, 41, 46. D. C. Grahame, R. Parsons, J. Am. Chem.
441–501. Soc. 1961, 83, 1291–1296.
21. D. C. Grahame, B. A. Soderberg, J. Chem. 47. J. M. Parry, R. Parsons, Trans. Faraday Soc.
Phys. 1954, 22, 449–460. 1963, 59, 241–256.
22. H. D. Hurwitz, J. Electroanal. Chem. 1965, 48. Ya. M. Kolotyrkin, Yu. V. Alekseev, Yu. A.
10, 35–41. Popov, J. Electroanal. Chem. 1975, 62,
23. E. Dutkiewicz, R. Parsons, J. Electroanal. 135–149.
Chem. 1966, 11, 100–110. 49. B. Damaskin, S. Karpov, S. Dyatkina et al.,
24. J. Lipkowski, C. Nguen Van Huong, C. Hin- J. Electroanal. Chem. 1985, 189, 183–194.
nen et al., J. Electroanal. Chem. 1983, 143, 50. B. B. Damaskin, O. A. Baturina, Russ. J.
375–396. Electrochem. 1999, 35, 1186–1193.
25. L. Stolberg, J. Lipkowski, D. E. Irish, J. Elec- 51. W. R. Fawcett, A. Motheo, J. Electroanal.
troanal. Chem. 1987, 238, 333–353.
Chem. 1994, 375, 319–338.
26. A. Hamelin, S. Morin, J. Richter et al.,
52. B. B. Damaskin, Russ. J. Electrochem. 1994,
J. Electroanal. Chem. 1989, 272, 241–252.
30, 91–94.
27. L. Stolberg, J. Lipkowski, D. E. Irish, J. Elec-
53. A. Frumkin, N. Balashova, V. Kazarinov, J.
troanal. Chem. 1990, 296, 171–189.
Electrochem. Soc. 1966, 113, 1011–1018.
28. L. Stolberg, J. Lipkowski, D. E. Irish, J. Elec-
54. A. Frumkin, O. Petry, A. Kossaya et al., J.
troanal. Chem. 1992, 322, 357–372.
Electroanal. Chem. 1968, 16, 175–191.
29. A. Hamelin, S. Morin, J. Richter et al.,
J. Electroanal. Chem. 1990, 285, 249–262. 55. A. Frumkin, O. Petry, Electrochim. Acta
30. A. Hamelin, S. Morin, J. Richter et al., 1970, 15, 391–403.
J. Electroanal. Chem. 1991, 304, 195–209. 56. A. N. Frumkin, O. A. Petrii, B. B. Damas-
31. W. R. Fawcett, J. Electroanal. Chem. 1972, kin, J. Electroanal. Chem. 1970, 27, 81–100.
39, 474–477. 57. B. Grafov, E. Pekar, O. Petrii, J. Electroanal.
32. A. Frumkin, Z. Phys. Chem. 1925, 116, Chem. 1972, 40, 179–186.
466–484. 58. A. N. Frumkin, B. B. Damaskin, O. A. Petrii,
33. A. Frumkin, Z. Physik 1926, 35, 792–802. J. Electroanal. Chem. 1974, 53, 57–65.
34. R. Parsons, J. Electroanal. Chem. 1964, 7, 59. A. N. Frumkin, Potentsialy nulevogo zaryada
136–152. (Potentials of Zero Charge), Nauka, Moscow,
35. J. O’M. Bockris, M. A. V. Devanathan, 1982.
K. Müller, Proc. R. Soc. London 1963, A274, 60. A. N. Frumkin, O. A. Petrii, Electrochim.
55–79. Acta 1975, 20, 347–359.
36. J. O’M. Bockris, E. Gileadi, K. Müller, Elec- 61. A. Frumkin, Philos. Mag. 1920, 40, 363–375.
trochim. Acta 1967, 12, 1301–1321. 62. A. Frumkin, Philos. Mag. 1920, 40, 375–385.
37. B. Damaskin, A. Frumkin, A. Chizhov, 63. A. Frumkin, A. Slygin, Acta Physicochim.
J. Electroanal. Chem. 1970, 28, 93–104. URSS 1936, 5, 819–840.
38. M. V. Sangaranarayanan, S. K. Rangarajan, 64. A. N. Frumkin, O. A. Petrii, Doklady Akad.
Can. J. Chem. 1981, 59, 2072–2079. Nauk SSSR 1971, 196, 1387–1390.
39. R. Guidelli, G. Aloisi, J. Electroanal. Chem. 65. O. A. Petrii, Yu. G. Kotlov, Elektrokhimiya
1992, 329, 39–58. 1968, 4, 1256–1260.
40. P. Nikitas, J. Electroanal. Chem. 1994, 375, 66. V. E. Kazarinov, O. A. Petrii, V. V. Topolev
319–338. et al., Elektrokhimiya 1971, 7, 1365–1368.
348 3 Specific Adsorption
67. V. S. Bagotzky, Yu. B. Vassiliev, J. Weber 89. A. Hubbard, J. Stickney, M. Soriaga et al.,
et al., J. Electroanal. Chem. 1970, 27, 31–46. J. Electroanal. Chem. 1984, 168, 143–166.
68. N. A. Balashova, V. E. Kazarinov in Electro- 90. P. Andricacos, P. Ross, J. Electroanal. Chem.
analytical Chemistry (Ed.: A. J. Bard), Marcel 1984, 167, 301–308.
Dekker, New York, 1969, pp. 135–197, 91. A. Hubbard, Heterog. Chem. Rev. 1994, 1,
Vol. 3. 3–39.
69. B. B. Damaskin, O. A. Petrii, V. E. Kazari- 92. M. J. Weaver, I. Villegas, Langmuir, 1997,
nov, Elektrokhimiya 1972, 8, 1373–1377. 13, 6836–6844.
70. E. Gileadi, Collect. Czech. Chem. Commun. 93. V. E. Kazarinov, Elektrokhimiya 1966, 2,
1971, 36, 464–475. 1389–1394.
71. E. Gileadi, Isr. J. Chem. 1971, 9, 405–412. 94. N. V. Pospelova, A. A. Rakov, V. I. Veselov-
72. E. Gileadi, G. E. Stoner, J. Electrochem. Soc. skii, Elektrokhimiya 1969, 5, 797–803.
1971, 118, 1316–1319. 95. A. A. Yakovleva, R. K. Bairamov, V. I. Vese-
73. E. Gileadi, Discuss. Faraday Soc. 1973, 56, lovskii, Elektrokhimiya 1975, 11, 515–518.
228–234. 96. V. I. Naumov, Yu. M. Tuyrin, Elektrokhimiya
74. E. Gileadi, Electrochim. Acta 1987, 32, 1973, 9, 1011–1016.
221–229. 97. V. I. Naumov, V. E. Kazarinov, Yu. M.
75. W. Lorenz, Z. Phys. Chem. 1961, 218, Tuyrin, Elektrokhimiya 1973, 9, 1412–1413.
272–276. 98. N. V. Pospelova, A. A. Rakov, V. I. Vese-
76. W. Lorenz, Z. Phys. Chem. 1973, 254, lovskii, Elektrokhimiya 1969, 5, 1318–1320.
123–133. 99. L. M. Mirkind, M. I. Fioshin, Doklady Akad.
77. A. Frumkin, B. Damaskin, O. Petrii, Z. Nauk SSSR 1964, 154, 1163–1166.
Phys. Chem. 1975, 256, 728–736. 100. O. A. Petrii, Electrochim. Acta 1996, 41,
78. K. Vetter, J. Schultze, Ber. Bunsen-Ges. Phys. 2307–2312.
Chem. 1972, 76, 920–927. 101. D. E. Yates, S. Levine, T. W. Healy, J. Chem.
79. J. Schultze, K. Vetter, J. Electroanal. Chem. Soc., Faraday Trans. 1 1974, 70, 1807–1818.
1973, 44, 63–81. 102. T. Hiemstra, W. H. Van Reimsdijk, G. H.
80. K. Vetter, J. Schultze, Ber. Bunsen-Ges. Phys. Bolt, J. Colloid Interface Sci. 1989, 133,
Chem. 1972, 76, 927–933. 91–104.
81. K. Vetter, J. Schultze, J. Electroanal. Chem. 103. T. Hiemstra, W. H. Van Reimsdijk, G. H.
1974, 53, 67–76. Bolt, J. Colloid Interface Sci. 1989, 133,
82. J. Schultze, F. Koppitz, Electrochim. Acta 105–117.
1976, 21, 327–336. 104. B. B. Damaskin, Elektrokhimiya 1989, 25,
83. F. Koppitz, J. Schultze, Electrochim. Acta 1641–1648.
1976, 21, 337–343. 105. M. A. Vorotyntsev, A. A. Kornyshev, Elek-
84. A. N. Frumkin, Zh. N. Malysheva, O. A. trostatika sred s prostranstvennoi dispersiei (in
Petrii et al., Elektrokhimiya 1972, 8, 599–603. Russian) (Electrostatics of the Spatially Dis-
85. B. E. Conway, E. Gileady, Trans. Faraday persed Media), Nauka, Moscow, 1993.
Soc. 1962, 58, 2493–2509. 106. H. Tamura, T. Oda, M. Nagayama et al.,
86. B. E. Conway, E. Gileady in Modern Aspects J. Electrochem. Soc. 1989, 136, 2782–2786.
of Electrochemistry (Eds.: J. O’M. Bockris, 107. E. A. Nechaev, Zh. Fiz. Khim. 1978, 52,
B. E. Conway), Butterworths, London, 1964, 1494–1496.
pp. 347–442, Vol. 3. 108. M. I. Urbakh, E. A. Nechaev, Elektrokhimiya
87. A. N. Frumkin, Doklady Akad. Nauk SSSR 1980, 16, 1264–1268.
1964, 154, 1432–1436.
88. B. B. Damaskin, O. A. Petrii, V. V. Batrakov,
Adsorption of Organic Compounds on Elec-
trodes, Plenum Press, New York, 1971.
3.2 State of Art: Present Knowledge and Understanding 349
This means that the concept of specific potentials of the metal and electrolyte, re-
adsorption involves the use of nonther- spectively. It is often convenient to take φ S
modynamic methods and model assump- as zero.
tions. The surface free charge density in the
For the orientation of the reader, we diffuse layer is denoted by σ d and it
summarize the fundamental elements of may be regarded as made up of the
the characteristic features of the double- charges contributed by the surface excess
layer model relating to the specific ad- or deficiency of each ionic species (i )
sorption on the basis of the correspond-
ing IUPAC recommendations [1]. For a σd = σid = zi ndi = zi i (1)
detailed survey including historical back-
the sum being overall ionic species in the
ground, see elsewhere [2–7]. solution.
On the basis of the formal relationship
3.2.2
The Structure of the Double Layer d(φ M − φ S ) d(φ M − φ2 )
=
dσ M dσ M
Recapitulation of fundamental notions In
d(φ2 − φ S )
the case of nonspecific adsorption it is + (2)
dσ M
assumed that ions retain their solvation
shell, and in the position of closest the differential capacitance of the whole
approach to the interface they are separated interphase can be conceived as composed
from it by one or more solvent layers. of two components, so Eq. (2) may be
The locus of the electrical centers of written as
nonspecifically adsorbed ions in their
C −1 = (C i )−1 + (C d )−1 (3)
position of closest approach is the outer
Helmholtz plane (OHP). that is, the interphase behaves as a
The position of this plane for the anion series combination of C i , the inner-
and the cation of the electrolyte may differ. layer capacitance, and C d , the diffuse-layer
It may also differ for the different cations capacitance.
and the different anions of solution. The
region in which nonspecifically adsorbed The double layer in the case of specific
ions are accumulated and distributed by adsorption In the case of specifically ad-
the contrasting action of the electric field sorbed ions it is assumed that they pen-
and thermal motion is called the diffuse etrate into the inner layer and may (but
layer. The region between the OHP and not necessarily) come in contact with the
the interface is called the inner (compact) metal surface. They are usually assumed
layer. to form a partial or complete monolayer.
In accordance with the IUPAC defi- The locus of the electrical centers of this
nitions [1] denoting by x2 , the distance layer of specifically adsorbed ions is the
of OHP from the interface, and by φ2 , inner Helmholtz plane (IHP) assumed to be
the mean electric potential at OHP, the at a distance x1 from the metal surface. In
potential drop across the inner layer is certain cases the amount of charge of the
(φ M − φ2 ), while in the diffuse layer it is ions specifically adsorbed in the IHP (σ i ) is
(φ2 − φ S ) where φ M and φ S are the inner higher than the charge on the metal phase
3.2 State of Art: Present Knowledge and Understanding 351
(with the reverse sign); thus the charge of adsorption. The potential drop across the
the diffuse layer has the same sign as that inner layer may be assumed to depend on
on the metal (superequivalent adsorption). the charge σ M and specifically adsorbed
If the contribution to the solution charges σji . From this assumption it fol-
charge from ions whose centers lie in lows generally that
the region x1 < x < x2 is denoted by
σ i , the electroneutrality condition for the M ∂(φ M − φ2 )
d(φ − φ2 ) = dσ M
interphase will be as follows: dσ M σji
σ M = −σ S = −(σ i + σ d ) (4) ∂(φ M − φ2 )
+ dσji (9)
dσji M i
The separation of σ S into σ i , and σ d σ ,σk
cannot be made without the introduction or
of a model. Use of the GC theory and
the assumption that one of the ionic d φ M − φ2 1 ∂ φ M − φ2
≡ i =
components is not specifically adsorbed dσ M C ∂σ M
σji
enables the other σjd to be calculated and
so the specifically adsorbed charge due to ∂ φ M − φ2 dσji
the other ion to be obtained: +
∂σji dσ M
σ M ,σki
σji = σjS − σjd (5) (10)
Thus the inner-layer capacitance may be
In view of the short-range of the forces related to a series of partial capacitances
causing specific adsorption, the quantities that depend on σ M :
σji are often equivalent to less than one
monolayer of ions. Consequently the real ∂σ M
i
C σ M
=
(11)
distribution is often represented in models ∂ φ M − φ2 i
σj
as a monolayer of ions with their centers
on the IHP.
and each σji for the specifically ad-
The charge in the diffuse layer is now
sorbed ions:
σ d = −σ M − σji (6) ∂σji
i
C σj =i
(12)
and the corresponding differential capaci- ∂ φ M − φ2 M i σ ,σk
tance (Csd ) of the diffuse layer is
Here the subscript σki means that all σji are
dσ d kept constant except one.
Csd = − (7)
d(φ2 − φ S )
3.2.3
Thus Eq. (3) can be replaced by Thermodynamics and Models
i −1
1+d σj 3.2.3.1 The Extent of Adsorption
C −1 = (C i )−1 + (C d )−1
dσ M In the previous section it was emphasized
(8) that the charge distribution in the double
The inner-layer capacity can be further ana- layer cannot be described without model
lyzed to express its dependence on specific assumptions concerning both the specific
352 3 Specific Adsorption
system, the charge on both sides of the amount adsorbed (or surface excess) as
double layer expressed through zi i a function of bulk activity of the same
according to species at constant temperature. In the
case of electrified interphases the electrical
σ M = −σ S = − zi i = − zi i state of the interphase must also be
(29) kept constant for the determination of
does not depend on this choice as the relation between surface and bulk
1 concentrations.
zi i = ni zi − nr xi zi Two ways for this are followed in the
A
practical determination of the isotherms.
1
= ni zi (30) These are the determination at constant
A σ M or at constant (φ M − φ S ) (or at
1 nw some constant cell potential). From a
zi i = ni zi − xi zi
A xw thermodynamic point of view, isotherms
1 with respect to relative surface excesses
= ni zi (31) may be determined at constant charge
A
or at any well-defined constant potential.
considering the very fact that owing to However, the interpretation and physical
the electroneutrality
of the bulk solution meaning of the results may be significantly
phase, xi zi = 0. more difficult in the case when constant
cell potential is used.
3.2.3.2 Adsorption Isotherm In the case of adsorbed layers, by analogy
Adsorption in systems of neutral mole- with the two-dimensional pressure of
cules is frequently expressed in terms spread films, a surface pressure defined
of an adsorption isotherm that gives the as the change of interfacial tension (γ )
Name Isotherms
caused by the addition of a given species to For this case the Gibbs adsorption
a base solution is introduced. At constant equation takes the form
cell potential this is
−dγ = σ M dE+ + − dµsalt (35)
π = γbase − γ (32)
−dγ = σ M dE− + + dµsalt (36)
where γbase is the interfacial tension of the
base solution. where E+ and E− are the potentials with
The analogue quantity at constant respect to a reference electrode reversible
for the cation and anion, respectively, −
charge is
= ξbase − ξ (33) and + are the corresponding relative
surface excesses (see Eq. 13) of the ions.
where The relative surface excesses can be
ξ = γ + σE (34) obtained from the following relationship:
In both cases, all other variables should dγ
− = ± (37)
be kept constant. However, in reality the dµsalt E±
interfacial concentration of other species
may be changed by the introduction of one A very important indicator of the occur-
of the components. rence of specific adsorption is the so-called
Various isotherms were proposed on the Esin–Markov coefficient, which in a gen-
basis of different physical models. Three eralized form can be given by the following
approaches should be distinguished [1]. equation:
1. The solution phase is considered as ∂E± 1 ∂E±
= (38)
a continuum and the behavior of the ∂µsalt σ M RT ∂ ln asalt σ M
adsorbed species is similar to that of
Originally the notion of Esin–Markov co-
the adsorbate at the vapor–condensed
efficient was introduced for the condition
phase interface.
of σ M = 0, that is, for the effect observed
2. The adsorption process is a replacement
at the potential of zero charge (Epzc ).
reaction in which the adsorbing species
In the absence of specific adsorption the
replaces another species, normally the
Esin–Markov coefficient is zero, that is,
solvent (see previous section).
no shift of the PZC should occur when the
3. The interphase is regarded as a two-
concentration (activity) of the electrolyte is
dimensional solution.
changed. The origin of the shift at σM = 0
Some of the isotherms described in the in the case of specific adsorption can be
literature are shown in Table 1. understood easily. Holding the potential
at Epzc , if the addition of the salt is fol-
3.2.3.3 Indication and Evaluation of lowed by the specific adsorption of the
Specific Adsorption anion or the cation, this process should re-
In order to show how the occurrence of sult in a change in the electric state of the
specific adsorption can be demonstrated interphase. The charge of the specifically
through thermodynamic relationships, we adsorbed species should be compensated
consider a simple case: solution of a salt by the charge of the counterions on the so-
MeX containing Me+ and X− ions in lution side of the double layer. This means
contact with Hg. that at the preselected potential value, σ M
356 3 Specific Adsorption
does not remain zero following the adsorp- In the case of mercury it is accepted
tion. In order to attain again the σ M = 0, that specific adsorption of F− plays no role
the potential of the electrode should be under certain conditions.
shifted to a new value. In general, specific However, this conclusion for mercury
adsorption of anions is indicated by nega- electrodes cannot automatically be carried
tive shifts in potential at constant charge over to other metals shown, for instance,
density, while specific cationic adsorption in studies of Ag electrodes [9, 10].
results in the opposite shift. The ‘‘method of mixed electrolytes’’ first
The determination of the extent of presented in two fundamental papers by
specific adsorption, remaining at the Hurwitz [11] and Dutkiewicz and Par-
above-mentioned simplest case (MeX), sons [12] was an important contribution
follows the following pattern. to the methods of evaluation of specific ad-
In these simplest cases the evaluation sorption. In Ref. [11], Gibbs’s adsorption
of the specific adsorption is based on equation has been derived in a general
the assumption that one type of ion form for the case of mixed solutions of
in the electrolyte is not involved in the strong electrolytes at constant molal ionic
specific adsorption. Mostly anionic specific strength.
adsorption is calculated assuming that It was predicted that in the case of
the cation (often an alkali ion) is not the binary systems NaF-NaX or KF-
specifically adsorbed. This means that KX (X = Cl− , Br− , I− ), investigated at
the cation is only in the diffuse layer. constant molal ionic strength at a mercury
Determining − and from Eq. (37) electrode, the treatment would lead to
+
(by measuring the interfacial tension), it is the specifically adsorbed surface excess of
possible to calculate, from + the extent Cl− , Br− , I− , provided that the specific
d
of − , excess concentration of the anion adsorption of F− is considered negligible,
in the diffuse layer, by the relationships of which seems a reasonable assumption.
GC theory. Likewise, it was demonstrated that the
The amount of specifically adsorbed determination of Cs+ for Cs+ from
anion is then obtained by the difference mixtures of the type LiCl-CsCl is possible
if the specific adsorption of Li+ is assumed
sp d
− = − − − (39) to be weak.
In Ref. [12] the specific case of KI + KF
(In the case of dilute solutions − ≈ ).
− system was discussed in detail.
The reliability of such a determination Their derivation, which could be an illus-
depends on the validity of GC theory. In the trative example on how to treat a problem
case of strong specific anion adsorption, simultaneously in terms of thermodynam-
this method furnishes reliable results. The ics and model approach, is as follows.
errors in the determination of − d are
The electrocapillary equation for a pure
greatest when specific adsorption is weak mercury electrode in contact with an
and the solution is dilute. aqueous solution containing KI and KF
For the interpretation of inner-layer at constant temperature and pressure may
models in the absence of specific adsorp- be written
tion, it is very important to find cases and
systems where there is no specific anion −dγ = K + dµ̄K+ + I − dµ̄I− + F − dµ̄F−
adsorption at all. (40)
3.2 State of Art: Present Knowledge and Understanding 357
where γ is the interfacial tension, µ̄i the diffuse layer, then it follows that
electrochemical potential of species i, and
i is the surface excess of species i relative Id− x d−
= = I (45)
to the water. If we introduce the chemical F− 1−x F−
potentials of the salts and eliminate the
electrochemical potentials of the anions, because the concentration of ions of the
we may write Eq. (40) in the form same charge in the diffuse layer is directly
proportional to their bulk concentration.
−dγ = σ dE+ + I − dµKI + F − dµKF Thus it may be seen from Eq. (45) that
(41) the second term in the coefficient of
where σ is the charge per unit area on d ln x in Eq. (44) is simply the surface
the mercury surface, E+ is the potential concentration of iodide in the diffuse layer,
of the mercury electrode with respect to and Eq. (44) reduces to
a reference electrode reversible to the
−dγ = σ dE+ + Ii − RT d ln x (46)
potassium ion in the working solution,
and µj is the chemical potential of the Finally, since the concentration of K+ is
salt j . constant and the activity coefficients are
Making the very plausible assumption constant to a good approximation,
that in a mixed-salt solution of constant
total concentration, we may write dE+ = dEN.C.E. (47)
the only feasible way to understand double- The direct in situ study of the specific
layer phenomena and to create a reliable adsorption of ions and organic species
theoretical basis for the interpretation of was dictated by the desire to obtain more
the phenomena observed. and more direct information on the pro-
In principle, in the case of nonspe- cesses occurring on the surface of the
cific adsorption the nature of the electrode electrode. This was one of the most impor-
metal should play a secondary role; there- tant factors leading to the elaboration of
fore, the fundamental relationships found nonelectrochemical methods for the study
for this type of adsorption in the case of of specific adsorption. For instance, radio-
mercury should be of general validity, in- tracer method, ellipsometry, and various
dependent of the nature of the metal. In spectroscopic methods (such as Raman
reality, the experimental results obtained spectroscopy) should be mentioned here.
with mercury, the methods for the evalu- The introduction of new methods re-
ation of these experimental data, and the sulted in a change in the philosophy on
theoretical efforts to interpret them served how to separate specific and nonspecific
as very important basis for the study of adsorption. According to the approach
any kind of electrode. Some of these, for presented in the previous section, the
instance, methods of Hurwitz and of Par- determination of specific adsorption re-
sons and coworkers and the Esin–Markov
quired the deduction of the adsorption in
effect, were mentioned in the previous
the diffuse double layer, calculated on the
section.
basis of model assumption, from the to-
It is evident that such methods as the
tal adsorption determined by well-defined
application of capillary electrometer or the
and reliable thermodynamic methods. In
dropping electrode cannot be used in the
the case of some new methods, the specific
case of solid electrodes. On the other hand,
adsorption of a given species is studied in
the measurement of surface tension and
the presence of a great excess of a sup-
surface stress of solid electrodes is not easy
and in addition, the interpretation of the porting electrolyte. If this excess is several
data is very problematic [13–24]. orders of magnitude higher than the con-
Beginning from the fifties of the last centration of the species studied, there
century, the investigation of the behav- could be no doubt that in the case of a mea-
ior of solid electrodes came into the surable adsorption of these species (more
foreground. The available methods, for in- than 10−11 to 10−12 mol cm−2 ) the value
stance, wetting (contact) angle method, the obtained should be ascribed to specific ad-
measurement of differential capacity, were sorption. This consequence can be clearly
far from furnishing reliable data in the case demonstrated in the case of radiotracer
of solid electrodes. adsorption studies.
The practical aspects of electrocataly- For instance, if the adsorption of 35 S-
sis (fuel cell–oriented research) required labeled H2 SO4 is studied in the presence of
data on the specific adsorption at elec- 1 mol dm−3 HClO4 and the concentration
trodes with high real-surface area, while of labeled sulfate is 10−4 mol dm−3 , the ra-
for theoretical considerations, results ob- tio of the adsorption of HSO4 − and ClO4 −
tained with well-defined electrode surfaces ions in the diffuse part of the double layer
(single-crystal surfaces) were of great im- should be 10−4 as well. This means that
portance. the extent of the nonspecific adsorption
3.2 State of Art: Present Knowledge and Understanding 359
of HSO4 − species under such conditions the choice of the components, the ther-
should be less than 10−13 mol cm−2 . modynamic charge Q (the total charge) of
Despite the development and application the electrode at a given potential should
of new mostly nonelectrochemical meth- be different. This total charge should
ods, ‘‘classical’’ electrochemical methods be distinguished from the so-called ‘‘free
remained in use for solid electrodes. charge’’ corresponding to the double-layer
Whatever the method used should be, model believed to exist on the solution side
the study of the specific adsorption in and (with opposite sign) on the metal side
the presence of a supporting electrolyte of the double layer formed at the interface
has the advantage that the results can be (see later).
obtained directly without any calculations The study of the UPD of metal ions
based on model assumptions concerning belongs to the group of adsorption studies
the structure of the double layer. in the presence of a great excess of
supporting electrolyte as in most cases
3.2.4.2 Underpotential Deposition: A studied the concentration of metal ions
Limiting Case of Specific Adsorption is chosen to be very low in comparison
In the literature the specific adsorption and with that of the ions of the supporting
the underpotential deposition (UPD) pro- electrolyte.
cesses are discussed separately. For a long The most important exception to this
period the adsorption of hydrogen on Pt, rule is the specific adsorption, the UPD, of
Rh, Pd electrodes at more positive potential H+ ions or hydrogen adsorption on some
values than the hydrogen equilibrium po- noble metal electrodes. These systems
tential was considered as some specific and were studied from several molar acid
interesting phenomenon characteristic for concentrations to very high pH values. For
some noble metals and hydrogen. detailed discussion of UPD see Sects. 3.2
However, beginning from the sixties it and 3.3.
became quite evident that the adsorption
of metal ions on foreign metal substrates
3.2.4.2.1 A Widely Used and Misinter-
is often accompanied by significant charge
preted Misconception: The Electrosorp-
transfer commensurable with the charge
tion Valency In the classical double-layer
involved in hydrogen adsorption at noble
models it is assumed that specifically ad-
metal electrodes [25–28]. These phenom-
sorbed ions retain their original charge
ena were termed as UPD characterized by
during the adsorption process.
the general equation
In the sixties, Lorenz and coworkers have
MSn+ + ne−
−−
−−−
−Mad (51) suggested that a purely electrostatic bond
between the ion and the metal cannot be
From thermodynamic point of view, it is considered as a physical reality and there
not possible to decide whether we have should be a degree of charge transfer
M n+ and ne− or Mad at the interface. If similar to the formation of a covalent
the equilibrium characterized by Eq. (51) bond [29–33]. The specific adsorption on
is considered as a reality, only two from the basis of this concept was formulated as
the three reaction partners can be consid- follows
ered as independent components of the
z(1−λ)
(thermodynamic) system. Depending on R z = Rads + λze (52)
360 3 Specific Adsorption
the interphase is maintained. The chem- 2. Reference electrode reversible with re-
ical potential of the gold is, of course, spect to K ions (EK )
invariant at constant temperature and
pressure. ∂γ
QK = z+ ν+ F KA =
The chemical potential of the monolayer ∂EK µTA ,µT
T may be determined from the potential (62)
(ET ) of the working electrode with respect 3. Reference electrode reversible with re-
spect to T ions (ET )
to an electrode of pure T, both in
equilibrium with T ions in the solution,
∂γ
that is, the electromotive force (emf) of the QT = zT F T = (63)
∂ET µKA ,µTA
cell T|T+ |T|Au is considered
z− , z+ and zT are the charge of ions, ν− is
(µT (θ) − µoT ) the number of anions in salt TA, and ν+ is
ET = − (57)
zT F the number of cations in salt KA. QA , QK ,
and QT are the so-called thermodynamic
where µT (θ) is the chemical potential of T or total charges of the electrode. This
at coverage θ on the gold surface and µoT is thermodynamic charge in general differs
the standard chemical potential of pure T. from the ‘‘free charge’’ believed to exist
It follows directly from Eq. (56) that on the two sides of the double layer.
three surface excesses are accessible to They can be considered equal to each
experiment: other in the case of ideally polarizable
electrodes without any charge transfer. In
such situation the charge of the electrode
∂γ
TA = − (58) has a single value.
∂µTA µKA ,µT
However, in the case of adsorption
∂γ processes connected with charge transfer
KA = − (59) (for instance, UPD), the thermodynamic
∂µKA µTA ,µT
charge of the electrode has more than one
∂γ value as owing to the equilibrium pro-
T = − (60)
∂µT µTA ,µKA cesses (with charge transfer), the number
of the degrees of freedom for the choice
From these equations, three Lippmann of thermodynamically independent com-
equations can be derived depending on ponents decreases, that is, it will be less
the choice of reference electrode reversible than the number of chemical components
with respect to one of the ions present in distinguishable in the system.
the system. In the case discussed above we have the
following choices.
1. The potential of the working electrode is The system is built up from
measured against a reference electrode
reversible with respect to A ions (EA ) 1. KA and T
2. TA and T
3. KA and TA
∂γ
QA = z− ν− F TA =
∂EA µKA ,µT The QA , QK , and QT charge values cor-
(61) respond to the charge to be introduced
3.2 State of Art: Present Knowledge and Understanding 363
electrically together with the correspond- at the conclusion that the results obtained
ing material to create 1 cm2 of interphase confirm that the charge associated with
of the electrode keeping constant all the the unusual states is at least formally
parameters characterizing the system. compatible with the exchange of one
electron per three surface platinum atoms.
3.2.4.4 Anion Adsorption on Well-defined In accordance with this view, the oc-
Crystal Surfaces currence of the unusual peaks on the
Fundamental questions connected with cyclic voltammograms observed in acetic
the characterization of the double layer acid–acetate system was ascribed to a
at well-defined single-crystal faces of solid one-electron-transfer process of acetic
metals have been reviewed and surveyed acid/acetate ions according to the equa-
several times by Hamelin [6, 69–72]. tions [83]
Since the early 1980s, cyclic voltammetry
−− + −
has been used to characterize single-crystal CH3 COOH−−−
−CH3 COOads + H + e
electrodes in terms of surface order, pres-
at pH < 2
ence or absence of defects contaminations,
and so on. In some cases the voltammo- CH3 COO −−−CH3 COOads + e−
− −−−
grams have also been used for identifica-
tion of adsorbates, mostly on the basis of at pH > 5
electric charge calculation. However, since
the double-layer charging and surface re- There are, however, many evidences
actions such as UPD and anion adsorption demonstrating that the existence of the
may occur in parallel, it is difficult to break unusual states cannot be easily explained
down the total voltammetric charge into all with the charge transfer to or from the
of its individual components. Therefore, anions (or other species). For instance, re-
interpretation from voltammetry alone, cent spectroscopic studies do not support
used as a tool for the adsorbed species the occurrence of electron transfer in the
identification, may be ambiguous. This case of anion adsorption [77, 79, 84–88].
has been, for instance, the case with the A very important and confusing ob-
interpretation of the ‘‘unusual adsorption servation is that the charge involved in
states’’ on the Pt(111) electrode [73–82]. the unusual states in various solutions
Although at a very early stage of the is nearly independent of the nature of
studies the appearance of the unusual the anion, including organic anions. This
peak on the voltammograms of Pt(111) was aspect of the unusual states cannot be
explained by anion adsorption for a long easily reconciled with the charge trans-
period, the view that it should be ascribed fer to or from the anion as a mechanistic
to hydrogen adsorption (or incorporation origin of the pseudocapacitive current gen-
of hydrogen) was widely accepted by erated by the reversibly adsorbed anions on
many researchers. Recently the hydrogen Pt(111).
adsorption model is rejected on the Results obtained from careful exam-
basis of reexamination of the problem ination of the pH effects on the po-
by the most eminent proponents of the tentiodynamic behavior of the Pt(111)
model studying the unusual desorption electrode in acidified perchlorate solu-
states on Pt(111) electrodes by CO and I tions [82] advocate in the favor of the
displacement method. Firstly, they arrived assumption that water and not ClO4 −
364 3 Specific Adsorption
anions are involved in the unusual ad- (ClO4 − ) is kept constant when the concen-
sorption states in the double-layer region tration of the adsorbing anion is changed.
of Pt(111). Presumably H2 O dissociation It is considered that the effect of the ad-
and OH deprotonation are responsible for sorption of the anion of the supporting
the two main electrode processes in the electrolyte on the measured surface excess
double-layer region of Pt(111) in NaClO4 of the investigated anion can be mini-
solutions. mized under this condition. In addition,
For instance, in situ Fourier transform it is assumed that in the presence of an
infrared (FTIR) spectroscopy has been excess of supporting electrolyte the mea-
used by Faguy and coworkers [79] to study sured surface excess is essentially equal to
the potential dependent changes in anion the surface concentration.
structure and composition at the surface In the case of sulfate adsorption the sur-
of Pt(111) electrodes in HSO4 − -containing face excesses determined from the charge
solutions. From the infrared (IR) differen- density data were compared with the Gibbs
tial normalized relative reflectance data, excesses determined from radiochemical
the maximum rate of intensity changes measurements using 35 S-labeled sulfate
for three IR bands can be obtained. Two solutions. Very good agreement of the data
modes associated with the adsorbed an- was reported [89].
ion and one mode assigned to species is In order to obtain insight into the nature
not adequately described as either sulfate of the adsorbed species, Esin–Markov co-
or bisulfate; the data are more consis- efficients for SO4 2− adsorption from two
tent with an adsorbed H3 O+ −SO4 2− ion series of solutions were determined: (1) at
pair, possibly with the three unproto- constant pH and variable K2 SO4 concen-
nated sulfate oxygens interacting with Pt tration; (2) at constant K2 SO4 concentra-
sites. tion and variable pH. The Esin–Markov
In a series of papers by Lipkowski and coefficients were used to identify the na-
coworkers, specific adsorption SO4 2− , Cl− , ture of the adsorbed species (SO4 2− or
and Br− on Au(111) electrode from 0.1 M HSO4 − ). The authors arrived at the con-
HClO4 + x M K2 SO4 and 0.1 M HClO4 + clusion that SO4 2− ion is the adsorbed
x M KCl or KBr or KI solution was studied species even if HSO4 − predominates in
using chronocoulometry and evaluating the bulk of the solution. Similar results
the data on the basis of thermodynamic were reported in Ref. [96].
of ideally polarizable electrode (x varies This conclusion, however, seems to be
between 5 × 10−6 and 5 × 10−3 M) [53, 56, questionable, in the light of thermody-
89–94]. A further study was devoted to namic considerations and other experi-
OH− adsorption [95]. mental results (see later).
We should make a distinction between Results obtained for Br− , Cl− , and
these cases and the case of adsorption −
I adsorption were compared. The main
from a mixed electrolyte at constant ionic conclusions are as follows: All the three
strength discussed in Sect. 3.2.3.3. In this halides form a chemisorption bond with
latter case, the investigated anion replaces the gold surface. The bonds are quite
progressively the anion of the supporting polar at the negatively charged surface;
electrolyte, while in the method by Lip- however, its polarity drops significantly
kowski and coworkers the concentration at the positively charged surface. At low
of the anion of the supporting electrolyte charge densities and coverages, the bond
3.2 State of Art: Present Knowledge and Understanding 365
polarity is determined by the ability of free and neutral media. The adsorption of sul-
electrons to screen the dipole formed by fate in alkaline solutions is inhibited by
the adsorbed anion and its image charge the more strongly specifically adsorbed
in the metal. At high charge densities and OH− ions. The potential dependence of
coverages, the chemisorption bond has surface coverage, estimated on the basis
a predominantly covalent character. The of spectroelectrochemical data, indicated
strength of the halide adsorption and the that a considerable repulsion exists among
covalent character of the chemisorption the adsorbed sulfate species. The spectral
bond increase progressively by moving data were compared to those obtained for
from chloride to iodide. sulfate adsorption on Pt(111) and on un-
An interesting comparative study of the derpotentially deposited Ag on the Pt(111)
adsorption of bisulfate anion on different surface.
metals was carried out by Shingaya and Voltammetric study of the adsorption
Ito [88]. Adsorption of sulfuric acid species of 14 anions on Ag(111) was reported in
on various metal(111) electrodes in a 0.5- Ref. [98].
M H2 SO4 solution was investigated by IR The specific anion adsorption of chloride
reflection absorption spectroscopy (IRAS). and sulfate on Cu(111) from acidic aque-
Pt, Au, Ag, Cu, and Rh(111) were used as ous electrolytes has been studied by Broek-
electrodes. According to this study, HSO4 − mann and coworkers [99] using scanning
is adsorbed on all of the surfaces including tunneling microscopy (STM) and cyclic
Pt, Rh, Au, Ag, and Cu in a 0.5-M H2 SO4 voltammetry. It was found that at posi-
solution. It was found that considerable tive potentials √
adsorbed
√ chloride forms a
amounts of sulfate in addition to bisulfate well-ordered ( 3 × 3)R30◦ superstruc-
coexist on Ag(111) in a 0.5-M H2 SO4 ture. At negative potentials the bare copper
solution. An interesting finding that both surface can be observed. In the case of
DSO4 − and SO4 2− are coadsorbed on sulfate adsorption at positive potentials,
Au(111) in 0.5-M D2 SO4 + D2 O solution. high-resolution STM images revealed an
The interconversion of HSO4 − /H3 O+ and additional species that was assigned to
H2 SO4 that was seen on Pt(111) was not coadsorbed water molecules. Close-packed
observed on the other electrodes. The sulfate rows are separated by zigzag chains
frequencies of the SO3 symmetric and of these water molecules.
S−OH stretching bands of a bisulfate Adsorption of sulfate ions on the three
adsorbed on the electrode surfaces are basal planes of monocrystalline copper and
related to the electronic state of each silver electrodes in 0.1 M HClO4 was stud-
M(111) surface. ied by radiometric and electrochemical
A somewhat different conclusion was re- methods [100, 101].
ported for Ag(111) by Adzic and coworkers The adsorption of OH− ions on
in a study [97] in which cyclic voltamme- the Pt(111) plane has been studied
try and in situ IR spectroelectrochemistry by Drazic and coworkers [102] using
were employed to investigate the ionic ad- fast cyclic voltammetry in sodium hy-
sorption on the Ag(111) electrode surface droxide solutions (0.03 to 1 M), under
in sulfate-containing solutions of differ- quasi-equilibrium and Tafel approxima-
ent pH. It has been concluded that the tion conditions. It was shown that the
sulfate anion is the predominant species OH− ion adsorption is an electrosorp-
adsorbed at the electrode surface in acidic tion process with one electron exchanged
366 3 Specific Adsorption
between an OH− ion and the platinum electrode potential and constant charge
surface. The electrosorption process fol- coincide.
lows the Frumkin adsorption isotherm
with low-intensity repulsive interactions of 3.2.5
the adsorbed species (f = 2–3). The esti- Methods of Investigation and Illustrative
mated values of the standard electrochemi- Examples
cal rate constant (k o = 5.6 × 10−4 cm s−1 )
and the standard exchange current den- 3.2.5.1 General Remarks
sity (j00 = 5.45 × 10−2 A cm−2 ) indicate a During the last two to three decades, so-
rather fast electrochemical process. phisticated methods of solid-state surface
The thermodynamic and inner-layer science were introduced to study the ad-
characteristics of halide ions adsorbed on sorption phenomena at electrochemical
bismuth single-crystal planes from so- interfaces. These new techniques of inves-
lutions of methanol and propanol was tigation enable us to interpret adsorption
studied by an Estonian group in a se- phenomena at molecular level.
ries of publications [103–108]. The ad- The main characteristic of most of the
sorption of Cl− , Br− , and I− ions from new experimental methods is the simul-
methanolic solutions on different faces taneous use (coupling) of electrochemical
of a bismuth single-crystal electrode was techniques with other nonelectrochemical
studied by impedance and chronocoulo- methods. Only some of these methods
metric techniques. The charge of specif- can be very briefly mentioned in the
ically adsorbed anions σ1 was calculated following sections. More detailed descrip-
using the Hurwitz–Parsons–Dutkiewicz tion of the methods and their application
method. Fitting the results to the simple can be found in monographs and text-
virial isotherm, the values of the stan- books [109–111].
dard Gibbs energy of adsorption −GoA
and the mutual repulsion coefficient B 3.2.5.2 In Situ Techniques
of the adsorbed anions were estimated.
The dependence of the integral capaci- 3.2.5.2.1 Optical Techniques
tance of the inner layer on the electrode
charge σ was calculated according to UV-visible Techniques
the traditional inner-layer model using
the electrode charge as the independent Ellipsometry [112, 113] Ellipsometry is one
electrical variable. It was found that the of the earliest optical technique to be
adsorption of halide ions decreases in the applied to the study of adsorption pro-
sequence of the Bi single-crystal planes cesses [112, 113]. It involves the analysis
(011) > (001) > (111). of the phase change and the change in
As a continuation of this work, in accor- amplitude ratio of polarized light reflected
dance with the method suggested by Lip- from a surface.
kowski and coworkers, instead of electrode A significant limitation of the method
charge the electrode potential was consid- is that a model for the adsorbed layer
ered as independent electrical variable for should be made for the evaluation of
the evaluation of the experimental data. the experimental data. As the theory of
It was found that under comparable ellipsometry is based on the assumption
conditions the results obtained at constant that the reflection occurs from an ideally
3.2 State of Art: Present Knowledge and Understanding 367
smooth surface, the interpretation of data One of the most commonly applied IR
obtained with surfaces with nonideal techniques developed to overcome these
topography encounters difficulties. problems is the external reflectance tech-
nique. In this method, the strong sol-
Specular reflectance spectroscopy [114] The vent absorption is minimized by simply
principle of the method is the recording of pressing a reflective working electrode
a reflection of the incident monochromatic against the IR transparent window of
beam by the surface of the electrode as a the electrochemical cell. The sensitivity
function of wavelength, potential, or time. problem, that is, the enhancement of
Similar to ellipsometry, a model of the the signal/noise ratio in the case of ex-
interface is required to evaluate the ex- ternal reflectance techniques is solved
perimental data. Several techniques were by various approaches. These are, for
elaborated to enhance the sensitivity of instance, electrochemically modulated in-
reflectance measurements (for instance, frared spectroscopy (EMIRS), in situ FTIR
multiple reflectance). (which use potential modulation), and
In order to avoid the problems connected polarization modulation infrared reflec-
with measurement of absolute reflectivi- tion absorption spectroscopy (PM-IRAS,
ties, changes in reflectivity are detected as FTIR) [86, 117–123].
the potential of the electrode is modulated A new development is the elabora-
in the case of the so-called ‘‘modulated tion of synchrotron far infrared spec-
specular reflectance spectroscopy.’’ troscopy (SFIRS) for adsorption stud-
ies [124].
Vibrational Spectroscopy (Infrared and Ra-
man Scattering Spectroscopies) Although Raman Spectroscopy The major problems
UV-visible techniques were proven useful, in the case of the application of IR
they are not suitable for identifying ad- spectroscopy, the strong absorption by
sorbates or orientation of adsorbates. This the solvent and window material, can
type of information could be provided by be avoided using Raman scattering tech-
the various versions of vibrational spec- niques, since only visible radiation is
troscopy. involved in this case.
During the last decade, vibrational spec- Raman scattering cross sections of typ-
troscopy has contributed to a great extent ical adsorbates are low, therefore two
to the advance in electrochemical surface fundamental techniques were elaborated
science to enhance them: resonance Raman scat-
tering (RRS) and surface-enhanced Raman
Infrared (IR) spectroscopy [115, 116] For in scattering (SERS) [125–127].
situ application of IR spectroscopy for
the study of adsorption, there were two Second harmonic generation anisotropy from
fundamental problems to overcome: single-crystalline electrode surfaces [128] In-
terfacial second harmonic generation
1. Most solvents (especially water) absorb (SHG) is an optical spectroscopy that
IR very strongly. is inherently surface-sensitive. All effects
2. How to separate the signals coming that modify the electron density profile
from the adsorbed layer from the of the interface affect the SHG pro-
noises. cess. Therefore, SHG depends on the
368 3 Specific Adsorption
Solution
Adsorbed layer
Is Ia
D
Scheme 1
370 3 Specific Adsorption
Detector
Solution flow
Electrode
Scheme 2
Electrode Electrode
Bottom of
the cell
Solution Gap
D D
liquid in a thin layer near the surface Au(111) was investigated and interpreted
has a higher viscosity because of the in Ref. [164]. Other interesting results
high concentration of ions and the high obtained in various systems have been
electrical field in this region. reported in Refs. [48–50, 165–175].
Some of the illustrative examples for the Atomic force microscopy (AFM) similar
application of EQCN technique are col- to STM is a scanning atomic-scale probe
lected in [157]. The electrical double-layer microscopy; however, AFM does not rely
region of optically polished gold in 0.1-M on tunneling. It detects the interatomic
NaF, KF, RbF, and CsF solutions in the forces between the tip and surface atoms.
absence and with addition of specifically This method provides a very powerful
adsorbing anions SO4 2− , OH− , Cl− , and complement to STM. It can be used in
Br− was studied, arriving at the conclu- cases when faradaic current is flowing,
sion that EQCN can be used to investigate that is, it offers the possibility of studying
the electrical double-layer phenomena on surfaces in the course of electrochemical
metal electrodes. processes. Nonconducting samples can be
A combined interferometric and EQCN imaged as well.
technique was elaborated for the study
of the energetics of adsorption phenom- 3.2.5.3 Ex Situ Methods
ena [159]. In the application of in situ methods, the
EQCN technique is very suitable for presence of the bulk electrolyte consti-
the study of the deposition and disso- tutes a significant limitation. As it was
lution processes of adatoms on various shown in the previous sections, by the
electrodes [55, 159–163]. use of thin-layer cells or carrying out
the measurements in the presence of a
3.2.5.2.4 Scanning Tunneling Microscopy thin-solution layer, the role of the bulk
(STM) and Atomic Force Microscopy (AFM) electrolyte phase can be reduced.
STM was introduced into electrochemistry An alternative route suggested in the lit-
during the late 1980s. The method is erature [114, 176–178] to overcome these
based on the phenomena of quantum- limitations is the emersion of the elec-
mechanical tunneling. It allows atomic trode from the electrochemical cell and
imaging of the structural properties of both transferring it into an ultrahigh vacuum
bare and adsorbate-covered surfaces. The (UHV) chamber, where studies using
principle of the method is as follows: surface-sensitive methods can be car-
A sharp tip of a chemically inert metal ried out.
is brought so close to the electrode However, this procedure involves the
surface that tunneling can occur and a loss of the electrochemical environment
tunneling current flows maintaining the and potential control. Auger electron
potential difference between the tip and spectroscopic (AES) studies [179] show
the electrode. Scanning the tip across that despite the evacuation required by
the surface of the sample at constant AES, chemisorbed species remain at
current or height from the surface, a the surface and their identification and
topographical image of the surface with quantitization can be carried out in
atomic resolution can be obtained. vacuum.
The coadsorption structure of Cu According to Hubbard [179], cationic
and halides on Pt(111), Pt(100), and species present in the emersed layer
372 3 Specific Adsorption
of liquid or in the electrical double solution. This common potential scale rep-
layer (EDL) can also be determined us- resents an important link between UHV
ing AES. However, anions X− , present studies and electrochemistry. The poten-
in the emersed EDL and solution, typ- tial range accessible by the investigation
ically undergo hydrolysis to volatile HX of substrates with different work functions
species, which are subsequently lost to the can be expanded by the coadsorption of
vacuum. electronegative and electropositive species,
In contrast to the latter statement, Kolb such as halogens or alkali metals, which
and coworkers reported somewhat differ- induce large work function changes. The
ent results claiming that under certain adsorption of anions or cations from
conditions the electrode can be removed electrolytes, achieved in UHV at low tem-
from the electrolyte into an inert atmo- perature, is of particular interest. Solvation
sphere or even into vacuum with the processes and surface reactions, which
double layer apparently intact and the bulk result from the electrostatic interaction
electrolyte completely absent [114]. They between the partially ionically adsorbed
measured the work function of a gold species and coadsorbed H2 O, the most
electrode after emersion as a function of frequently used solvent, can be studied
potential in order to prove that the poten- by applying surface-sensitive techniques.
tial drop across the double layer remains A variety of methods have been developed
unchanged. allowing the investigation of the surface
The chemical composition of the structure and chemical composition.
emersed double layer was tested by ESCA. The structure of sulfuric acid species
It was found for nonspecific adsorbing adsorbed on a Pt(111) electrode has
ions that the ex situ obtained data are been successfully determined by using
in very good agreement with predictions the UHV modeling [186]. The micro-
from classical double-layer theory. scopic information about the structure
In the case of immersion technique of an adsorbed electrolyte anion and wa-
the opposite way (from UHV chamber ter as a solvent is indispensable to a
to the solution phase) is followed for true molecular level understanding of
the determination of the PZC of Au(111), the electrochemical double layers and
Pt(111) electrodes [180]. chemical reactions at the electrode sur-
faces.
3.2.5.3.1 Electrochemical Double-layer
Modeling Under Ultrahigh Vacuum 3.2.5.4 Illustrative Examples
Conditions Simulations of the elec- 1. The influence of the surface structure
trochemical double layer under UHV on the anion adsorption. Potential
conditions allow the characterization dependence of the adsorption of sulfate
of the chemical interaction between ions at various crystal faces of copper as
electrode surface and electrolyte con- studied by radiotracer technique [101]
stituents [181–186]. The electron work is shown in Fig. 1.
function measured for an adsorbate- 2. Effect of the nature of the an-
covered metal in UHV is considered to be ion on the potential dependence of
equivalent to the electrochemical potential the adsorption. Data obtained by the
for the same adsorbate-–electrode pair in method of Lipkowski and coworkers at
3.2 State of Art: Present Knowledge and Understanding 373
Cu(111) Cu(110)
4
1
10−14 Γ/ ions cm−2
Cu(100) Cu(poly)
4
0
−0.7 −0.6 −0.5 −0.4 −0.3 −0.2 −0.1 0 −0.7 −0.6 −0.5 −0.4 −0.3 −0.2 −0.1 0
E
[V]
Fig. 1 Potential dependence of sulfate adsorption on Cu electrodes from 5 × 10−4 M H2 SO4
in 0.1 M HClO4 ; (•), positive scan; (), negative scan. Potentials were measured versus
Ag|AgCl|1 M Cl− . (Reprinted from Ref. [101] with permission from Elsevier Science.)
374 3 Specific Adsorption
80
I−
10−13 Γ/ions cm−2
60
Br−
Cl−
40
SO42−
20
0
−800 −400 0 400 800
E vs SCE
[mV]
Fig. 2 Comparison of SO4 2− , Cl− , Br− , and I− adsorption at the
Au(111) electrode surface from 0.1 M HClO4 + 10−3 M K2 SO4 and
0.1 M HClO4 + 10−3 M KCl, KBr, and KI solutions. (Reprinted from
Ref. [93] with permission from Elsevier Science.)
A-1
40
20
[µA cm−2]
i
−20
C-2
−40
−60 C-1
20
III II
I
[µA cm−2]
0
j
−20
−E H
[mV]
600 700 800 900 1000 1100 1200 1300 1400
10
8 D1
6 D3
D2
4
[µA cm−2]
2
0
i
−2
−4 A1
A3 A2
−6
−8
−10
−100 0 100 200 300 400 500 600 700 800
∆E
(a) [mV]
180
160 Adsorption
Desorption
140
120
[µC cm−2]
100
zF Γ
80
60
40
20
0
−100 0 100 200 300 400 500 600 700 800
∆E
(b) [mV]
Fig. 5 (a) Cyclic voltammogram in the system
Au(100)/5 × 10−3 M Ag2 SO4 + 0.5 MH2 SO4 at T = 298 K.
|dE/dt| = 7 mV s−1 . An and Dn with n = 1, 2, 3 denote cathodic
adsorption and anodic desorption peaks, respectively. (b) (E)
isotherm of the system Au(100)/4.2 × 10−4 M Ag2 SO4
+ 0.5M H2 SO4 at T = 298 K. (Reprinted from Ref. [187] with
permission from Elsevier Science.)
3.2 State of Art: Present Knowledge and Understanding 377
4
2
0
−2
[Hz]
∆f
−4
−6
100 ng cm−2
−8
−10
John Wiley & Sons, New York, 1980, 25. D. M. Kolb, in Advances in Electrochem-
pp. 488–552. istry and Electrochemical Engineering (Eds.:
4. S. Trasatti, in Trends in Interfacial Elec- H. Gerischer, C. W. Tobias), John Wiley
trochemistry (Ed.: A. F. Silva), D. Reidel & Sons, New York, 1978, pp. 125–271,
Publishing Company, Dordrecht, 1986, Vol. 11.
pp. 25–48. 26. R. R. Adžić, in Advances in Electrochem-
5. R. Parsons, in Trends in Interfacial Elec- istry and Electrochemical Engineering (Eds.:
trochemistry (Ed.: A. F. Silva), D. Reidel H. Gerischer, C. W. Tobias), John Wiley
Publishing Company, Dordrecht, 1986, & Sons, New York, 1985, pp. 159–260,
pp. 71–81. Vol. 13.
6. A. Hamelin, in Trends in Interfacial Elec- 27. A. Aramata, in Modern Aspects of
trochemistry (Ed.: A. F. Silva), D. Reidel Electrochemistry (Eds.: J. O’M. Bockris
Publishing Company, Dordrecht, 1986, et al.), Plenum Press, New York, 1977,
pp. 83–102. pp. 181–250, Vol. 31.
7. R. Parsons, in Trends in Interfacial Elec- 28. S. Szabó, Int. Rev. Phys. Chem. 1991, 10,
trochemistry (Ed.: A. F. Silva), D. Reidel 207–248.
Publishing Company, Dordrecht, 1986, 29. W. Lorenz, Z. Phys. Chem. 1962, 219,
pp. 373–385. 421–423.
8. D. H. Everett, Pure Appl. Chem. 1970, 21, 30. W. Lorenz, Z. Phys. Chem. N. F. 1967, 54,
583–596. 191–195.
9. G. Valette, R. Parsons, J. Electroanal. Chem. 31. W. Lorenz, Z. Phys. Chem. 1970, 244, 65–84.
1986, 204, 291–297. 32. W. Lorenz, Z. Phys. Chem. 1971, 248,
10. G. Valette, R. Parsons, J. Electroanal. Chem.
161–170.
1985, 191, 377–386.
33. C. Engler, W. Lorenz, Z. Phys. Chem. 1978,
11. H. D. Hurwitz, J. Electroanal. Chem. 1965,
259, 1188–1190.
10, 35–41.
34. K. J. Vetter, J. W. Schultze, Ber. Bunsen-Ges
12. E. Dutkiewicz, R. Parsons, J. Electroanal.
Phys. Chem. 1972, 76, 927–933.
Chem. 1966, 11, 100–110.
35. K. J. Vetter, J. W. Schultze, J. Electroanal.
13. G. Láng, K. E. Heusler, J. Electroanal. Chem.
Chem. 1974, 53, 67–76.
1995, 391, 169–179.
36. A. Aramata, S. Taguchi, T. Fukuda et al.,
14. K. E. Heusler, G. Láng, Electrochim. Acta
1997, 42, 747–756. Electrochim. Acta 1998, 44, 999–1007.
15. G. Valincius, Langmuir 1998, 14, 37. G. Horányi, A. Aramata, J. Electroanal.
6307–6319. Chem. 1997, 437, 259–262.
16. J. Lipkowski, W. Schmickler, D. M. Kolb 38. G. Horányi, A. Aramata, J. Electroanal.
et al., J. Electroanal. Chem. 1998, 452, Chem. 1997, 434, 201–207.
193–197. 39. G. Horányi, ACH Models Chem. 1997, 134,
17. W. Haiss, R. J. Nichols, J. K. Sass et al., J. 33–47.
Electroanal. Chem. 1998, 452, 199–202. 40. G. Horányi, M. Wasberg, J. Electroanal.
18. W. Schmicler, E. Leiva, J. Electroanal. Chem. Chem. 1996, 413, 161–164.
1998, 453, 61–68. 41. P. Zelenay, M. Gamboa-Aldeco, G. Horányi
19. R. Guidelli, J. Electroanal. Chem. 1998, 453, et al., J. Electroanal. Chem. 1993, 357,
69–78. 307–326.
20. G. Valincius, J. Electroanal. Chem. 1999, 42. K. Varga, P. Zelenay, G. Horányi et al., J.
478, 40–49. Electroanal. Chem. 1992, 327, 291–306.
21. B. M. Grafov, Russ. J. Electrochem. 1999, 35, 43. N. M. Marković, H. A. Gasteiger, P. N. Ross
1029–1032. Jr., Langmuir 1995, 11, 4098–4108.
22. G. Láng, K. E. Heusler, J. Electroanal. Chem. 44. Y. Shingaya, H. Matsumoto, H. Ogasawara
1999, 472, 168–173. et al., Surf. Sci. 1995, 335, 23–31.
23. R. Guidelli, J. Electroanal. Chem. 1999, 472, 45. R. Gómez, H. S. Yee, G. M. Bommarito
174–177. et al., Surf. Sci. 1995, 335, 101–109.
24. B. M. Grafov, J. Electroanal. Chem. 1999, 46. N. M. Marković, H. A. Gasteiger, C. A. Lucas
471, 105–108. et al., Surf. Sci. 1995, 335, 91–100.
3.2 State of Art: Present Knowledge and Understanding 379
181. I. Villegas, M. J. Weaver, J. Phys. Chem. and Applications (Ed.: A. Wieckowski), Mar-
1996, 100, 19 502–19 511. cel Dekker, New York, 1999, pp. 1–17.
182. Y. Shingaya, M. Ito, Electrochim. Acta 1998, 190. L. Blum, M. D. Legault, D. A. Huckaby in
44, 889–895. Interfacial Electrochemistry, Theory, Experi-
183. G. Pirug, H. P. Bonzei, Surf. Sci. 1998, 405, ment, and Applications (Ed.: A. Wieckowski),
87–103. Marcel Dekker, New York, 1999, pp. 19–31.
184. E. M. Stuve, K. Bange, J. K. Sass, in Trends 191. M. L. Berkowitz, I. C. Yeh, E. Spohr, in In-
in Interfacial Electrochemistry (Ed.: A. F. terfacial Electrochemistry, Theory, Experiment,
Silva), D. Reidel Publishing Company, and Applications (Ed.: A. Wieckowski), Mar-
Dordrecht, 1986, pp. 255–280. cel Dekker, New York, 1999, pp. 33–45.
185. G. Pirug, H. P. Bonzel, in Interfacial Electro- 192. G. Brown, P. A. Rikvold, S. J. Mitchell et al.,
chemistry, Theory, Experiment, and Applica- in Interfacial Electrochemistry, Theory, Experi-
tions (Ed.: A. Wieckowski), Marcel Dekker, ment, and Applications (Ed.: A. Wieckowski),
New York, 1999, pp. 269–286. Marcel Dekker, New York, 1999, pp. 47–61.
186. Y. Shingaya, M. Ito, in Interfacial Electro- 193. A. Ignaczak, J. A. N. F. Gomes, J. Elec-
chemistry, Theory, Experiment, and Applica- troanal. Chem. 1997, 420, 209–218.
tions (Ed.: A. Wieckowski), Marcel Dekker, 194. M. T. M. Koper, Electrochim. Acta 1998, 44,
New York, 1999, pp. 287–300. 1207–1212.
187. S. Garcia, D. Salinas, C. Mayer et al., Elec- 195. W. Schmickler, (Ed.), J. Electroanal. Chem.
trochim. Acta 1998, 43, 3007–3019. 1998, 450(2), 157–351.
188. S. Trasatti, (Ed.), Electrochim. Acta 1996, 196. P. A. Bopp, A. Kohlmeyer, E. Spohr, Elec-
41(14), 2071–2338. trochim. Acta 1998, 43, 2911–2918.
189. J. W. Halley, S. Walbran, D. Lee Price, in In- 197. E. Spohr, Electrochim. Acta 1999, 44,
terfacial Electrochemistry, Theory, Experiment, 1697–1705.
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 383
the competition between the depth and of nearest neighbors, while treating the
width of substrate potential wells, and other interactions, as well as interactions
the dimension and magnitude of the lat- of nearest neighbors with their neighbors
eral interaction energy of the adsorbed in the MFA. Compared to the MFA
system. (or BWA) and the QCA, a further
The consideration of interactions be- advancement, especially at T < Tcrit , is
tween the adsorbed particles is a sig- represented by the low-temperature series
nificant improvement of the Langmuir expansion (LTSE) [60, 73, 77–80]. This
approach. In the simplest case, one takes approach not only considers equilibrium
into account only pair interactions between adsorbate clusters up to a certain size,
nearest neighbors, which may be attrac- but also various configurations of a
tive or repulsive, respectively. With the cluster with constant size, for example,
assumption of randomly distributed ad- the lattice geometry is taken explicitly
species, which are not correlated, one ob- into account. Exact isotherms within the
tains the Bragg–Williams Approximation lattice gas treatment may also be obtained
(BWA) [60, 63, 73–76]. The interactions by MC simulation [62], usually based on
of the test particle are considered in the an MC algorithm in a Grand Canonical
mean field of all other particles being pro- ensemble [38, 56, 57].
portional to the coverage θ multiplied by The detailed discussion and application
the total number of particles in the lat- of the aforementioned isotherms will be
tice (mean-field approximation or MFA). given in the paragraph on phase transi-
Combination with Eq. (4) leads directly to tions in organic monolayers (Sect. 3.3.4.3).
Frumkin-type equations. It involves the
statistical distribution of adspecies among 3.3.3
energetically uniform sites and neglects Kinetics of 2D Phase Formation and
local fluctuations and correlations. The Dissolution
loop in the isotherm below the critical
temperature Tcrit is related to an artificial 3.3.3.1 Diffusion, Adsorption,
system forced to be homogeneous under Autocatalysis
all conditions. Especially, the coexistence The formation of 2D ordered, steady state
of the two phases with different densi- adlayers at electrified interfaces is the
ties at the equilibrium phase transition result of several complex, nonequilibrium
under conditions of short-range attractive processes. In the simplest case, this
interactions is excluded. The models based sequence involves mass transfer of the
on the BWA are more reliable if the lat- molecules/ions from the bulk electrolyte
eral interactions can be neglected, or if towards the surface, adsorption and/or
the range of interactions is large [52]. The charge transfer, and association at the
MFA breaks down severely under con- electrode surface [15, 16] (Fig. 1).
ditions in which phase transitions occur
and ordered adlayers form [34–36, 54, 56, Mass transport controlled by diffusion The
60, 73]. limiting case of nonfaradaic adlayer for-
Isotherms based on the so-called mation controlled exclusively by diffusion
quasi-chemical approximation (QCA) are as r.d.s. was discussed by Koryta [81] and
a significant improvement over the Delahay [82]. The time dependence of the
MFA [77, 78]. This gives an exact treatment surface excess , in case of semiinfinite
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 389
Fig. 1 Schematic
representation of the interfacial
equilibria involved in the A
formation of ordered A
monolayers.
A ad A A
A
A ad A
Γ cx = 0 δc /δx c
x
III II I
linear diffusion, is obtained from Fick’s The former yields, in the limit of a
first law according to negligible rate of desorption,
t
∂c θ = 1 − exp(−kad cx=0 t) (7)
(t) = D dt (5)
0 ∂x x=0
where kad is the adsorption rate.
with x as the distance from the plane This theory was extended by Schuh-
of adsorption at the electrode. If the mann to an adsorption process obeying
rate of adsorption is sufficiently high, so a Frumkin isotherm [91]. The same treat-
that the subsurface concentration cx=0 ment is also valid for interfacial faradaic
is zero during the phase formation, one reactions [90, 92].
obtains [82, 83] Several groups derived approximate
expressions for the coupling of diffu-
2c Dt sion/adsorption [85, 87, 88, 93–96] or
θ= = (6) diffusion/charge transfer-controlled pro-
m m π
cesses [2, 92, 97] either in the time or in
m is the maximum surface excess the frequency domain.
of the condensed phase. Equation (6)
is formally equivalent to the well- Autocatalytically controlled surface reactions
known Cottrell equation for faradaic pro- Michailik and coworkers [98] proposed an
cesses [84]. Diffusion-controlled adsorp- autocatalytic surface process representing
tion in the frequency domain was treated the time dependence of formation of a 2D
theoretically and experimentally by Me- condensed film of triphenylethyl phospho-
lik–Gaikazyan [85, 86], Lorenz [87], and nium sulfate ((TPEP)2 SO4 ) deposited at a
Armstrong [88]. mercury–electrolyte interface [99]:
∂θ
Adsorption control The kinetics of the ad- = ka (1 − θ)l θ p (8)
sorption step in the absence of mass trans- ∂t
port control was treated by Lorenz [87] l and p are integers (l, p ≥ 0), with l as the
and Delahay [89, 90] assuming either a reaction order with respect to the initial
Langmuir- or a Temkin-type formalism. state and p the order with respect to the
390 3 Specific Adsorption
products. The analytical rate equation θ(t), Frenkel [112]. The total Gibbs energy re-
derived with l = 2 and p = 1, was shown quired to create a 2D cluster by deposition
to describe quantitatively the deposition of of N particles (ions, molecules) from solu-
(TPEP)2 SO4 onto a mercury electrode [98]. tion is:
Lorenz [100] combined diffusion, intrin-
sic adsorption, and surface association GT (N ) = GP (N ) + GPB (N ) (9)
as separate contributions and developed
The first (favorable) term represents the
the first model, which attributes a slow
free energy gain due to the formation of the
(dynamic) adsorption step to the 2D asso-
new phase. The second (unfavorable) term
ciation of adsorbed species at an electro-
accounts for the creation of a new phase
chemical interface.
boundary. GT (N ) reaches its maximum
The above-mentioned approaches suffer
for a critical particle number Nc . Clusters
from a conceptual difficulty assuming that
with a smaller number of particles than Nc
the adsorbed monomers are distributed
will tend to dissolve, while larger clusters
homogeneously at the electrode. This is
will grow further. The relation (9) is valid
not unambiguously valid in case of 2D or-
for any arbitrary geometrical form of the
dered ML (cf. experimental section 3.3.4).
cluster [113]. The special case of a circular
On the basis of i–t transients for 1 M
patch of radius r and height h shall be
pyridine in 1 M NaOH, Armstrong [101] considered in more detail.
showed that the formation of the adsorbate Phase formation controlled by a charge
film proceeds according to a 2D nucle- transfer reaction Men+ + ne− ↔ M across
ation and growth mechanism, analogous the interface, such as 2D electrocrystal-
to faradaic processes such as electrocrys- lization [114, 115], UPD [2, 22, 116], and
tallization and the deposition of anodic 2D anodic passivation [117, 118] gives the
films [17]. This approach accounts for the following expression for GT
inhomogeneous distribution of adspecies
at the electrode–electrolyte interface, and πr 2 h
shall be discussed in the following sections GT (r) = neo η + 2πrε (10)
v̄
in more detail.
v̄ is the specific particle volume related
3.3.3.2 Nucleation and Growth to the number N of deposited particles
Mechanisms by N = (πr 2 h)/v̄. η = (E − Erev ), with
3.3.3.2.1 Nucleation Erev being the equilibrium potential, is
Homogeneous nucleation – concept of the the overvoltage and represents the driving
critical cluster A metal surface that is force of the faradaic process. ε is the line
homogeneously flat, like mercury or quasi- tension, which has to be overcome by
perfect silver single crystals [2], offers no the expansion of the nucleus [118]. Metal
specific adsorption sites. In analogy to 3D deposition can occur only if η is negative,
phenomena, the nucleation of a new 2D so that the Gibbs energy GT (r) of a
phase may be treated on the basis of cluster as a function of its size r first rises,
the classical nucleation theory (CNT) as reaches a maximum and then decreases
formulated by Gibbs [102] and further de- (Fig. 2). The critical radius rc and the
veloped by Volmer [103, 104], Farkas [105], critical free energy GTc (r) are given by
Stranski and Kaishev [106–109], Becker εv
rc = − (11a)
and Döring [110], Zeldovich [111], and neo η
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 391
∆G Tc(r )
with Erev being the equilibrium poten-
tial at which γ # = 0 [3, 119]. γ # is
a nonlinear function of the applied elec-
rc r trode potential. The free energy GT c (r)
and radius rc of the critical cluster are
derived as
πε2
GT c = − (14a)
γ #
π r 2hneoη/v and
ε
rc = − (14b)
γ #
J = Zα ∗ cnc α1
S + M1 ←−→ S
βs1
− GT c
= Zα ∗ c1 exp (16) − M1 + M1 . . . S − Mi−1
kT
αi
+ M1 ←−→ S − Mi + · · · (17)
where α ∗ is the rate at which the βsi
monomers are added to the critical clus-
ter of nc atoms, ions, or molecules; Instead of the steady state of (quasi)
cnc is the equilibrium concentration of homogeneous nucleation, during which
critical clusters, and Z is the nonequi- nuclei appear at constant rate, both the
librium Zeldovich factor [111, 121]. The total number of critical nuclei and the
equilibrium concentration of critical clus- duration of the nucleation process are
ters is related to the concentration of confined by the maximum number of
monomers, c1 , by the Boltzmann ex- preexisting sites No . If there is a uniform
pression with the energy barrier given probability with time of converting these
by the free energy of formation of the sites into critical nuclei S − Mnc , one
critical nucleus. The period required to obtains the following first-order nucleation
reach a steady state of critical nuclei at law as an approximate solution of the
constant supersaturation, for example, a reaction sequence (17)
stable distribution of clusters of the new
phase controlled by ‘‘monomer’’ transport N (t) = No [1 − exp(−J t)] (18)
in the ambient phase and incorporation
of building units of clusters, is identi- The quantity J is the steady state
fied as initial delay or induction period nucleation rate per active site and may be
t [122–127]. Toshev and coworkers re- expressed analogous to J in Eq. (16) [17,
ported that t in homogeneous nucleation 122, 123]. Two limiting cases of the
is usually longer than for heterogeneous exponential law of nucleation, as given
nucleation [128]. by Eq. (18), are of particular importance:
of the active sites. Staikov and cowork- on lose their physical meaning and the
ers reported that, at constant supersatu- use of atomic forces of interactions be-
ration, J (terrace) < J (step) < J (kink) [2, comes more reasonable. The atomistic
130]. Active sites represent not only surface approach for the calculation of the nu-
imperfections, but may also be gener- cleation rate on supersaturation was first
ated by surface oxidation/reduction, re- introduced by Walton [140] and then de-
construction processes or ad/desorption veloped to a general nucleation theory by
of anions or solvent molecules [131, 132]. Stoyanov and Milchev [141, 142]. These
Fleischmann and coworkers [17, 133] authors considered the excess free energy
and Retter [134] pointed out that the of creation of a new ‘‘surface’’ (second
combination of active intermediates to term in Eq. (9)) by the difference be-
form a critical nucleus at a specific site may tween the individual binding energies of
lead to the power law of n–step nucleation N monomers (atoms, molecules) in the
bulk, N 1/2 , and the dissociation ener-
N (t) = J No t n (21) gies of an N –mer cluster by including the
binding energies of the monomers in the
J represents the modified nucle-
cluster as well
as the interaction with the
ation rate.
substrate, i . Substrate-induced strain
In case of homogeneous nucleation,
is accounted by the introduction of a
No may be attributed to the number of
strain energy per monomer, εM , which
available lattice sites [135]. The nucleation
yields
mechanisms described earlier assume that
the surface concentration of monomers re-
mains constant throughout the transition. GPB (N ) = N 1/2 − i + εM N
The formation of 2D condensed films of 5- N
bromocytosine [136, 137] on mercury and (22)
anodic adenine-mercury complexes [138] Applications of this atomistic approach to
are examples of the so-called ‘‘truncated’’ metal deposition on foreign substrates are
nucleation. This mechanism is based on summarized in [2]. Buess–Herman [143]
the condition that the formation of critical pointed out that, in contrast to charge
nuclei and their competition with growth transfer–induced 2D adlayers, critical nu-
and/or ad/desorption processes tend to clei of surfactant molecules are quite large,
decrease the available monomer concen- often being composed of more than 100
tration with time. At a certain moment, molecules. In these cases, the CNT is still
the nucleation process ceases. valid.
Atomistic approach – small cluster model of 3.3.3.2.2 Growth The expansion of a su-
nucleation With increasing supersatura- percritical nucleus through continued in-
tion the number of monomers constituting corporation of monomers is called growth.
the critical cluster reduces up to a few If the resulting new phase is isotropic,
atoms or molecules, and in some partic- as for instance in the case of a cir-
ular cases of nucleation on active sites, cular island, a single, time-independent
this number was found to be close to growth rate, kG , suffices to characterize
zero [2, 139]. Macroscopic quantities, such the process. For anisotropic growth, there
as surface, surface free energies, and so will be several growth rates [2]. For 2D
394 3 Specific Adsorption
growth, expansion of the clusters is pos- coefficient [145, 146]. The corresponding
sible only at its periphery. Referring to coverage function is given by
the elementary process involved on an ide-
ally smooth surface, one may distinguish
∂θ 2πhA2 Dρ
two main mechanisms: (1) growth con- = = K1 (27)
trolled by the rate of monomer incorporation ∂t MSE m
and (2) mass transport (surface diffusion)-
controlled growth [17, 118]. When more than one critical nucleus is
Assuming that the incorporation is the formed, competition between the grow-
rate–determining step, the number N (t) ing centers for the limited areas SE takes
of monomers belonging to an isolated place. Growth ceases at the point where
circular cluster of height h obeys the these patches merge, domain bound-
equation aries are created, coalescence may often
be ruled out [147]. Computer simulations
∂N (t)
= 2πkG r(t)h (23) predict for the 2D growth controlled
∂t
by direct incorporation of monomers
with (Eden model), rather compact patches
kG
r(t) = t (24) with fractal dimensions close to 2 [148].
ρ
In situ scanning tunneling microscopy
ρ is the number of particles per unit (STM) experiments with hydrogen-bonded
area [17]. Normalization to the real sur- 2D condensed films of purine bases
face area SE and to the maximum on defect-free HOPG-surfaces confirmed
surface excess of the condensed phase these findings [149]. In the diffusion-
m yields the fractional coverage of this limited aggregation model, proposed by
phase as Witten and Sander, the resulting clus-
2M ter has an open structure with fractal
∂θ 2πhkG
= t = K1 t (25) dimensions of 1.66 [150]. More complex
∂t ρSE m mechanisms involve anisotropic and/or
The first in situ atomic-scale visualization multistep growth, induction periods of
of this growth mechanism was recently growth [117, 133, 151–154], as well as
demonstrated for the incorporation of defect-mediated processes [2, 17]. The case
chloride ions at kink positions of the of 1D growth was treated quantitatively
c(2 × 2) chloride adlayer on Cu(100) in in the formation of surface hemimicelles
0.01 M HCl [144]. of amphiphils [155]. Lorenz and cowork-
If the rate of advance of the growing ers pointed out that type and density of
circular center is controlled by symmetrical surface defects, such as monatomic steps
hemicylindrical (surface) diffusion about on single-crystal surfaces, influence dra-
an axis perpendicular to the 2D nucleus matically the steady state structures as
one obtains well as the kinetics of 2D phase forma-
√ tion processes. They demonstrated, for
r(t) = A Dt (26)
instance, that barrierless nucleation and
where A is a constant depending on 2D growth on an equidistantly spaced
potential, molecular weight, and density stepped surface predicts rectangular i–t
of the monomer, and D is the diffusion transients [2].
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 395
40
(a) (b) (c)
30
[nA]
20
−i
10
0
0 10 20 30 40 50 60
t
(a) [s]
Fig. 3 (a) Current–time curves
associated with the formation of single
monolayers on a rectangular
quasi-perfect Ag(100) electrode in 6 M 4
AgNO3 at 45 ◦ C and an overvoltage of
2 mV, as reported by Bostanov and 3
−ln Po
directly to the steady state nucleation rate 6 mV is shown in Fig. 3(b). The nucleation
J . The statistical analysis is based on the rate is strongly potential-dependent with
Poisson distribution: ln J ∼ 1/η (cf. Eqs. (11 and 16)) [156–159].
The shape of the mononucleation tran-
µN exp(−µ) sient is directly related to the absolute rate
PN = (28)
N! of growth, since no overlapping of nu-
where clei takes place. Figure 3(a) illustrates the
isotropic growth of an expanding circular
µ = J SE t; J in (m−2 s−1 ) (29) silver nucleus. Growth stops at the edges
of the rectangular substrate surface. The
PN represents the probability of having growth rate constant was found to be of
N nuclei formed at a time t, where t the order of 1 cm s−1 V−1 [156–159]. The
is a continuous variable, whereas N is carefully chosen geometry of the silver
discrete: N = 0, 1, 2, and so on [123]. template provides, in addition, direct ac-
Usually one evaluates the probability of cess to the location of the initially formed
observing no nucleation event, Po , dur- critical nucleus.
ing a given time interval t. These data The growth of a 2D condensed organic
are experimentally accessible and rather film created by mononucleation on a
unambiguous [123, 156–164]. According hemispherical mercury electrode is often
to Eq. (25), ln Po should be directly pro- expressed in terms of a growing spher-
portional to t, with the proportionality ical disk by expansion of its periphery
constant J (Fig. 3b). (Fig. 4a) [160, 164]:
Typical examples of mononucleation are
∂q kG (t − to )
the deposition of silver on electrolytically i= = 2πrqm kG sin
∂t r
grown Ag(100) crystals (Fig. 3) [156–159],
and the formation of 2D condensed or- (30)
ganic films, such as isoquinoline [160], where r denotes the radius of the electrode,
thymine [163], or coumarin [162–164] on qm the monolayer charge, kG the growth
mercury electrodes. These experiments rate, and to the onset of growth. Figure 4(b)
demonstrate that the rate of nucleation shows a set of experimental and simulated
is more strongly dependent on the ap- transients for coumarin on mercury [164].
plied supersaturation than the growth The stochastic nature of the experiment
rate. Extreme care has to be taken in is represented by the scatter of the
mononucleation experiments to avoid arti- mononucleation transients along the time
facts from edges or defect contributions of axis. The shape of the i–t traces is identical
the substrate or catalytic effects of impuri- because of the use of a small, spherical
ties [161, 162]. (mercury) electrode.
Figure 3 illustrates, as just a small
section of a train, three individual i–t tran- Polynucleation At high supersaturation,
sients for the deposition of a single silver many nuclei are created independently.
monolayer on a dislocation-free Ag(100) The completion of the transition proceeds
surface having a rectangular cross section. by subsequent growth of a large number
The formation of the critical nuclei is in- of nuclei. The resulting transients, for
dicated by arrows. The corresponding plot instance triggered by a potential step, are
of −lnPo (t) versus t for an overvoltage of no longer stochastic, but deterministic,
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 397
1.5
1.0
[µA]
0.5
i
0
−0.5
0 0.5 1.0 1.5 2.0 2.5
t
(a) (b) [s]
Fig. 4 (a) Scheme illustrating the growth of a single nucleus
according to constant radial growth. (b) Stochastic current–time
transients obtained for 0.5 M MaF + 5 mM coumarin at a hanging
mercury drop electrode (HMDE) following a single potential step
from E1 = −0.400 V to E2 = −0.520 V. The lines were drawn using
Eq. (30) with kG /r = 17.5 s−1 , qm = 108 nC, and appropriate values
for to (reprinted from Ref. [164], copyright 1995 by Bulgarian
Chemical Society).
and do not depend on the shape and This statistical law makes it possible to
area of the electrode surface. In the initial correlate single cluster behavior to that
stages of formation of the new phase, of a collection of many clusters and,
the individual centers can be assumed to alternatively, to derive nucleation and
grow rather independently of each other. growth laws from observations involving
The extended fractional coverage θx , which many overlapping clusters. The validity of
represents the hypothetical coverage of the Eq. (32) requires a random distribution of
unhindered growth of individual nuclei, a large number of clusters, in which, each
is obtained by solving the convolution must be small with respect to the total
integral [17, 134, 165] available electrode area.
The combination of Eqs. (31 and 32)
∂θx ∂N with the exponential law of nucleation
= f1 (t − x) dx (31)
∂t ∂t t=x (Eq. (18)), and growth of circular 2D clus-
ters by direct incorporation of monomers
with ∂N /∂t given by the respective
(Eq. (25)) yields, as an example [172, 173]
law of nucleation (cf. Eq. (18–21)) and
f1 (t − x) representing the growth (cf.
2t
Eq. (25, 27)). The extended coverage θx θ = 1 − exp −2K1 No t 2 −
J
and the ‘‘true’’ coverage θ, as obtained
2
in reality by internuclear collision and + 2 [1 − exp(−J t)] (33)
overlap of the growing centers at an ad- J
vanced stage of the transition, are related
and the limiting cases
within the framework of the Avrami theo-
rem [166–171] θ = 1 − exp(−2K1 No J t 2 )
θ = 1 − exp(−θx ) (32) for J −−−→ ∞ (33a)
398 3 Specific Adsorption
2K1 No J 3 The same strategy was applied in the
θ = 1 − exp − t derivation of rate equations for n-step nu-
3
cleation according to a power law (cf.
for J −−−→ 0 (33b) Eq. (21)) [133, 134], the combination of
nucleation laws with anisotropic growth
for instantaneous or progressive one-step regimes [153], as well as truncated nucle-
nucleation [3, 17, 118]. ation due to time-dependent concentration
Examples of the kinetic mechanisms gradients of monomers [136]. MC simu-
represented by Eq. (33) and its lim- lations verified that the Avrami theorem
iting cases are the 2D formation of is valid for instantaneous [184], progres-
organic and ionic monolayers of cy- sive [185], and n-step nucleation according
tosine [174], 5-bromocytosine [173], iso- to a power law [184–187].
quinoline [175], coumarin [164], guani- The above-mentioned expressions of
dinium nitrate [176], for chloride [177] polynucleation and growth processes, as
and hydroxide [178] and several other so- triggered by single potential step experi-
called anodic films on mercury [179, 180], ments E1 → E2 , may be represented by
as well as the transition between the the general Avrami equation [15, 16, 188]
√ √
( 3 × 3) and the (1 × 1) Cu–UPD layer
on Au(111) [181]. θ = 1 − exp(−bf t m ) (35)
In case of surface diffusion-controlled bf is a constant incorporating both rates
growth (Eq. (27)), one obtains accordingly of nucleation, J , and growth, kw ; m is re-
lated to the sum of the dimension and the
θ = 1 − exp − 2K1 No time exponent in the nucleation law. The
values of m provide a set of diagnostic cri-
1 teria to determine possible mechanisms
× t − [1 − exp(−J t)] (34) of the nucleation and growth processes
J
involved (Table 1). Unfortunately, there ex-
and ists no unique relation between m, bf ,
and the actual kinetic mechanism on the
θ = 1 − exp(−2K1 No t) basis of single potential step experiments.
However, in carefully designed secondary
for J −−−→ ∞ (34a) growth double potential step experiments
θ = 1 − exp(−K1 No J t 2 ) (E1 → E2 → E3 ), one may separate nucle-
ation and growth exploiting their different
for J −−−→ 0 (34b) dependence on supersaturation [156, 157,
189, 190]. The rate of nucleation is a much
Phase formation kinetics based on sur- steeper function of potential than the rate
face diffusion-controlled growth processes of growth. The program sequence starts
have been suggested for 2D monolay- with a potential pulse E1 → E2 of large
ers of camphor-10-sulfonate on mer- amplitude and short duration τ to create
cury [182], physisorbed uracil films on a fixed number of critical nuclei, while
Au(hkl) [183], as well as the commensu- ensuring their insignificant growth. The
√
rate/incommensurate transition c( 2 × second step, E2 → E3 , is chosen such
√ √
2 2)R45◦ → c(p × 2 2)R45◦ of bromide that (1) further nucleation is unlikely and
on Au(100) [51]. (2) only preformed nuclei start to grow
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 399
1 I SD 154
1<m<2 EL SD 51, 154, 182, 183
2 I DI 173, 181
P SD
2<m<3 EL DI 173, 174
3 P DI 16, 164, 175, 176
m>3 PL DI or SD 134
0.15
1 2 3 4 5
0.10
[Fm−2]
C
0.05
0
0 0.5 1.0 1.5 2.0 2.5
t
[s]
Fig. 6 Capacitance transients for 0.5 M aqueous NaF + 0.5 mM
coumarin following a single potential step from E1 = −1.000 V
(inside the capacitance pit III) to E2 = −0.4265 V at different
waiting (aging) times to at E1 : (1) 2.5s, (2) 5.0s, (3) 10s, (4) 20s,
(5) 50s. The experimental transients could be well represented by
Eq. (40) in combination with Eq. (36) (reprinted from Ref. [164],
copyright 1995 by Bulgarian Chemical Society).
402 3 Specific Adsorption
(a)
Tab. 2 Adlayer structures of halides on Au(hkl) and Ag(hkl), adapted from Ref. [21]
basis of selected examples with halide given in Ref. [21]. Selected results of these
or oxyanion–containing adlayers, it will studies, as obtained by in situ STM and/or
be demonstrated that the observed struc- SXS for ordered halide adlayers on Au(hkl)
tures and 2D phase transitions can be and Ag(hkl), are summarized in Table 2.
understood as a result of competing adsor- Before the detailed discussion of specific
bate–adsorbate and adsorbate–substrate examples, several general trends shall be
interactions. pointed out.
Ordered halide adlayers are only ob-
3.3.4.1.2 Phase Formation in Halide served above a critical potential Ec , which
Adlayers often represents rather high anion cover-
ages. The potentials of adlayer formation
In Situ Results on Adlayer Structures For and phase transitions between different
Cl− , Br− , and I− ordered structures have ordered adlayers shift ∼60 mV per decade
been observed in in situ studies on single- of halide concentration. The potential
crystal electrodes of the face centered cubic ranges of the ordered phases depend
(fcc) metals Au [51, 252–264], Ag [53, 55, strongly on Epzc and the strength of
251, 265–269], Cu [270–272], Pt [23, 251, metal–halide interactions. The ordered
273, 274], [275–277], Pd [278], Rh [279], phases are stabilized by repulsive ad-
and Ni [280]. A comprehensive review is sorbate–adsorbate interactions. Halides
406 3 Specific Adsorption
substrate are different, the resulting phase The iodide species are arranged in a
is a compromise between a simple com- centered rectangular unit cell with sides
√
mensurate structure, which minimizes the 3 and p, where p decreases with in-
interface energy (determined by the cor- creasing potential. The adsorbate coverage
rugation in the adsorption energy), and changes from about 0.355 to 0.40 in
a uniformly compressed incommensurate a 0.56 V wide potential range. At the
√
phase, which is favored by the elastic lat- most positive extreme of the c(p × 3)
eral interactions in the adlayer. Frenkel phase, the structure corresponds to the
√
and Kontorova [68] and Frank and van c(5 × 3) high-order commensurate unit
der Merwe [69] showed the lowest energy cell with a nearest-neighbor iodide spac-
state corresponds to a system of com- ing of 0.439 nm. Qualitatively, the SXS
mensurate regions separated by domain data show two discontinuous changes in
walls (also called solitons or misfit dis- coverage, indicating first-order phase tran-
√ √ √
locations). When the domain walls are sitions for ( 3 × 3)R30◦ ↔ c(p × 3)
√
diffuse, such that the adsorbed species are and c(p × 3) ↔ rotated-(hex) (Fig. 8).
equally spaced along the incommensurate The phase behavior of Ag(111)/I− is com-
direction, the structure is referred to as pared with that of iodide on an Au(111)
a uniformly compressed incommensurate surface, which exhibits an almost identical
phase. For these phases and dislocation substrate geometry. Both incommensu-
√
structures, the displacement of the adsor- rate phases, the c(p × 3) and the ro-
bates from the commensurate positions tated – (hex) phase, were observed by in
can result in a long-range vertical modula- situ STM, as well as SXS [252, 255, 258].
√ √
tion (Moire-pattern) of the adlayer, which The commensurate ( 3 × 3)R30◦ phase
is often observed directly by STM [252, 253, was only found in a limited potential range
258, 261, 267]. by STM and ex situ LEED studies [258],
A whole sequence of adlayer structures which might suggest that this structure
is illustrated by the system Ag(111)/I− , exists only as a short-range ordered phase.
where ordered phases are observed in The potential-dependent coverage of the
a wide range of potentials and surface incommensurate phases on Au(111) is also
coverages due to the strong adsorption plotted in Fig. 8 (filled circles). For com-
and high polarizability of iodide [259, 267]. parison, the potential scale for the data on
Starting at negative potentials in which Au(111) was shifted by −0.50 V to account
the halide layer is disordered, a commen- for the difference in Epzc of Ag(111) and
√ √ √
surate ( 3 × 3)R30◦ , an UI c(p × 3), Au(111) [259].
and an incommensurate rotated hexago- The phase behavior of iodide on Ag(111)
nal structure were found with increasing and Au(111) is an elegant example of
potential, independently in SXS- and STM the stepwise loss of commensuration with
studies [259, 267] (Fig. 8). The coverage increasing potential accompanied by two
θ increases from 1/3 in the commensu- first-order phase transitions. This phase
rate phase with iodide species residing sequence has been predicted by Bak and
in threefold hollow sites to 0.442 at the coworkers [286] using Landau theory and
√
most positive potentials. The c(p × 3) Kadar [287] using a generalized three-state
structure is obtained by uniaxial com- Potts model.
√ √
pression of the ( 3 × 3)R30◦ adlattice The orientation of the adlayers in
along the (11̄0) direction of the substrate. the incommensurate hexagonal phases
408 3 Specific Adsorption
0.40
θ
0.38 B
0.36
A
0.34
(c)
Coverage
0.04 M KClO4 + 0.01 M KBr. (b) θtot 0.3
(referenced to the Ag(100)-(1 × 1) Chrono (θ tot)
0.2 I
lattice) for three KBr concentrations θSXS
in (0.05 − x) M KClO4 on Ag(100). The
solid lines represent the 0.1
Monte-Carlo-simulation isotherms as
calculated by Rikvold and coworkers in 0
Ref. [291] (reprinted with permission −1.4 −1.2 −1.0 −0.8 −0.6 −0.4
from Ref. [55], copyright 2001 by
Elsevier Science Ltd.). ESCE
(a) [V]
0.5
0.4 10 mM
1 mM
Coverage
0.3 0.1 mM
0.2
0.1
0
−1.4 −1.2 −1.0 −0.8 −0.6 −0.4
ESCE
(b) [V]
0.53
Negative
c ( 2 × 2 2)
0.51 Positive
Ec
0.49
0.30 0.40 0.50 0.60
ESCE
(b) [V]
4.08 Å and 4.56 Å, respectively. The crys- Ag(100)/Br− [53]), suggests that the elastic
tallographic analysis of the X-ray peak interactions between the Br− adsorbates,
intensities suggests a registry of the which favor hexagonal packing, are more
√ √
c( 2 × 2 2)R45◦ adlayer with all adsor- significant than the adsorbate–substrate
bates residing in bridge sites, rather than interaction energy difference between the
a mixed occupation of top and hollow two phases. The corrugation potential for
sites [259]. This result is also supported bromide on Au(100) appears to be weaker
by an in situ STM study of Cuesta and than on Ag(100). Similar trends were
coworkers [261], in which no differences found in halide adlayers on Au(111) and
in the apparent height of the halide Ag(111) [21], as well as in the corrugation
√ √
adsorbate in the c( 2 × 2 2)R45◦ unit potentials extracted from the calculations
cell were observed. The preference for the of Ignaczak [245].
√ √ √ √
c( 2 × 2 2)R45◦ phase, with respect to The commensurate c( 2 × 2 2)R45◦
the higher coordinated, and commonly on phase transforms at E > 0.40 V into
√
fcc(100) surfaces found c(2 × 2) phase (cf. a UI c( 2 × p)R45◦ adlayer, where
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 413
√
the incommensurability ε = 2 2/(p − 1) This behavior is consistent with a cu-
√
continuously varies from 0 (commensurate mulative disorder in the c( 2 × p)R45◦
phase) to 0.12. The nearest-neighbor spac- phase [259]. The results for bromide ad-
√
ing in the c( 2 × p)R45◦ structure is al- sorption on Au(100) under electrochemical
√
ways the commensurate spacing aAu 2 = control indicate that continuous C → UIC
4.078 Å, the next-nearest-neighbor spacing transitions are not restricted to the vacuum
decreases from 4.56 Å in the commensu- environment.
rate phase to 4.13 Å resulting in an almost The kinetics of the phase transitions
undistorted hexagonal lattice at the most of bromide adlayers on Au(100)-(1 × 1)
positive potentials [51, 259]. As the lat- were studied by chronocoulometric poten-
tice compresses the bromide species slide tial step experiments [51]. Current tran-
√
along the ‘‘rails’’ determined by the under- sients indicate for the disorder ↔ c( 2 ×
√
lying gold atoms. The adlayer compression 2 2)R45◦ transitions in both directions
is similar to that in the rotated hexagonal an instantaneous (hole) nucleation pro-
phase on Au(111) [254]. cess coupled with a Langmuir-type ad-
The continuous change of the bromide sorption/desorption mechanism. Rising
coverage with potential upon the transi- transients, which could be modeled with
√ √ √
tion c( 2 × 2 2)R45◦ → c( 2 × p)R45◦ the exponential law of nucleation and 1D or
can be approximated by the power law θ = surface diffusion-controlled growth, were
√ √
0.122 (E − Ec )β + 0.50 with β = 0.40 and obtained for the c( 2 × 2 2)R45◦ →
√ ◦
Ec = 0.375 V. The measured exponent is c( 2 × p)R45 transition when stepping
smaller than the theoretical value β = 0.50 into the potential regime of high incom-
predicted by Pokrovsky and Talapov for mensurability, (the uniformly compressed
the C → UIC transition within a model adlayer phase). The signal-to-noise ratio
of noninteracting domain walls [70]. The in the range of the weakly incommen-
theoretical prediction is only supposed surate phase was too high, and hence
to apply close to the transition, where no direct conclusion on the ‘‘continu-
the domain walls are narrow relative to ous’’ character of this phase transition
their separation. The experimental pre- around Ec (as predicted from the above-
cision does not permit to quantitatively mentioned SXS results) from transient
extract the exponent close to the transi- experiments was accessible. The dissolu-
√
tion, in which the theoretical prediction tion of the UIC ( 2 × p)R45◦ phase is
could be unambiguously tested. At higher characterized by monotonous i–t curves,
which may be related to the consider-
incommensurabilities the bromide cover- √ √
age is not only determined by the proximity able disorder of the c( 2 × 2 2)R45◦
to the C/UIC phase transition, but rather when formed from the incommensurate
by the compressibility of the bromide ad- phase [51].
layer, which may explain the deviations
in the exponent. The width of the X-ray 3.3.4.1.3 Phase Transitions in Adlay-
diffraction (XRD) peaks in the incommen- ers of Oxoanions The specific ad-
surate direction scale quadratically with sorption of oxoanions, such as sul-
the wave vector component in this di- fate/bisulfate, hydroxide, or phosphate
rection, and increase continuously by an considerably affects the shape of voltam-
order of magnitude as the C/UIC tran- metric and capacitance curves [21, 23,
sition is approached from the UIC side. 231–233, 293]. The presence of sharp
414 3 Specific Adsorption
Au(111)-( 3 × p ) Au(111)-(1 × 1)
Disordered ( 3 × 7)
3
I II III
2
P1 P2
[µA cm−2]
1
j
−1
P2′
−2
−0.4 −0.2 0 0.2 0.4 0.6 0.8 1.0
ESCE
(a) [V]
(1 × 1) ( 3 × 7)
1.6
log tmax
log imax
[ms]
[µA]
0.750 0.816
1.2
−0.4 1.6
0.812 1.4
−0.8 1.2
0.800 0.808 0.816
[µA]
0.8
i
ESCE
0.808 [V]
0.804
0.4
0.800
0.796
0
0 100 200 300 400 500
t
[ms]
Fig. 13 Current transients i(t) for Au(111), miscut < 0.5◦ , in 0.05 M H2 SO4
obtained after a single potential step from E1 = 0.75 V (region II) to various
final potentials in region III. The experimental traces are given as individual data
points, the solid lines represent theoretical curves calculated with the
parameters of the numerical fit to a model combining (a) an adsorption process
(Eq. 7) and (b) one-step nucleation according to an exponential law with
surface diffusion-controlled growth (Eq. 34), (reprinted from Ref. [299].
Copyright 1997 by VCH Verlagsgesellschaft mbH Weinheim).
to a substrate of its own kind, for example, deposit [116, 319]. This correlation is not
the chemical potentials of monolayer valid at single-crystal electrodes [320–322].
and bulk are different. In comparison New theories of UPD involving semiem-
to molecular adsorption, one observes a pirical models, DFT- and band structure
significant charge transfer (faradaic reaction calculations taking into account contri-
!) between the deposited metal ion and the butions of the electronic subsystem (re-
polarized electrode [116]. lated to work function differences) and
The UPD is often demonstrated in of the ionic cores of the adsorbate, have
cyclic voltammetry. The formation (dis- been developed to model the trends ob-
solution) of the first monolayer is seen served on single-crystal electrodes (cf.
by pronounced current peaks at E > Ref. [323]). These quantum statistical ap-
Erev , the bulk deposition occurs at E < proaches are complemented by statistical
Erev . For many polycrystalline surfaces, mechanical models [67] and (dynamic)
it has been shown that the differences Monte Carlo simulations [197] on UPD
in peak potentials for the oxidative dis- systems.
solution of metal monolayers and bulk The use of single-crystal electrodes
correlate linearly with the difference in in UPD–studies revealed pronounced
work function, , of substrate and structure specificity in metal monolayer
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 417
0.8
< 0.5°
0.6
0.750 V 0.804 V
[µA] 0.4
0.7°
i
3.0°
1.6°
0.2
4.7°
0
0 100 200 300 400 500
t
[ms]
Fig. 14 Typical set of i–t transients with E1 = 0.75 V/E2 = 0.804 V for
stepped Au(111) s crystals with various miscut angles (as indicated) in
0.05 M H2 SO4 . The solid lines represent calculated curves as obtained
with the parameters of the numerical fit to an adsorption-nucleation
model (cf. Fig. 13), (reprinted from Ref. [299], copyright 1997 by VCH
Verlagsgesellschaft mbH Weinheim).
formation as was shown in the early coadsorption onto the exact determination
work of Schultze [320–322], Lorenz [324], of coverage-potential isotherms was ex-
Bewick and coworkers [48–50], or plored by the twin-electrode thin-layer
Adzic [325]. Meanwhile a considerable technique [329] and studies with the quartz
amount of data is available, which crystal microbalance (QCM) [330]. Multi-
illustrate the role of electrode material step adsorption isotherm was taken as a
and surface crystallography, the influence first evidence for ordered UPD films [2,
of coadsorbed anions and/or solvent 22, 320–322]. The kinetics of adlayer
molecules, as well as intermetallic deposition/dissolution, as monitored in
electronic and ionic interactions on (1) chronoamperometric potential step ex-
the formation/dissolution and (2) on periments, was phenomenologically in-
the structural properties of these metal terpreted with homogeneous diffusion-
monolayers (cf. reviews in [2, 22, 116, adsorption models (cf. summary in [2])
326–328]). or as first-order phase transitions con-
UPD of metals has been extensively trolled by nucleation and growth [48–50].
studied with a variety of methods. Cyclic Optical properties of the metal mono-
voltammetry is frequently employed to layers were studied in the visible and
phenomenologically describe the phase be- near UV-range by electroreflectance spec-
havior in UPD systems and to determine troscopy [25] and second harmonic gen-
adsorption isotherms assuming full dis- eration [331]. The use of ex situ UHV-
charge of the metal ions [2, 116, 326–328]. techniques (LEED, reflection high-energy
The role of partial discharge and anion electron diffraction (RHEED), XPS, UPS,
418 3 Specific Adsorption
20
A2
[µA cm−2]
10 B2
j
B1 A1
Fig. 15 Cyclic voltammogram for Cu
−10
UPD on a well-ordered Au(111)
1.0 electrode in 0.1 M H2 SO4 + 1 mM
CuSO4 , scan rate 1 mV s−1 (reproduced
from L. Kibler, Preparation and
Coverage
( 3 × 3)R30°
Characterization of Noble Metal
0.6 Single-Crystal Electrodes. Copyright
Random 2000 by International Society of
adsorption Electrochemistry), and electrochemically
0.2 (1 × 1) derived Cu coverage (normalized charge
referring to one complete Cu UPD ML)
0.0 0.1 0.3 0.4
as a function of potential, determined by
0.2
potential steps in the positive direction.
ESCE The Cu adlayer structures are also
[V] shown (adapted from Ref. [360]).
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 419
electrochemical methods, such as voltam- The coverage was estimated to 0.2 ML em-
metry [320–322, 337, 338], chronocoulom- ploying QCM [351, 352] and chronocoulo-
etry [339, 340], (2) in situ techniques such metric [339, 340] measurements. These
as STM [341–343], and AFM [344–346], experimental results are corroborated by
FTIR [199, 347], X-ray absorption spec- a recent theoretical work of Blum and
troscopy [348, 349], SXS [350, 351], and coworkers [67, 356–358] and Rikvold and
QCM [352, 353] and (3) ex situ UHV tech- coworkers [197, 359].
niques (LEED, RHEED, AES) [242, 354,
355]. Current transients and adsorption kinetics
The cyclic voltammogram of Cu UPD The shape of the current peaks, the
on Au(111) shows two well-defined hysteresis in the peak positions between
pairs of current peaks A1 /A2 and the cathodic and anodic potential sweeps
B1 /B2 corresponding to energetically dif- (particularly for B1 and B2 ) and
ferent adsorption/desorption processes lattice gas simulations [197, 359] suggest
(Fig. 15) [320–322, 337]. In the first step that monolayer formation occurs via
(peak A1 ), the transition between ran- several first-order 2D phase transitions.
domly adsorbed copper and (hydrogen) Single potential step experiments revealed
sulfate ions into an ordered layer of monotonously falling transients for peak
copper atoms (electrosorption valency √
A (disordered Cu adlayer → ( 3 ×
γ ∼ 1.8 [339, 340]) and coadsorbed sul- √1 ◦
3)R30 ) and rising transients for peaks
√ √
fate ions takes place. The resulting
√ √ A2 (( 3 × 3)R30◦ → disordered Cu
( 3 × 3)R30◦ structure was first ob- √ √
adlayer), B1 (( 3 × 3)R30◦ → (1 × 1)),
√ √
served by ex situ LEED and RHEED and B2 ((1 × 1) → ( 3 × 3)R30◦ )
experiments [354], and later confirmed (Fig. 16) [181]. If the potential step width
by in situ SXS [350], STM [341–343], is increased (‘‘higher overpotentials’’),
and AFM [344]. QCM [352, 353], chrono- the surface reaction becomes faster,
coulometric [339, 340], and FTIR- and the current maxima are shifted
measurements [199, 347] proved the to shorter times and higher current
coadsorption of sulfate. The SXS study densities. At low overpotentials, the entire
of Toney and coworkers [350] revealed phase transition process can be slowed
that copper atoms form a commensu- down into the timescale of seconds
rate honeycomb lattice (occupation of or minutes. The transients for peaks
threefold hollow sites, 2/3 ML coverage), A2 , B1 , and B2 could be modeled
while sulfate species are adsorbed in the by instantaneous (hole) nucleation and
centers (1/3 ML coverage) above the plane 2D growth (Bewick–Fleischman–Thirsk
of the copper atoms. Three oxygen atoms (BFT) – theory) in parallel to a Langmuir-
of each sulfate species are chemically type adsorption/desorption process of
bound to Cu atoms, and one points to- species, with both proceeding at different
wards the electrolyte [350]. In the second electrode sites [181]. Models based on site-
step (peak B), a pseudomorphic mono- unspecific coupling of adsorption and
layer of Cu (1 × 1) on Au(111) (occupation nucleation [361] or solely on homogeneous
of threefold hollow sites) is formed [343]. adsorption processes with lateral attraction
EXAFS experiments indicate the disor- failed [362].
dered coadsorption of sulfate species on Additional support for the nucleation
top of the copper monolayer [348, 349]. and growth controlled phase formation
420 3 Specific Adsorption
−1200 300
190 mV 237 mV
A1 A2
[µA cm−2]
233 mV
−400 350 mV 190 mV 100
229 mV
0 0
0.0 0.2 0.4 0.6 0.8 0.0 0.5 1.0 1.5
t t
[s] [s]
40 mV 115 mV
−40 90 mV 30 mV 150 150
B1 B2
[µA cm−2]
−20 41 mV
98 mV
50 50
43 mV 90 mV
0 0
0.0 2 4 6 8 0.0 0.5 1.0 1.5 2.0
t t
[s] [s]
Fig. 16 Measured current transients ( ) for (a) Langmuir-like adsorption according to
the two-dimensional phase transitions during Cu Eq. (7) and instantaneous (hole) nucleation/2D
UPD on Au(111) in 0.05 M H2 SO4 + 1 mM growth represented by Eq. (33a) as two parallel
CuSO4 (peaks A1 , A2 , B1 , and B2 in Fig. 15). The processes are also shown (•). (Reprinted from
potential steps are indicated in the figure. The Ref. [181], copyright 1994 by Elsevier Science
results of a least-square fit to a model combining Ltd.)
t = 0.0 s
t = 0.25 s
t = 1.00 s
t = 12.7 s
B1
25 mV
180 mV
t=0s 100 nm
t = 12.7 s
B2
76 mV
37 mV
t=0s 100 nm
40 60
ClO4− Cl − 80 Br −
30 40
40
[µA cm−2 ]
20
20
j
0 0
10
−20
0 −40 (4 × 4)
(5 × 5)
(2.2 × 2.2) −40
−10 −80
−0.2 0.0 0.2 0.4 0.6 0.0 0.1 0.2 0.3 0.4 −0.1 0.1 0.3 0.4
ESCE ESCE ESCE
[V] [V] [V]
Fig. 19 Cyclic voltammograms for UPD of Cu on Au(111) in 0.1 M HClO4 , 0.05 M H2 SO4 , 0.1 M
HClO4 + 1 mM NaCl, and 0.1 M HClO4 + 1 mM NaBr. The copper concentration was always 1 mM,
scan rate 5 mV s−1 . The ordered structures in the medium coverage region are also indicated (with
..
permission by M. Holzle, Ref. [360]).
4 4
Au(100) Au(110)
2 2
[µA cm−2]
0 0
j
−2 −2
−4 −4
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
ESCE ESCE
[V] [V]
Fig. 20 Cyclic voltammograms for UPD of Cu on Au(100) and Au(110) in 0.1 M
H2 SO4 + 1 mM Cu2+ , scan rate 1 mV s−1 .
two steps in 0.40 V > E > 0.17 V. The ma- 3.3.4.2.3 Underpotential Deposition of
jor charge contribution is consumed in the Copper Ions on Pt(hkl) The UPD of Cu
broad, more positive feature. No ordered on Pt(hkl) has been studied by electro-
adlayer has been reported in this poten- chemical techniques [375–386], electrore-
tial region. At potentials negative of the flectance [375], radioactive labeling [387,
sharp current spike, an ordered Cu–UPD 388], IR-spectroscopy [389, 390], electro-
(1 × 1) ML was found [372]. The overall chemical quartz crystal microbalance
deposition process appears to be strongly (EQCM) [391], in situ STM [389, 392–396],
influenced by the presence of monatom- and a variety of X-ray techniques [200, 385,
ically high gold islands, created during 397–401], as well as ex situ UHV-methods
the lifting of the Au(100)-(hex) recon- such as LEED, AES, and XPS [389, 399,
struction [373]. On Au(110) the Cu ML is 402–406].
formed in a broad, single peak adsorption Figure 21 shows a slow scan voltam-
process [374]. mogram of Cu–UPD on a high-quality
40
A2
30 0
[µC cm−2]
−200
qM
20
−400
−600
[µA cm−2]
[V]
B2
−10
B1 A1
(1 × 1) ( 3 × 3) ( 3 × 7)
−20
( 3 × 3)
Pt(111) electrode in 0.1 M H2 SO4 [407]. 390, 402]. During the second UPD peak,
Two pairs of narrow deposition (A1 , B1 ) B2 , the sulfate ions are displaced from
and dissolution (A2 , B2 ) peaks, which the honeycomb center, and a (1 × 1) pseu-
are separated by a small, but distinct domorphic ML is formed with the Cu
hysteresis, are developed clearly. Inte- residing in the threefold hollow sites of
grating the current between 0.70 V and the Pt surface, on top of which the sul-
√ √
0.55 V (RHE) gives a monolayer charge of fate ions arrange in a ( 3 × 7) pattern,
approximately 500 µC cm−2 . The mecha- θ = 0.20 [389, 396, 397.] The reverse pro-
nism of Cu deposition on Pt(111) in the cesses take place during the dissolution
presence of sulfuric acid is described as of the UPD phase upon positive potential
follows: Bisulfate ions adsorbed on the excursion.
√ √
Cu-free Pt(111) surface form a ( 3 × 7) Systematic studies with stepped
structure [301] which gradually transforms Pt(111) s electrodes having (111) terraces
√ √
into ( 3 × 3)R30◦ [389, 390, 402]. Upon and monatomic (110) [386, 407] or
passing the first UPD peak A1 , sulfate ions (100) [379, 380] steps demonstrate that
and Cuδ+ coadsorb in a honeycomb lattice the fingerprint region of Cu–UPD
(2/3 ML of Cu, threefold hollow sites). The depends strongly on terrace size and step
√ √
resulting ( 3 × 3)R30◦ structure was density. At low coverages, Cu adsorbs
observed by in situ STM [390, 395, 396] preferentially on steps [382, 383, 407].
and SXS [408], and is in agreement with Figure 22 illustrates, as an example, typical
ex situ XPS and LEED investigations [389, current versus potential profiles for the
30
B2 A2
[µA cm−2]
20
j
10
D2 C2
0
Pt(111)
Pt(10.10.9)
Pt(776)
Pt(554)
Pt(775)
Pt(332)
Pt(221)
Pt(331)
0.4 0.5 0.6 0.7
E RHE
[V]
Fig. 22 Current versus potential profiles of the stripping of a Cu–UPD ML deposited on
stepped Pt(n n n-2) electrodes as recorded during a positive potential scan from 0.35 V to
0.80 V, 1 mV s−1 , with 1 mM Cu2+ in 0.1 M H2 SO4 [407].
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 427
0.8
0.8 4
0.6
[mA cm−2]
0.4
j
0.6 3
0.2
[mA cm−2]
[mA cm−2]
0.0
0.4 2
j
j
0 2 4
t
[s]
0.2 1
E = 0.65 V E = 0.68 V
E = 0.64 V E = 0.67 V
E = 0.63 V E = 0.66 V
0.0 0
coadsorbed anions, which shift the onset these UPD phases were reported by
of deposition towards more negative Finnefrock and coworkers [200]: At the
potentials according to the following initial potential E1 = 0.20 V the epitaxial
sequence: F− < ClO4 − < (H)SO4 (2)− < Cu(1 × 1) layer is present at the surface.
Cl− < Br− < I− [390]. Rather complex At the final potential E2 = 0.35 V (past
bilayer structures have been reported B2) the incommensurate CuCl structure
(Fig. 24, cf. summary in [22]). The is formed (Fig. 25a). The corresponding
voltammetric profiles represent two-stage transient current, which is related to the
processes. At potentials above the first dissolution of Cu and the reorganization
UPD peak, chloride ions are adsorbed in of the Cl- adlayer, decays exponentially to
a disordered phase. During the first UPD- zero at times well below 1 s. The X-ray
step (disorder–order phase transition) a response at the scattering position (0.765,
Cu ML is formed, which is covered 0, 0.5), which is sensitive to the 2D order
by an incommensurate, close to (4 × of the Cu species within the UPD phase, is
4) Cl− -adlayer. The latter transforms slower and completely different in shape
during the second UPD peak at negative (Fig. 25c). The sigmoidal SXS-transients
potentials, and on top of an epitaxial (1 × were attributed to a 2D nucleation-
√
1) Cu adlayer, into a (2 × 3) phase and-growth process. These experimental
accompanied by partial desorption of results indicate that the processes of
Cl− [402, 405, 412, 413]. The extend of Cu and anion adsorption/desorption and
Cu and Cl− codeposition on Pt(111) s- the formation of the ordered adlayer
electrodes depends critically on terrace occur at rather different timescales.
width and step density [386]. Pioneering This conclusion is not accessible from
time-resolved SXS experiments on the the analysis of just the electrochemical
kinetics of phase transitions between current response! The reversed process,
120
1ML-(1 × 1) ∼ 0.5 ML Cu 1ML-(1 × 1) ∼ 0.5 ML Cu Cu
A2 A2 A2
80
B2
[µA cm−2]
40 B2
j
0
Cu(1 × 1)
B1
−40 B1 B1
A1 A1
Incomm-(hex)
A1 Cl− Incomm-(hex) Br − ( 3 × 3)R30° I−
−80
0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6
E SCE
[V]
Fig. 24 Cyclic voltammograms for UPD of Cu on Pt(111) in 0.05 M H2 SO4 + 1 mM NaCl, 0.05 M
..
H2 SO4 + 0.1 mM NaBr, and 0.05 M H2 SO4 + 0.1 mM NaI (with permission by M. Holzle,
Ref. [360]). The halide adlayer structures at intermediate Cu coverages between the deposition peaks
A1 and B1 , and the Cu coverage within the respective bilayers are also indicated.
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 429
350
Disordered Cl Incomm-(hex) Disordered Cl
EAg/AgCl on Cu(1 × 1) CuCl bilayer on Cu(1 × 1)
[V] 300
250
(a)
200
4
4
2 2
0
[mA]
0 10.0 10.5
i
−2
(b)
−4
(c)
1300
1200
[a.u.]
I
1100
1000
0 10 20 30 40
t
[s]
Fig. 25 Potential step experiment of the transition between the commensurate
Cu(1 × 1) and the incommensurate CuCl UPD adlayer on Pt(111) in 0.1 M
H2 SO4 + 1.0 mM Cu2+ : (a) potential perturbation, (b) current transients (the
inset shows a magnified region of the entire transient) and (c) time dependence
of the scattered X-ray intensity at (0.765, 0, 0.5), which represents a
characteristic diffraction rod of the incommensurate CuCl bilayer (reprinted from
Ref. [200], copyright 1998 by American Physical Society).
D3 D1
j
0 0 0
A3 A1
[µA cm−2]
−20 −60 −40
A2 A2 A2
−12 0.4
15 −0.350 −0.449
−0.350 −0.450 0.3
−8
−0.200 −0.480
10 −0.446
j
−0.430 0.2
−0.450 −4
[mA cm−2]
5 −0.420 −0.443
0.1
−0.390
0 0
10 30 50 70 0 30 60 90 0 0.5 1.0 1.5 2.0 2.5
t t t
(a) [ms] (b) [ms] (c) [s]
Fig. 26 Cyclic voltammograms for UPD of Pb on well-prepared Ag(111) electrodes in different electrolytes. (a) 0.5 M NaClO4 + 1 mM HClO4 + 3 mM
PbClO4 , (b) 0.5 M NaAc + 1 m MHClO4 + 3 mM PbOAc, and (c) 0.5 M NaClO4 + 0.1 M Na2 HCit + 3 mM Pb(ClO4 )2 , scan rate 1 mV s−1 . The lower
panels a, b, and c show for each electrolyte typical current transients of formation of a complete UPD Pb ML. The transients in (a) have been reproduced
from Ref. [130], copyright 1997, NRC Canada. The transients of Pb UPD in citrate-containing solution (panel (c) solid lines) were fitted to a model based
..
on instantaneous hole nucleation combined with a time-dependent rate of growth (with permission by M. Holzle, Ref. [360]).
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces
431
432 3 Specific Adsorption
2D Me
Time
1D Me
Terrace
Terrace
E 3D Me E 2D Me E1D Me 0 nm 20 nm
(a)
Time
Terrace Pit
E3D Me E 2D Me E1D Me 0 nm 80 nm
(b)
Fig. 27 (a) In situ STM line scan plot showing the step decoration by a 1D Pb
phase and the initial stage of 2D phase formation on a stepped Ag(111) substrate
and (b) line scans of formation of a 2D Pb UPD phase near a terrace and a
monatomic deep pit for Pb UPD on Ag(111) in 4 mM Pb(ClO4 )2 + 0.01 M HClO4
(by courtesy of E. Amman from Ref. [434]).
place. The dissolution of the metal adlayer the UPD-range may cause surface alloy-
proceeds in the reverse regime. ing [427].
The sequence of formation/dissolution The completed Pb UPD is metallic, and
of the experimentally observed 1D and 2D represents an incommensurate, hexagonal
Pb-phases is in agreement with theoret- ML that is compressed compared with the
ical predictions based on a Nernst-type bulk metal by 0.1–3.2%, and rotated from
equation [2]: the substrate (011̄)-direction by ±4.5◦ [426,
427, 429–431]. The rotation of the ad-
o RT aPb2+ layer with respect to the substrate lattice
EPb,nD = EPb,3D + ln gives rise to a characteristic Moire pat-
zF aPb,nD
(43) tern as observed in several in situ STM
with the dimensionality n, EPb,3D o the studies [360, 426, 427] (Fig. 28). The in-
formal potential of Pb in the bulk phase, teraction between solvent molecules and
and aPb2+ the Pb activity in the elec- the Pb adatoms does not influence the
trolyte. The activity of the condensed metal structure of the complete ML deposited
phases shift towards lower values with de- in ClO4 − or acetate-containing electrolyte,
creasing dimensionality due to stronger since the UPD phase is essentially iden-
substrate–adsorbate interactions, and in tical to that of vapor-deposited Pb on
consequence, the corresponding equilib- Ag(111) at full coverage [420, 435]. The
rium potentials are shifted in the positive monolayer compression in the vacuum
direction [2]. Extended polarization within experiment (1–2%) is slightly less than for
3.46
Acetate, cathodic
Acetate, anodic
Perchlorate, cathodic
3.44 Perchlorate, anodic
a nn
[A]
3.42
3.40
10 nm × 10 nm 0 50 100 150
EPb/Pb2+
(a) (b) [mV]
Fig. 28 (a) Atomic resolution STM image of a plot. The solid and dashed lines are linear
Pb UPD ML in citrate-containing electrolyte at least-squares fits to the data for perchlorate and
potentials past the main deposition peak A2 acetate, respectively. These lines have slopes of
..
(with permission by M. Holzle, Ref. [360]). 0.424 ± 0.01 Å/V (perchlorate) and
(b) Dependence of the Pb ML near-neighbor 0.420 ± 0.01 Å/V (acetate); their offsets are
distance ann on the electrode potential E for Pb 3.390 Å (perchlorate) and 3.395 Å (acetate).
UPD on Ag(111). The near-neighbor spacing for (Reprinted from Ref. [431], copyright 1995 by The
bulk Pb is 3.501 Å, well above the range of the American Chemical Society.)
434 3 Specific Adsorption
Pb UPD, probably caused by the stronger The order of these 2D systems is produced
Pb–Ag bonds compared to Pb–Pb [319], by a chemical reaction at the interface
although the rotational angle φ is about as the system approaches equilibrium.
the same. Unfortunately, they appear to contain a
The equilibrium change of the ML significant number of pinholes and other
Pb–Pb distance, aPb – Pb with the applied defects [443–445]. Constrains exist also for
electrode potential provides an estimate of monolayers based on the LB technique.
the 2D compressibility of the lead ML These studies are limited to the choice of
√ molecules with a very low solubility in the
aPb−Pb 3 ∂aPb−Pb subphase [438, 439].
κ2D = (44)
zeo ∂E T Alternatively, molecular and ionic mono-
layers can also be obtained on conducting
The experimental value κ2D = (1.25 ± surfaces in an electrochemical environ-
0.05) Å2 /eV is in qualitative agreement ment [3, 15, 16, 20, 143]. This approach
with the results of a simplified model based offers the advantage that formation and
on a 2D free electron gas [334, 431]. properties of a wide variety of adlay-
The rotational epitaxy angle estimated ers can be controlled as required by
by Toney and coworkers from in situ the applied electrode (substrate) poten-
SXS investigations (4.5◦ ) is only slightly tial and subsequently characterized by
smaller than the value predicted by the structure-sensitive in situ techniques in
model of Novaco and McTague [71, 72]. real space and real time. For these
The adlayer orientation with the lowest reasons, potentiostatically or galvanos-
energy is not the high symmetry direction. tatically generated monolayers on well-
Instead, the adlayer prefers to rotate to defined metal electrodes have become
nonsymmetry directions, taking advantage attractive model systems and provide an
of the lowest energy shear wave. important testing ground for fundamen-
A comparable discussion of Pb-UPD on tal issues in 2D physics and chemistry,
the other low-index phases of silver is such as phase transitions in adlayers
summarized in [2, 22]. and substrates surfaces [3, 7, 15, 16, 20,
143, 446]. Since the pioneering work
3.3.4.3Phase Transitions in Organic of Frumkin [447, 448], many ‘‘equilib-
Monolayers rium’’ and ‘‘dynamic’’ adsorption stud-
ies at metal–electrolyte interfaces have
3.3.4.3.1 General Aspects Organic mo- been performed with mercury [3, 15, 16,
nolayers on well-defined metal substrates 20, 449], gallium [18], and several amal-
may be obtained in various ways. Typ- gams [19], or low-melting-point electrodes
ical strategies are molecular beam epi- of sp-metals, such as Bi, Pb, Sn, or Zn,
taxy (MBE) [436], so-called ‘‘self-assembled onto which organic molecules are usually
monolayers’’ (SAM) [437] or LB films weakly adsorbed [449, 450]. The potentio-
(LB) [438, 439]. SAM’s are molecular as- static formation of 2D condensed mono-
semblies formed by the spontaneous in- layers was first reported by Lorenz [451]
teraction of a surfactant with a solid or and Vetterl [45] who found an unusual ca-
liquid substrate. Examples are thiols on pacitance versus potential hysteresis for
gold or silver [437, 440, 441] as well as alkyl- saturated aqueous solutions of nanoic
trichlorosilanes on oxide surfaces [442]. acid and the so-called ‘‘capacitance pits’’
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 435
for various purine and pyrimidine bases. pyridine [460] and several derivatives of
Since then, many other systems have uracil [469]. The phenomenological kinet-
been reported and quantitatively analyzed. ics of film formation and dissolution, as
Examples include carboxylic acids, cam- monitored by i–t, q M –t, or C –t tran-
phor and related compounds, purine and sients after single or multiple potential
pyrimidine and their derivatives, pyridine step protocols, have been described by
and bipyridines, quinolines, coumarin, models that involve (hole) nucleation and
thiourea, tetraalkyl-ammonium salts, and growth mechanism [183, 219, 221, 466].
so on. The corresponding literature is re- The results of these ‘‘macroscopic’’ elec-
viewed in [3, 15, 16, 20, 143]. trochemical experiments are still scarcely
Despite the rather detailed phenomeno- complemented by structural studies, such
logical knowledge on the 2D phase as IR- or Raman Spectroscopies or the
formation in adlayers, as obtained at use of UHV techniques after emersion
the mercury–electrolyte interface, fur- of the electrode [454, 463, 472–476]. Re-
ther progress has been hampered by markable progress in developing a ‘‘true’’
the lack of information on the struc- atomistic/molecular picture of potential-
ture and molecular mechanisms of film induced 2D organic phase formation
formation. New experimental and theoret- on adsorbate-modified single-crystal elec-
ical perspectives became available when trodes was achieved by the application of
single-crystal electrodes were employed in situ STM and AFM. Direct evidence
as substrate materials. This step offers for the formation of 2D long-range or-
the advantage to combine classical elec- dered structures on Au(111) has been
trochemical experiments with the power reported for several purine and pyrimi-
of structure sensitive in situ techniques, dine bases [201, 219, 454, 463, 477–480],
such as scanning probe microscopies phenol [481], pyridine [482, 483], 2,2 -
(STM, AFM) [23, 452, 453], vibrational and 4,4 -bipyridine [484–490], phenan-
spectroscopies (IRAS [454], SEIRAS [28], throline [491, 492], octylthiole [493], tetra-
SFG [30],. . .), electroreflectance [455], and methyl-thiourea [494, 495], cysteine [496],
surface plasmon microscopy [456], neu- mercaptopyridine [497], and benzenethi-
tron and surface X-ray scattering [457, ole [498]. Tao and coworkers [484, 491] and
458]. The formation of condensed organic Wandlowski and coworkers [201, 478, 479,
films on solid electrodes was suggested 486, 487] have demonstrated that the dy-
by Batrakov in 1974 for the adsorption namics of these phase formation processes
of camphor on Zn(0001) [459]. Few other can be studied successfully as a function
examples have been reported in the lit- of potential and temperature, even at a
erature during the last decade, mostly nanoscale level. Itaya and coworkers have
based on voltammetric and capacitance employed iodine-modified Au(111)- and
measurements: pyridine on Ag(210) [460], Ag(111)-electrodes to image self-organized
thymine on Cd(0001) [461], Au(100) [462], arrays of crystal violet, methyl-
Au(111) [463], and Ag(hkl) [464], coumarin pyridinium-phenylendivinylene and por-
on Au(111) and Au(100)-(hex) [465], uri- phyrine derivatives [499–501]. The same
dine on Au(hkl) [220, 466], uracil and group also reported 2D ordered struc-
camphor on Au(hkl) and Ag(hkl) [183, tures of benzene, naphthalene, and an-
193, 219, 221, 467–471]. Quantitative ther- thracene on Pt(111), Rh(111), and Cu(111)
modynamic studies were carried out for electrodes [502, 503]. These results are
436 3 Specific Adsorption
E E
vacant or occupied. Below a critical tem- phase for the square lattice is given by
perature Tc , the 2D lattice gas can split
5
into two phases, a gaseous phase and θ = 1 − x z − 2 2x − 16 z−2
8 −1 14
a condensed phase. The corresponding 2x
first-order equilibrium phase transition in- 31
volves a latent heat [528]. Two advanced − 3 6x 20 − 16x 22 + 24 z−3 . . .
3x
approximations of lattice gas isotherms
(52)
have been applied to describe 2D con-
with z = Bc and x = exp(−aLTE /4). Equa-
densation in organic adlayers: The QCA
tion (52) represents a truncated series
considers nearest-neighbor site pairs, but
expansion, which is obtained with the
treats them (still unrealistically) as inde-
assumption that contributions of clusters
pendent of each other [60, 73]:
larger than trimers can be neglected.
Figure 30 compares the adsorption
ln(Bc) = ln z
isotherms (ai + ln(Bc)) versus θ, i = F,
(β − 1 + 2θ)(1 − θ) 2 θ QCA, LTSE, and the temperature depen-
=
(β + 1 − 2θ)θ 1−θ dence of θ in the vicinity of the phase
transition for the above three approxima-
(50) tions and the exact LGM. According to
β = 1 − 4θ(1 − θ) the LGM, a first-order phase transition
takes place at (aG + ln(Bc)) = 0 and at
a 1/2
Qc temperatures lower than Tc [60, 528, 532].
× 1 − exp (51)
2 The degree of coverage of the condensed
phase at the equilibrium phase transition
The LTSE [529] considers equilibrium depends on aG , the lattice gas pair interac-
adsorbate clusters (up to 15 monomers) as tion coefficient, as follows:
well as various configurations of a cluster
a −4 1/8
with constant size indicating that the G
lattice geometry is taken into account. The θ = 0.5 + 0.5 1 − sin h
4
corresponding isotherm of the condensed (53)
θ θ
1.0
1.0
1
1 2
2 3
0.75 4
1 0.5 3
2
3
0.5 Tc /Tcrit
−2 −1 0 1 ai + ln(Bc) 0 0.5 1.0
(a) (b)
Fig. 30 (a) Coverage of the condensed phase as a function of the activity term (ai + ln(Bc)) and
(b) temperature dependence (Tc /Tcrit ) of the coverage of the condensed phase, θ, for the following
models: (1) BWA or MFA (Frumkin isotherm), (2) QCA, (3) LTSE, and (4) LGM. (Reprinted from
Ref. [528], Copyright 1987 by Elsevier Science Ltd.)
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 439
25
20
[µF cm−2]
C
15
3 2 1
10
−0.7 −0.8
E SCE
(a) [V]
1.0
0.8
θ
Tcrit(MFA)
0.6
Tcrit(LGM)
Figure 32 shows a typical capacitance ver- with the coumarin molecule slightly
sus potential curve of 5 mM coumarin inclined with respect to the interface
in 0.5 M NaF at a HMDE. Three dif- and mI = 3.1 10−10 mol cm−2 [163]. The
ferent potential regions exist. Region capacitance decreases at more negative
I is interpreted as planar orientation potentials, and corresponds in region
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 441
0.4
O
0.3 O
[Fm−2]
C
0.2 I
II
0.1
III
0.3
0.2
3 Specific Adsorption
C
0.2
17 5 44 23 29 20
[Fm−2]
[Fm−2]
3 2 1
0.1
0.1
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0 0.2 0.4 0.6 0.8 1.0
t t
(a) [s] (b) [s]
1.0 100
0.2 Small
0.8 Medium 80
Large
0.6 60
J
C
vg
1 2 3 4 5 6
0.1
[Fm−2]
0.4 40
[cms−1]
[cms−1]
0.2 20
0
0 0.5 1.0 1.5 2.0 2.5 −0.40 −0.45 −0.50 −0.55 −0.60
t ESCE
(c) [s] (d) [V]
Fig. 33 Capacitance transients obtained from single potential step experiments for 5 mM coumarin in 0.5 M NaF at a stationary mercury electrode,
5 ◦ C.(a) deterministic transients with E1 = −0.400 V to E2 = −0.560 V (1), −0.565 V (2), and −0.570 V (3) The solid lines were calculated by nonlinear
last squares fitting to Eq. (35) with m = 3. (b) Transients as in (a), except that E2 = −0.540 V. The transients are now stochastic. The numbers represent
their sequence number in a set of 50 repeat experiments. (c) Transients as in (a), except that E2 = −0.520 V. All transients reflect mononucleation and
can be represented by Eq. (30). The coverage was obtained with Eq. (35). (Reprinted from Ref. [164], copyright 1994 by Bulgarian Chemical Society.)
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 443
sphere, the growth of the compact film, capacitance pits, thermodynamic and ki-
once it has been nucleated, is given netic aspects of the phase formation
by [544] of uracil and its derivatives have been
studied comprehensively at the mer-
2θ = 1 − cos(vg (t − to )/r) (58) cury–electrolyte interface (cf. reviews in [3,
15]). In 1995, Hölzle and coworkers [219]
where vg is the linear growth velocity, presented convincing evidence that 2D
to the moment of nucleation, and r the condensation of uracil is also observed
radius of the mercury droplet, θ is obtained on low-index gold single-crystal electrodes.
from Eq. (36). The potential dependence of On the basis of voltammetric, capacitance
the growth rate is plotted in Fig. 33(d) for and transient experiments [183, 219–221,
three different drop sizes. The nucleation 461, 464, 462, 467, 547] and their combina-
rate was obtained for each potential by tion with structure-sensitive methods such
analyzing a series of 550 mononucleation as in situ STM [454, 478–480], vibrational
transients. Extrapolating the growth curves spectroscopy [454, 473, 475,], SXS [458],
to to , and sorting in bins, provides the and UHV-based characterization strate-
probability Po that nucleation has not gies [463], a rather comprehensive un-
yet occurred in a given time interval t derstanding of the interfacial properties
(cf. Sect. 3.3.3.2.3). Representing − ln Po of uracil and several of its derivatives
as a function of time then produces on Au(hkl), Ag(hkl), and Cd(0001) has
a plot from which both, the induction emerged.
time and the steady state nucleation rate Figure 34 shows a typical cyclic voltam-
is obtained [163, 164, 545]. The latter is mogram for 12 mM Uracil in 0.05 M
plotted in Fig. 33(d). The experiments H2 SO4 . The first voltammetric scan, as
were carefully designed to ensure that obtained with a flame-annealed, thermally
only a small fraction of the observed √
reconstructed Au(111)-(p × 3) electrode
nucleation appears to originate at the after immersion at −0.100 V, is indicated
lumen of the capillary (e.g. at defect as dashed line. The solid line repre-
sites!) [164]. sents the steady state voltammogram,
The dissolution transients of the con- recorded after five complete potential cy-
densed coumarin film III depend strongly cles [478]. The phenomenological analysis
on the applied potential step regime, of these electrochemical data yields the
and show strong aging effects, which existence of four distinct interfacial re-
were attributed to the healing of domain gions labeled I to IV in Fig. 34 [20]:
boundaries. The corresponding ‘‘grain- In region I the molecules are ran-
boundary’’ dissolution model as developed domly (gaseous-like) adsorbed on the
recently by Poelman and coworkers [546] electrode until complete desorption oc-
is applicable (cf. Fig. 6) curs at sufficiently negative potentials.
The formation of condensed coumarin The formation of a 2D condensed phy-
monolayers was also reported on Au(111)- sisorbed uracil film at more positive
(1 × 1) and Au(100)-(hex) [465]. potentials is indicated by typical current
spikes P1 /P1 and P2 /P2 , which limit
3.3.4.3.5 Case Study II – Uracil on Au(hkl) the pit region II. A marked hystere-
Since the pioneering work of Vetterl [45], sis is observed at both pit edges when
who reported for the first time so-called changing the direction of the potential
444 3 Specific Adsorption
(b) 30 nm
P2
P1'
I II IV
2 µA cm−2
III
P3
P2'
scan. The stability range of II increases Hg < Ag(111) < Ag(100) < Au(111)
with decreasing temperature. The respec- √
−(p × 3) < Au(100)−(1 × 1)
tive critical temperatures assumes the
following substrate sequence [469, 470] ∼ Au(100)−(hex) (59)
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 445
(a) 10 nm (b)
Fig. 35 Unfiltered in situ STM images of the physisorbed uracil film on
√
Au(111)-(p × 3)/0.05 M H2 SO4 + 3 mM uracil: (a) high-resolution image
at E = −0.05 V, iT = 2 nA, vT = +0.01 V. The primitive unit cell is indicated;
(b) proposed packing model. (Reprinted from [478], copyright 1997,
American Chemical Society.)
446 3 Specific Adsorption
b
α
a
a
(a) 7 nm (b) 7 nm
a b
b α
a
[112]
[110]
1 nm 1 nm
(c) (d)
Fig. 36 High-resolution STM images and packing models of chemisorbed uracil
on Au(111)-(1 × 1) (a); E = 0.60 V, iT = 2 nA, vT = −0.65 V and
Au(100)-(1 × 1) (b); E = 0.60 V, iT = 1.8 nA, vT = −0.52 V in 0.05 M H2 SO4 .
Characteristic dimensions of the adsorbate mesh and the respective elementary
cells are indicated in the experimental images and in the proposed packing
models depicted in (c) and (d), respectively.
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 447
nC4 = O
experiment for the dissolution of the
chemisorbed uracil monolayer on
−0.005/a.u.
]
40 tC=O stretching region (For clarity, the
[ms ‘‘loss’’ spectra are plotted with an
1750 1650 1550 1450
inverted sign !) (b) Time dependence of
the integrated intensity of the uracil
n
(a) [cm ]−1 vibration υC4=O (1590 cm−1 ). The solid
lines represent fits to an Avrami-type
0.12 equation (cf. Eq. (35)) with m = 1.12.
(c) Corresponding current transient.
Intensity (n C2 = O)
m = 1.48
−8 Ref. [475]).
EC (c)
−12
0 20 40 60 80 100
t
[ms]
Raman scattering [555], chronocoulome- electrode. Molecular stacks are formed via
try, and SHG [556], SXS [458], FTIR [557], the coordination of the two ring nitrogen
SEIRAS [489], and in situ STM [484, atoms with the positively charged gold
486–489]. Lipkowski and coworkers [556] electrode in region I (Fig. 38a, b). The
found, on the basis of electrochemical data, distance between adjacent rows amounts
that 2,2 -bipyridine exhibits multistate to (0.96 ± 0.5) nm, the intermolecular
adsorption on Au(111). Structural details distance is (0.38 ± 0.2) nm, and the
of this complex phase behavior were molecules are tilted from the normal
resolved by in situ STM (Fig. 38) and to the axis of the chain by an
SEIRAS. angle α = (28 ± 2)o [484, 486, 488]. The
2,2 -BP was dissolved in aqueous simultaneous imaging of the hexagonal
electrolyte and potentiostatically deposited substrate surface and of the organic adlayer
onto the unreconstructed Au(111)-(1 × 1) pattern pointed to a commensurate (4 ×
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 449
N N
+ + + + + +
Stacking phase I
(e) 7.0 nm
Stacking phase II
II I
2 µA cm−2
ESCE
-0.2 0 0.2 0.4 0.6 [V]
√
2 3) unit cell, which contains three positions [486]. The structure-determining
molecules, for example, AI ≈ 0.38 nm2 . role of the substrate crystallography onto
The nitrogen atoms coordinate with the 2,2 -BP stacking pattern is supported
the substrate surface in twofold bridge by experiments with the quadratic
450 3 Specific Adsorption
35 nm × 35 nm
(a) (b)
(c) (d)
Fig. 39 Sequence of STM images for 3 mM 2,2 -BP on Au(111)/0.05 M H2 SO4
after a single potential scan (10 mV s−1 ) from 0.50 V to 0.24 V (potential-induced
dissolution of I): (a) E = 0.50 V, (b) E = 0.24 V, t = 3 min, (c) E = 0.17 V,
t = 4.5 min, (d) E = 0.17 V, t = 8.0 min; iT = 2 nA. (Reprinted from Ref. [486],
copyright 1998 by Elsevier Science Ltd.)
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 451
II, have been observed in sulfate- and stacking rows, and rather slow perpen-
in perchlorate-containing neutral or acidic dicular to it. Positional and directional
electrolytes [484, 486–489]. fluctuations within the ordered and the dis-
Details of the dynamics of the ordered phases appear. Under the shown
rather slow potential and/or temperature- experimental conditions the ordered 2,2 -
induced structural transitions could be BP stacking layer dissolved completely at
monitored by in situ STM in real space and 33.60 ◦ C (Fig. 40d). Closer inspection of
time [486]: The first example starts with a the disordered regions reveals a granular
highly ordered 2,2 -BP stacking layer I, structure on a length scale exceeding the
equilibrated at Ei = 0.500 V on Au(111)- pixel resolution of the image in the x –y
(1 × 1) (Fig. 39a), and then the potential direction. This means that a noticeable
was scanned slowly towards negative part of the modulation in the disordered
values. The range was restricted ensuring area is not a time effect caused by the fast
that the gold electrode still bears a positive motion of adsorbed species, but rather re-
charge. At E = 0.240 V the first point flects, at least in part, the static positions of
dislocations occur (Fig. 39b). The active particles in the disordered phase. This sug-
centers are formed not only on defects of gests the existence of adsorbate–substrate
the adsorbate or substrate lattices, such coordination complexes. Similar observa-
as point defects or domain boundaries, tions were reported for the 2D fluid–solid
steps or kink positions. They are equilibrium of cesium and oxygen coad-
distributed evenly across entire terraces. sorbed on Ru(0001) under UHV condi-
The nucleated holes grow anisotropically tions [558].
with time and more negative electrode In the present case, the adsorbed species
potentials until the ordered adlayer is ‘‘feel’’ the effect of the substrate corru-
dissolved completely. Growth proceeds gation potential. This situation is differ-
preferentially along the directions of the ent from purely 2D system, like gua-
commensurate molecular stacking rows. nine or adenine on HOPG [210] or the
During this transition one can identify two physisorbed films of uracil or thymine on
phases clearly, solidlike patches exhibiting Au(hkl) [219, 478, 479], where the adsor-
characteristic long-range order of 2,2 - bate–substrate interactions are very weak.
BP in region I, and a fluid medium There, the respective phase transitions are
disordered on the atomic scale (Fig. 39b). governed entirely by the applied electri-
The observed order/disorder transition is cal field and directional, lateral attrac-
first order [486]. tive interactions via hydrogen bonds [20,
The above interpretation is supported by 453].
experiments on the temperature-induced The potential and/or temperature con-
dissolution of the ordered 2,2 -BP phase trolled dissolution of the above order/
I [486]. The system was stabilized at Ei = disorder transition is completely reversible
0.200 V and T = 29.7 ◦ C (Fig. 40a), and in sulfuric acid solution. A representa-
then the temperature was slowly ramped tive time sequence, after application of
with 0.02 ◦ C min−1 . Firstly, point and line a potential scan from Ei = 0.100 V to
dislocations (‘‘one-dimensional holes’’) ap- Ef = 0.500 V, is plotted in Fig. 41. Within
pear (Fig. 40b). They grow anisotropi- a few seconds at Ef the 2,2 -BP molecules
cally by successive stripping of molecular start to order. No preferential nucleation
stacks, fast along the directions of the sites, which might correlate with defect
452 3 Specific Adsorption
40 nm × 40 nm
(a) (b)
(c) (d)
Fig. 40 Temperature-induced dissolution of the 2,2 -BP stacking phase I for
3 mM 2,2 -BP in 0.05 M H2 SO4 /Au(111). (a) The system was equilibrated at
E = 0.20 V and 29.7 ◦ C for 20 min. Then the temperature was ramped with
0.02 ◦ C min−1 , (b) 33.14 ◦ C, (c) 33.51 ◦ C, (d) 33.60 ◦ C. All images were
obtained in constant current mode (iT = 0.25 nA) at the same surface area.
(Reprinted from Ref. [486], copyright 1998 by Elsevier Science Ltd.)
structures of the substrate surface, were positions at the end of the growing front.
found. First stacks are formed; their direc- (2) The second growth mechanism pro-
tions correlate but exhibit a ‘‘missing row’’ ceeds via incorporation of entire molecular
arrangement (Fig. 41b). With increasing rows perpendicular to the main stack di-
observation time this metastable phase is rection. The growth and the existence of
displaced by a densely packed stacking do- other metastable structures are strongly
main (Fig. 41c) with dimensions typical influenced by the nature of the support-
for phase I. The latter grows anisotropi- ing electrolyte anions [20, 484, 486, 488,
cally in two directions: (1) parallel along 489]. Defects within the adlayer change
the main stacking direction. Individual their shape or heal with increasing ob-
molecules and small stacks are still in servation time. The detailed mechanism
search of energetically favorable adlattice most probably involves place exchange
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 453
(a) 35 nm x 35 nm (b)
(c) (d)
(e) (f)
Fig. 41 Potential-induced formation of the 2,2 -BP stacking structure I on
Au(111) after a potential scan from region II (10 mV s−1 ): (a) Ei = 0.10 V to
Ef = 0.50 V, (b) E = 0.50 V, t = 1 min, (c) E = 0.50 V, t = 5 min,
(d) E = 0.50 V, t = 8 min, (e) E = 0.50 V, t = 11 min, (f) E = 0.50 V,
t = 42 min; iT = 2 nA. Solution composition 3 mM 2,2 -BP/0.05 M H2 SO4 .
(Reprinted from Ref. [486], copyright 1998 by Elsevier Science Ltd.)
454 3 Specific Adsorption
processes and surface diffusion of 2,2 -BP- A main field of activities is focused on
gold complexes. A similar mechanism is structure and reactivity in two-dimensional
proposed for the ‘‘aging’’ of SAMs formed adlayers at electrode surfaces. Significant
by alkanthioles on gold surfaces [559]. new insights were obtained into the spe-
At the end of this chapter it is remarked cific adsorption and phase formation of
that the described structural details of the anions and organic monolayers as well as
above-mentioned phase transitions are not into the underpotential deposition of metal
accessible by macroscopic electrochemical ions on foreign substrates. The in situ
approaches, such as measurements of the application of structure-sensitive methods
interfacial capacitance or charging current, with an atomic-scale spatial resolution, and
but can be attributed to the often observed a time resolution up to a few microseconds
short or long-time transient responses. revealed rich, potential-dependent phase
behavior. Randomly disordered phases,
3.3.5 lattice gas adsorption, commensurate and
Conclusion and Outlook incommensurate (compressible and/or ro-
tated) structures were observed. Attempts
The macroscopic approach towards the have been developed, often on the basis of
‘‘static’’ electrochemical interface as de- concepts of 2D surface physics, to ratio-
veloped through the work of Lippmann, nalize the observed phase changes and
Gouy, Grahame, Frumkin/Damaskin, and transitions by competing lateral adsor-
Parsons is starting to change deeply. Over bate–adsorbate and adsorbate–substrate
the last decade a detailed microscopic pic- interactions.
ture of the bare and adsorbate-covered The thermodynamic and statistical me-
electrochemical interface has emerged. chanical analysis of first-order and/or con-
These advances could be attributed to four tinuous phase transitions demonstrated
major developments: (1) The use of well- clearly that models based on the MFA
defined single-crystal electrodes instead are often inadequate. Lattice gas mod-
of polycrystalline material. (2) The com- els (LGM), and approximations based
bination of complementary classical elec- on the quasi-chemical treatment (QCA)
trochemical techniques, based on macro- or series expansion techniques (e.g. the
scopic measurements of current, charge low-temperature series expansion (SE))
density, or interfacial capacitance, with are more appropriate. MC simulations
structure-sensitive in situ techniques, have also been developed as a powerful
such as vibrational spectroscopies (FTIR, approach to unravel the role of lateral
SEIRAS, surface-enhanced Raman scatter- interactions in various phase formation
ing (SERS), SFG), X-ray methods (SXS, processes.
EXAFS, . . .) and scanning probe micro- Impressive atomic- and molecular-scale
scopies (STM, AFM and related tech- details of steady state properties of 2D
niques). (3) The use of UHV techniques phase changes in ionic and molecular
for structure-sensitive investigations af- adlayers, and substrate surfaces have been
ter controlled emersion of the electrode obtained, mostly based on in situ STM,
and the transfer into a vacuum chamber. IR- and SXS-studies. Clear evidence was
(4) The development of modern theoreti- presented of the effect of crystallographic
cal concepts of the electrochemical double orientation, the coadsorption of ions of the
layer, and of processes taking place there. supporting electrolyte and of the solvent
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 455
90. P. Delahay, J. Phys. Chem. 1963, 67, 115. E. Budevski, V. Bostanov, T. Vitanov et al.,
135–137. Electrochim. Acta 1966, 11, 1697–1707.
91. D. Schuhmann, in Proprietes Electriques des 116. D. M. Kolb, in Advances in Electrochem-
Interfacial Charges (Ed.: D. Schuhmann), istry and Electrochemical Engineering (Eds.:
Masson, Paris, 1978, pp. 166–192. H. Gerischer, Ch. Tobias), John Wiley &
92. M. Noel, K. I. Vasur, Cyclic Voltammetry Sons, New York, pp. 125–272, Vol. 11.
and Frontiers in Electrochemistry, Aspect 117. R. Philipp, U. Retter, Thin Solid Films 1992,
Publications, London, 1990. 207, 42–50.
93. P. Delahay, C. T. Fike, J. Am. Chem. Soc. 118. Southhampton Electrochemistry Group, In-
1958, 80, 2628–2630. strumental Methods in Electrochemistry, Ellis
94. T. Berzins, P. Delahay, J. Am. Chem. Soc. Horwood, New York, 1990.
1955, 49, 906–909. 119. Cl. Buess-Herman, L. Gierst, Colloids Surf.
95. P. Delahay, I. Trachtenberg, J. Am. Chem. 1984, 12, 137–150.
Soc. 1957, 79, 2355–2362. 120. U. Retter, J. Electroanal. Chem. 1990, 296,
96. L. Rampazzo, Electrochim. Acta 1965, 14, 445–451.
733–741. 121. D. Kashchiev, Surf. Sci. 1969, 14, 209–220.
97. K. Jüttner, G. Staikov, W. J. Lorenz et al., 122. D. Kashchiev, Nucleation – Basic Theory
J. Electroanal. Chem. 1977, 80, 67–80. and Applications, Butterworth-Heinemann,
98. J. Michailik, B. B. Damaskin, Sov. Elec- Oxford, 2000.
trochem. 1979, 15, 478–481. 123. R de Levie, in Advances in Electrochem-
99. N. Emanuel, D. Knorre, Kurs Chimitschesoi istry and Electrochemical Engineering (Eds.:
Kinetiki; Wyschaja schkola, Moskwa, 1962. H. Gerischer, Ch. Tobias), John Wiley &
100. W. Lorenz, Z. Elektrochem. 1958, 62, Sons, New York, 1985, pp. 1–67, Vol. 13.
192–200. 124. S. Toshev, I. Markov, J. Cryst. Growth 1968,
101. R. D. Armstrong, J. Electroanal. Chem. 1969, 34, 436–445.
20, 168–170. 125. S. Toshev, I. Markov, Ber. Bunsen-Ges. Phys.
102. J. W. Gibbs, Collected Works, Longman’s Chem. 1969, 73, 184–192.
Green, London (1878), 1978. 126. D. Kashchiev, Surf. Sci. 1969, 18, 389–397.
103. M. Volmer, A. Weber, Z. Phys. Chem. 1926, 127. J. H. Sluyters, J. H. O. J. Wijenberg, W. H.
119, 277–301. Mulkder et al., J. Electroanal. Chem. 1989,
104. M. Volmer, Kinetik der Phasenbildung, 261, 263–272.
Steinkopf, Dresden, 1939. 128. S. Toschev, J. Gutzow, Phys. Status Solidi
105. L. Farakas, Z. Phys. Chem. 1927, 125, 239. 1967, 21, 683–689.
106. I. N. Stranski, R. Kaishev, Z. Phys. Chem. 129. I. Stranski, Z. Phys. Chem. 1928, 136,
1934, B26, 100–113. 209–220.
107. I. N. Stranski, R. Kaishev, Z. Phys. Chem. 130. G. Staikov, W. J. Lorenz, Can. J. Chem.
1934, B26, 114–131. 1997, 75, 1624–1634.
108. I. N. Stranski, R. Kaishev, Z. Phys. Chem. 131. A. Milchev, Electrochim. Acta 1986, 31,
1934, B26, 312–316. 977–980.
109. I. N. Stranski, R. Kaishev, Z. Phys. Chem. 132. A. Milchev, J. Electroanal. Chem. 1998, 457,
1934, A170, 295–299. 35–46.
110. R. Becker, W. Döring, Ann. Phys. 1935, 24, 133. P. M. Jacobs, F. C. Tompkins, in Chemistry
719–753. of the Solid State (Ed.: W. E. Garner), Butter-
111. J. B. Zeldovich, Acta Physicochim. (URSS) worths, London, 1955, pp. 181–212.
1943, 18, 1–12. 134. U. Retter, J. Electroanal. Chem. 1978, 87,
112. I. Ya. Frenkel, J. Chem. Phys. 1939, 7, 181–188.
200–210; 538–546. 135. W. Obretenov, I. Petrov, I. Nachev et al., J.
113. E. Budevski, G. Staikov, W. J. Lorenz, Elec- Electroanal. Chem. 1980, 109, 195–198.
trochim. Acta 2000, 45, 2559–2574. 136. C. Donner, H. Baumgärtel, L. Pohlmann
114. E. Budevski, in Comprehensive Treatise et al., Ber. Bunsen-Ges. Phys. Chem. 1996,
of Electrochemistry (Eds.: B. E. Conway, 100, 403–412.
J. O. M. Bockris, E. Yeager et al.), Plenum, 137. C. Donner, L. Pohlmann, H. Baumgärtel,
New York, 1983, pp. 399–450, Vol. 7. Surf. Sci. 1996, 345, 363–372.
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 459
138. R. Philipp, U. Retter, Z. Phys. Chem. 1996, 163. R. Srinivasan, R. de Levie, J. Phys. Chem.
196, 125–134. 1987, 91, 2904–2908.
139. I. Markov, Crystal Growth for Beginners, 164. Th. Wandlowski, R. de Levie, Bulg. Chem.
World Scientific, Singapore, 1995, Commun. 1995, 27, 231–246.
pp. 63–146. 165. E. Bosco, S. K. Rangarajan, J. Chem. Soc.,
140. D. Walton, J. Chem. Phys. 1962, 37, Faraday Trans. 1981, 177, 483–495.
2182–2190. 166. F. Canac, C. R. Acad. Sci. 1933, 196, 51–62.
141. S. Stoyanov, Thin Solid Films 1973, 18, 167. A. N. Kolmogoroff, Bull. Acad. Sci. USSR,
91–98. Ser. Math. Nat. 1937, 3, 355–361.
142. A. Milchev, S. Stoyanov, R. Kaischev, Thin 168. M. Avrami, J. Chem. Phys. 1939, 7,
Sold Films 1974, 22, 255–265, 267–274. 1103–1111.
143. C. Buess Herman, in Adsorption of 169. M. Avrami, J. Chem. Phys. 1940, 8, 212–224.
Organic Molecules at Metal Electrodes (Eds.: 170. M. Avrami, J. Chem. Phys. 1941, 9, 177–184.
J. Lipkowski, P. N. Ross), VCH, New York, 171. U. R. Evans, Trans. Faraday Soc. 1945, 41,
1992, pp. 77–118. 365–374.
144. O. M. Magnussen, L. Zitzler, B. Gleich 172. M. Y. Abyaneh, M. Fleischmann, Elec-
et al., Electrochim. Acta 2001, 46, trochim. Acta 1982, 27, 1513–1518.
3725–3733. 173. U. Retter, J. Electroanal. Chem. 1984, 179,
145. F. C. Frank, Proc. R. Soc. London, Ser. A 25–29.
1950, 201, 586–599. 174. U. Retter, J. Electroanal. Chem. 1980, 106,
146. S. Rangarajan, Faraday Symp. Chem. Soc. 371–375.
1978, 12, 101–114. 175. G. Quarin, Cl. Buess-Herman, L. Gierst,
147. S. Fletcher, A. Smith, Electrochim. Acta J. Electroanal. Chem. 1981, 123, 35–58.
176. Th. Wandlowski, G. R. Jameson,
1980, 25, 1019–1024.
R. de Levie, J. Phys. Chem. 1993, 97,
148. S. H. Liu, Solid State Phys. 1986, 39,
10 119–10 126.
207–211.
177. A. Bewick, M. Fleischmann, H. R. Thirsk,
149. N. J. Tao, C. Z. Li, F. Cunha et al., Scann.
Trans. Faraday Soc. 1962, 58, 2200–2209.
Microsc. 2002, in press.
178. R. D. Armstrong, M. Fleischmann, H. R.
150. T. A. Witten, L. M. Sander, Phys. Rev. Lett.
Thirsk, J. Electroanal. Chem. 1966, 11,
1981, 47, 1400–1403.
208–215.
151. L. Pospisil, J. Electroanal. Chem. 1986, 206, 179. J. A. Harrison, H. R. Thirsk, in Electroan-
269–283. alytical Chemistry (Ed.: A. J. Bard), Mar-
152. L. Pospisil, J. Phys. Chem. 1988, 92, cel Dekker, New York, 1971, pp. 67–148,
2501–2506. Vol. 5.
153. R. Philipp, J. Electroanal. Chem. 1990, 290, 180. C. Müller, J. Claret, M. Sartet, J. Electroanal.
67–78. Chem. 1987, 227, 147–158.
154. Th. Wandlowski, J. Electroanal. Chem. 1992, 181. M. H. Hölzle, U. Retter, D. M. Kolb, J. Elec-
333, 77–91. troanal. Chem. 1994, 371, 101–109.
155. U. Retter, Langmuir 2000, 16, 7752–7756. 182. R. Philipp, J. Dittrich, U. Retter et al., J.
156. E. Budevski, W. Bostanov, T. Vitanov et al., Electroanal. Chem. 1988, 250, 159–164.
Electrochim. Acta 1966, 11, 1697–1707. 183. Th. Wandlowski, Th. Dretschkow, J. Elec-
157. E. Budevski, W. Bostanov, T. Vitanov et al., troanal. Chem. 1997, 427, 105–112.
Phys. Status Solidi 1966, 13, 577–588. 184. R. G. Barradas, T. J. Vandernoot, J. Elec-
158. V. Bostanov, G. Staikov, D. K. Roe, J. Elec- troanal. Chem. 1982, 142, 107–119.
trochem. Soc. 1975, 122, 1301–1305. 185. R. G. Barradas, T. J. Vandernoot, J. Elec-
159. W. Obretenov, V. Bostanov, V. Popov, troanal. Chem. 1984, 176, 151–167.
J. Electroanal. Chem. 1982, 132, 273–276. 186. R. G. Barradas, M. Rennie, T. Vandernoot,
160. L. Gierst, C. Frank, G. Quarin et al., J. Electroanal. Chem. 1983, 144, 455–458.
J. Electroanal. Chem. 1981, 129, 353–363. 187. W. H. Mulder, J. H. O. J. Wijenberg,
161. R. Sridharan, R. de Levie, J. Electroanal. M. Sluyters-Rehbach et al., J. Electroanal.
Chem. 1987, 218, 287–296. Chem. 1989, 270, 7–19.
162. F. Thomas, Cl. Buess-Herman, L. Gierst, 188. Th. Wandlowski, J. Electroanal. Chem. 1992,
J. Electroanal. Chem. 1986, 214, 597–613. 333, 77–91.
460 3 Specific Adsorption
239. G. Gouy, Ann. Chim. Phys. 1906, 8, 263. J. X. Wang, R. R. Adzic, T. Wandlowski
291–364. et al., unpublished.
240. G. Gouy, Ann. Chim. Phys. 1906, 9, 75–139. 264. J. X. Wang, G. M. Watson, B. M. Ocko,
241. W. Schmickler; Chem. Rev. 1996, 96, unpublished.
3177–3200. 265. B. M. Ocko, O. M. Magnussen, J. X. Wang
242. D. M. Kolb, Z. Phys. Chem. N.F. 1987, 154, et al., Physica B 1996, 221, 238–244.
179–199. 266. G. Aloisi, A. M. Funtikov, T. Will, J. Elec-
243. T. Iwasita, F. C. Nart, in Advances in Elec- troanal. Chem. 1994, 370, 297–300.
trochemical Science and Engineering (Eds.: 267. T. Yamada, K. Ogaki, S. Okubo et al., Surf.
H. Gerischer, C. W. Tobias), VCH, New Sci. 1996, 369, 321–335.
York, 1995, pp. 126–216. 268. B. M. Ocko, Th. Wandlowski, Mater. Res.
244. G. Pacchioni, P. S. Bagus, M. R. Philpott, Soc. Symp. Proc. 1997, 451, 55–64.
Z. Phys. D 1989, 12, 543–546. 269. T. Teshima, K. Ogaki, K. Itaya, J. Phys.
245. A. Ignaczak, J. A. N. F. Gomes, J. Elec- Chem. B 1997, 101, 2046–2053.
troanal. Chem. 1997, 420, 71–78. 270. P. Broekmann, M. Wilms, M. Kruft et al., J.
246. M. T. M. Koper, R. A. van Santen, Surf. Sci. Electroanal. Chem. 1999, 467, 307–324.
1999, 422, 118–131. 271. J. Inukai, Y. Osawa, K. Itaya, J. Phys. Chem.
247. E. Spohr, in Advances in Electrochemical B 1998, 102, 1034–1040.
Sciences and Engineering (Eds.: R. C. Alkire, 272. P. Broekmann, M. Anastasescu, A. Spaenig
D. M. Kolb), Wiley-VCH, Weinheim, 1999, et al., J. Electroanal. Chem. 2001, 500,
pp. 1–76, Vol. 6. 241–254.
248. A. N. Frumkin, Z. Phys. Chem. 1933, 164A, 273. S. L. Yau, C. M. Vitus, B. C. Schardt, J. Am.
121–133. Chem. Soc. 1990, 112, 3677–3679.
274. C. A. Lucas, N. M. Markovic, P. N. Ross,
249. D. M. Kolb, Prog. Surf. Sci. 1996, 51,
Phys. Rev. B 1997, 55, 7964–7971.
109–173.
275. R. Vogel, I. Kamphausen, H. Baltruschat,
250. A. R. Despic, in Kinetics and Mechanisms
Ber. Bunsen-Ges. Phys. Chem. 1992, 96,
of Electrode Processes (Eds.: B. E. Conway,
525–530.
J. O’. M. Bockris, E. Yeager et al.), Plenum
276. J. Inukai, Y. Osawa, M. Wakisaka et al., J.
Press, New York, 1993, pp. 451–528.
Phys. Chem. B 1998, 102, 3498–3505.
251. R. R. Adzic, J. X. Wang, Electrochim. Acta
277. J. M. Orts, R. Gómez, J. M. Feliu, J. Elec-
2000, 45, 4203–4210. troanal. Chem. 1999, 467, 11–19.
252. X. Gao, M. J. Weaver, J. Am. Chem. Soc. 278. K. Sashikata, Y. Matsui, K. Itaya et al., J.
1992, 114, 8544–8551. Phys. Chem. 1996, 100, 20 027–20 034.
253. O. M. Magnussen, J. X. Wang, R. R. Adzic 279. L. J. Wan, S. L. Yau, G. M. Swain et al., J.
et al., J. Phys. Chem. 1996, 100, 5500–5508. Electroanal. Chem. 1995, 381, 105–111.
254. O. M. Magnussen, B. M. Ocko, R. R. Adzic 280. P. Müller, S. Ando, T. Yamada et al., J.
et al., Phys. Rev. B 1995, 51, 5510–5513. Electroanal. Chem. 1999, 467, 282–290.
255. B. M. Ocko, G. M. Watson, J. Wang, J. Phys. 281. H. H. Farell, in The Chemical Physics of Solid
Chem. 1994, 98, 897–906. Surfaces and Heterogeneous Catalysis (Eds.:
256. X. Gao, G. J. Edens, F. C. Liu et al., J. Phys. D. A. King, D. P. Woodruff), Elsevier, New
Chem. 1994, 98, 8086–8095. York, 1987, pp. 225–272, Vol. 36.
257. X. Gao, G. J. Edens, M. J. Weaver, J. Phys. 282. P. A. Dowden, Crit. Rev. Solid State Mater.
Chem. 1994, 98, 8074–8085. Sci. 1987, 13, 191–210.
258. T. Yamada, N. Batina, K. Itaya, J. Phys. 283. G. N. Kastanas, B. E. Koel, Appl. Surf. Sci.
Chem. 1995, 99, 8817–8823. 1993, 64, 235–249.
259. B. M. Ocko, O. M. Magnussen, J. X. Wang 284. P. J. Goddard, R. M. Lambert, Surf. Sci.
et al., Phys. Rev. B 1996, 53, R7654–R7657. 1977, 67, 180–194.
260. J. X. Wang, G. M. Watson, B. M. Ocko, J. 285. K. Kern, G. Comsa, in Chemistry
Phys. Chem. 1996, 100, 6672–6677. and Physics of Solid Surfaces (Eds.:
261. A. Cuesta, D. M. Kolb, Surf. Sci. 2000, 465, R. Vanselow, R. Howe), Springer, Berlin,
310–316. 1988, pp. 65–108.
262. S. Zou, X. Gao, M. J. Weaver, Surf. Sci. 286. P. Bak, D. Mukamel, I. Villain et al., Phys.
2000, 452, 44–57. Rev. B 1979, 19, 1610–1613.
462 3 Specific Adsorption
287. M. Kardar, A. N. Berker, Phys. Rev. Lett. 312. H. Uchida, N. Ikeda, M. Watanabe, J. Elec-
1982, 48, 1552–1555. troanal. Chem. 1997, 424, 5–12.
288. M. F. Toney, J. G. Gordon, M. G. Samant 313. Y. Shingaya, M. Ito, Electrochim. Acta 1998,
et al., Phys. Rev. B 1992, 45, 9362–9374. 44, 745–751.
289. X. Wang, R. Chen, Y. Wang et al., J. Phys. 314. K. Ataka, M. Osawa, Langmuir 1998, 14,
Chem. B 1998, 102, 7568–7576. 951–959.
290. Z. Racz, Phys. Rev. B 1980, 21, 4012–4016. 315. I. R. de Moraes, F. C. Nart, J. Electroanal.
291. S. J. Mitchell, G. Brown, P. A. Rikvold, J. Chem. 1999, 461, 110–120.
Electroanal. Chem. 2000, 493, 68–74. 316. Z. Shi, J. Lipkowski, S. Mirwald et al.,
292. W. Haiss, PhD Thesis, Free University J. Electroanal. Chem. 1995, 396, 115–124.
Berlin, Berlin, 1994. 317. T. Nishizawa, T. Namata, Y. Kinoshita
293. F. Silva, A. Martins, in Interfacial Elec- et al., Surf. Sci. 1996, 367, L73–L78.
trochemistry (Ed.: A. Wieckowski), Marcel 318. E. M. Patrito, P. Paredes Olivera, H. Sellers,
Dekker, New York, 1999, pp. 427–461. Surf. Sci. 1997, 380, 264–282.
294. H. Angerstein-Kozlowska, B. E. Conway, 319. D. M. Kolb, M. Przasnyski, H. Gerischer,
A. Hamelin et al., J. Electroanal. Chem. 1987, J. Electroanal. Chem. 1974, 54, 25–38.
228, 429–453. 320. J. W. Schultze, D. Dickertmann, Faraday
295. J. Clavilier, J. Electroanal. Chem. 1980, 107, Symp. 1977, 12, 36.
211–216. 321. J. W. Schultze, D. Dickertmann, Faraday
296. J. M. Feliu, M. T. Valls, A. Aldaz et al., Symp. 1977, 12, 172.
J. Electroanal. Chem. 1993, 345, 475–481. 322. J. W. Schultze, D. Dickertmann, Surf. Sci.
297. F. Silva, M. J. Sottomayor, A. Martins, 1976, 54, 489–505.
J. Electroanal. Chem. 1994, 375, 395–399. 323. E. Leiva, Electrochim. Acta 1996, 41,
298. G. J. Edens, X. Gao, M. J. Weaver, J. Elec- 2185–2206.
troanal.Chem. 1994, 375, 357–366. 324. W. J. Lorenz, H. D. Hermann, N. Wütrich
299. Th. Dretschkow, Th. Wandlowski, Ber. et al., J. Electrochem. Soc. 1974, 121,
Bunsen-Ges. Phys. Chem. 1997, 101, 1167–1177.
749–757. 325. R. Adzic, E. Yeager, B. D. Cahan, J. Elec-
300. A. M. Funtikov, U. Linke, U. Stimming trochem. Soc. 1974, 121, 474–484.
et al., Surf. Sci. 1995, 324, L343–L378. 326. K. Jüttner, W. J. Lorenz, Z. Phys. Chem. N.F.
301. A. M. Funtikov, U. Stimming, R. Vogel, J. 1980, 122, 163–185.
Electroanal. Chem. 1997, 428, 147–153. 327. G. Kokkinidis, J. Electroanal. Chem. 1986,
302. L. J. Wan, S. L. Yau, K. Itaya, J. Phys. Chem. 201, 217–236.
1995, 99, 9507–9513. 328. R. Adzic, in Advances in Electrochemistry
303. M. Wilms, P. Broekmann, M. Kruft et al., and Electrochemical Engineering (Eds.:
Surf. Sci. 1998, 404, 83–86. H. Gerischer, C. Tobias), John Wiley &
304. M. Wilms, P. Broekmann, C. Stuhlmann Sons, New York, 1984, pp. 159, Vol. 13.
et al., Surf. Sci. 1998, 416, 121–140. 329. K. Engelsmann, W. J. Lorenz, E. Schmidt,
305. P. Broekmann, M. Wilms, M. Kruft et al., J. Electroanal. Chem. 1980, 114, 1–10.
J. Electroanal. Chem. 1999, 467, 307–324. 330. O. Melroy, K. Kanazawa, J. J. G. Gordon
306. W. H. Li, R. J. Nichols, J. Electroanal. Chem. et al., Langmuir 1986, 2, 697–700.
1998, 456, 153–160. 331. G. L. Richmond, M. Robinson, V. L. Shan-
307. L. J. Wan, M. Hara, J. Inukai et al., J. Phys. non, Prog. Surf. Sci. 1988, 28, 1–70.
Chem. B 1999, 103, 6978–6983. 332. G. Lehmpfuhl, Y. Uchida, M. S. Zei et al.,
308. L. J. Wan, T. Suzuki, K. Sashikata et al., in Imaging of Surfaces and Interfaces (Eds.:
J. Electroanal. Chem. 2000, 484, 189–193. J. Lipkowski, P. N. Ross), John Wiley &
309. A. Cuesta, M. Kleinert, D. M. Kolb, Phys. Sons, New York, 1999, pp. 57–98.
Chem. Chem. Phys. 2000, 2, 5684–5690. 333. R. Kötz in Advances in Electrochemical Sci-
310. H. Angerstein-Kozlowska, B. E. Conway, ence and Engineering (Eds.: H. Gerischer,
A. Hamelin et al., Electrochim. Acta 1986, Ch. Tobias), VCH, Weinheim, 1990, pp.
31, 1051–1061. 75–109, Vol. 1.
311. Z. Shi, J. Lipkowski, M. Gamboa et al., 334. M. Toney, O. R. Melroy, in Electrochemical
J. Electroanal. Chem. 1994, 366, 317–326. Interfaces – Modern Techniques for in situ
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 463
383. P. Berenz, S. Tillmann, H. Massong et al., 410. A. Bittner, J. Winterlin, G. Ertel, Surf. Sci.
Electrochim. Acta 1998, 43, 3035–3043. 1997, 376, 267–278.
384. E. A. Meguid, P. Berenz, H. Baltruschat, 411. A. A. Aki, G. A. Attard, R. Price et al.,
J. Electroanal. Chem. 1999, 467, 50–59. J. Chem. Soc., Faraday Trans. 1995, 91,
385. E. Herrero, S. Glazier, L. J. Buller et al., 3585–3595.
J. Electroanal. Chem. 1999, 461, 121–130. 412. N. Markovic, H. A. Gasteiger, P. N. Ross,
386. L. Buller, E. Herrero, R. Gomez et al., Langmuir 1995, 11, 4098–4108.
J. Phys. Chem. B 2000, 104, 5932–5939. 413. I. A. Tidswell, C. A. Lucas, N. M. Markovic
387. K. Varga, P. Zelenay, A. Wieckowski, J. Elec- et al., Phys. Rev. B 1995, 51, 10 205–10 208.
troanal. Chem. 1992, 330, 453–467. 414. D. Dickertmann, F. D. Koppitz, J. W.
388. G. Horanyi, J. Electroanal. Chem. 1974, 55, Schultze, Electrochim. Acta 1976, 21,
45–51. 967–971.
389. Y. Shingaya, H. Matsumoto, H. Ogasawara 415. W. J. Lorenz, E. Schmidt, G. Staikov et al.,
et al., Surf. Sci. 1995, 335, 23–31. Faraday Symp. Chem. Soc. 1977, 12, 14–23.
390. I. Oda, Y. Shingaya, H. Matsumoto et al., 416. K. Jüttner, G. G. Staikov, W. J. Lorenz et al.,
J. Electroanal. Chem. 1996, 409, 95–101. J. Electroanal. Chem. 1977, 80, 67–80.
391. M. Watanabe, H. Uchida, N. Ikeda, J. Elec- 417. H. Bort, K. Jüttner, W. J. Lorenz et al.,
troanal. Chem. 1995, 380, 255–260. J. Electroanal. Chem. 1978, 90, 413–424.
392. H. Matsumoto, I. Oda, J. Inukai et al., 418. G. Staikov, K. Jüttner, W. J. Lorenz et al.,
J. Electroanal. Chem. 1993, 356, 275–280. Electochim. Acta 1978, 23, 305–313.
393. H. Matsumoto, I. Oda, J. Inukai et al., 419. G. Staikov, K. Jüttner, W. J. Lorenz et al.,
J. Electroanal. Chem. 1994, 379, 223–231. Electochim. Acta 1978, 23, 319–324.
394. G. Beitel, O. M. Magnussen, R. J. Behm, 420. K. Takayanagi, D. M. Kolb, K. Kambe et al.,
Surf. Sci. 1995, 336, 19–26. Surf. Sci. 1980, 100, 407–422.
395. K. Sashikata, N. Furuya, K. Itaya, J. Elec- 421. J. N. Jovicevic, V. D. Jovic, A. R. Despic,
troanal. Chem. 1991, 316, 361–368. Electrochim. Acta 1984, 29, 1625–1632.
396. Z. L. Wu, Z. H. Zang, S. L. Yau, Langmuir 422. V. D. Jovic, J. N. Jovicevic, A. R. Despic,
2000, 16, 3522–3528. Electrochim. Acta 1985, 30, 1455–1464.
397. H. S. Yee, H. D. Abruna, J. Phys. Chem. 423. V. D. Jovic, B. M. Jovic, A. R. Despic, J. Elec-
1993, 97, 6278–6288. troanal. Chem. 1990, 288, 229–243.
398. R. Durand, R. Faure, D. Aberdam et al., 424. K. J. Stevenson, D. W. Hatchett, H. S.
Electrochim. Acta 1992, 37, 1977–1982. White, Langmuir 1996, 12, 494–499.
399. N. Markovic, H. A. Gasteiger, C. A. Lucas 425. D. A. Koos, V. L. Shannon, G. L. Richmond,
et al., Surf. Sci. 1995, 335, 91–100. J. Phys. Chem. 1990, 94, 2091–2098.
400. C. A. Lucas, N. M. Markovic, I. M. Tidswell 426. U. Müller, D. Carnal, H. Siegenthaler et al.,
et al., Physica B 1996, 221, 245–250. Phys. Rev. B 1992, 46, 12 899–12 901.
401. E. Herrero, L. J. Buller, J. Li et al., Elec- 427. D. Carnal, P. I. Oden, U. Müller et al., Elec-
trochim. Acta 1998, 44, 983–992. trochim. Acta 1995, 40, 1223–1235.
402. R. Michaelis, M. S. Zei, R. S. Zhai et al., 428. U. Schmidt, S. Vinzelberg, G. Staikov, Surf.
J. Electroanal. Chem. 1992, 339, 299–310. Sci. 1996, 348, 261–279.
403. R. Michaelis, D. M. Kolb J. Electroanal. 429. M. G. Samant, G. L. Borges, J. G. Gordon
Chem. 1992, 328, 341–348. et al., J. Am. Chem. Soc. 1987, 109,
404. D. Aberdam, Y. Gauthier, R. Durand et al., 5970–5974.
Surf. Sci. 1994, 306, 114–124. 430. M. G. Samant, M. F. Toney, G. L. Borges
405. M. S. Zei, K. Wu, M. Eiswirthz et al., Elec- et al., Surf. Sci. 1988, 193, L29–L36.
trochim. Acta 1999, 45, 809–817. 431. M. F. Toney, J. G. Gordon, M. G. Samant
406. M. S. Zei, Z. Phys. Chem. 1999, 208, 77–91. et al., J. Phys. Chem. 1995, 99, 4733–4744.
407. Th. Wandlowski, E. Herrero, J. Feliu, 432. M. E. Hansen, E. Yeager, ACS Symp. 1988,
J. Electroanal. Chem., in preparation 378, 398–407.
408. C. A. Lucas, N. M. Markovic, P. N. Ross, 433. L. Laguren–Davidson, F. Lu, G. Salaita
Phys. Rev. B 1997, 56, 3651–3654. et al., Langmuir 1988, 4, 224–232.
409. A. A. Aki, G. A. Attard, J. Phys. Chem. B 434. E. Amman, MS Thesis, University of Bern,
1997, 101, 4597–4606. Bern, Switzerland, 1995.
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 465
435. K. Takayanagi, Surf. Sci. 1981, 104, 460. A. Hamelin, S. Morin, J. Richer et al.,
527–548. J. Electroanal. Chem. 1991, 304, 195–209.
436. E. Altman, R. Colton, Surf. Sci. 1993, 295, 461. A. Popov, R. Naneva, K. Dimitrov et al.,
13–33. Electrochim. Acta 1992, 37, 2369–2371.
437. A. Ullman, Chem. Rev. 1996, 96, 1533–1554. 462. B. Roelfs, H. Baumgärtel, Ber. Bunsen-Ges.
438. R. E. Pagano, R. Miller, J. Colloid. Interface Phys. Chem. 1994, 99, 677–681.
Sci. 1973, 45, 126–137. 463. B. Roelfs, E. Bunge, C. Schröter et al.,
439. A. Nelson, N. Auffret, J. Electroanal. Chem. J. Phys. Chem. B 1997, 101, 754–765.
1988, 244, 99–113. 464. M. H. Hölzle, D. Krznaric, D. M. Kolb,
440. M. Porter, T. B. Bright, D. Allara et al., J. Electroanal. Chem. 1995, 386, 235–239.
J. Am. Chem. Soc. 1987, 109, 3559–3568. 465. M. H. Hölzle, D. M. Kolb, Ber. Bunsen-Ges.
441. L. Strong, G. M. Whitsides, Langmuir 1988, Phys. Chem. 1994, 98, 330–335.
4, 546–558. 466. M. Scharfe, A. Hamelin, C. Buess-Herman,
442. J. Sagiv, J. Am. Chem. Soc. 1980, 102, 92–98. Electrochim. Acta 1995, 40, 61–67.
443. H. Finklea, D. A. Snider, J. Fedyk, Lang- 467. M. H. Hölzle, Th. Wandlowski, D. M. Kolb,
muir 1993, 9, 3360–3367. J. Electroanal. Chem. 1995, 394, 271–275.
444. P. Fenter, P. Eisenberger, K. S. Liang, Phys. 468. Th. Wandlowski, J. Electroanal. Chem. 1995,
Rev. Lett. 1993, 70, 2447–2450. 395, 83–89.
445. O. M. Magnussen, B. M. Ocko, M. Deutsch 469. Th. Wandlowski, M. H. Hölzle, Langmuir
et al., Nature 1996, 384, 250–252. 1996, 12, 6597–6615.
446. A. A. Kornyshev, I. Vilfan, Electrochim. Acta 470. D. Krznaric, B. Cosovic, M. H. Hölzle et al.,
1995, 40, 109–127. Ber. Bunsen-Ges. Phys. Chem. 1996, 100,
447. A. N. Frumkin, Z. Phys. Chem. 1925, 116, 1779–1790.
466–485. 471. H. Striegler, M. H. Hölzle, Th. Wandlowski
448. A. N. Frumkin, Z. Phys. 1926, 35, 792–802. et al., Poster 1a-41, presented at the 47th
449. B. Damaskin, O. A. Petrii, V. V. Batrakov, Annual Meeting of the ISE, November 1–6,
Adsorption of Organic Compounds on Elec- Budapest, 1996.
trodes, Plenum Press, New York 1972. 472. Th. Boland, B. Ratner, Langmuir 1994, 10,
450. R. Parsons, Chem. Rev. 1990, 90, 813–826. 3845–3852.
451. W. Lorenz, Z. Electrochem. 1958, 62, 473. W. Haiss, B. Roelfs, S. N. Port et al., J.
192–202. Electroanal. Chem. 1998, 454, 107–113.
452. D. A. Bonell, Scanning Tunneling Microscopy 474. M. Futamata, J. Phys. Chem. B 2001, 105,
and Spectroscopy, VCH, New York, 1993. 6933–6942.
453. N. J. Tao, in Imaging of Surfaces and In- 475. S. Promkin, Th. Wandlowski, unpublished.
terfaces (Eds.: J. Lipkowski, P. N. Ross), 476. Th. Wandlowski, K. Ataka, D. Mayer, Lang-
John Wiley & Sons, New York, 1999, muir 2002, la u255854.
pp. 211–248. 477. N. J. Tao, J. A. DeRose, S. M. Lindsay,
454. W. H. Liu, W. Haiss, S. Floate et al., Lang- J. Phys. Chem. 1993, 97, 910–919.
muir 1999, 15, 4875–4883. 478. Th. Dretschkow, A. S. Dakkouri, Th. Wand-
455. K. Takamura, F. Kusu, in Methods od Bio- lowski, Langmuir 1997, 13, 2843–2856.
chemical Analysis (Ed.: D. Glick), John Wiley 479. Th. Dretschkow, Th. Wandlowski, Elec-
& Sons, New York, 1987, pp. 155, Vol. 32. trochim. Acta 1998, 43, 2991–3006.
456. H. H. Rotermund, K. Krischer, B. Pettinger, 480. S. Soverby, personal communication. 1998.
in Imaging of Surfaces and Interfaces (Eds.: 481. K. M. Richard, A. A. Gewirth, J. Phys. Chem.
J. Lipkowski, P. N. Ross), John Wiley & 1995, 99, 12 288–12 293.
Sons, New York, 1999, pp. 139–210. 482. G. Andreasen, M. E. Vela, R. C. Salvarezza
457. I. Burges, V. Zamlymny, G. Szymanski et al., Langmuir 1997, 13, 6814–6819.
et al., Langmuir 2001, 17, 3355–3367. 483. W. B. Cai, L. J. Wan, H. Noda et al., Lang-
458. Th. Wandlowski, B. M. Ocko, O. M. Mag- muir 1998, 14, 6992–6998.
nussen et al., J. Electroanal. Chem. 1996, 484. F. Cunha, N. J. Tao, Phys. Rev. Lett. 1995,
409, 155–164. 75, 2376–2379.
459. V. V. Batrakov, B. B. Damaskin, Y. B. Ipa- 485. F. Cunha, N. J. Tao, X. W. Wang et al.,
tov, Elektrokhimiya 1974, 10, 216–220. Langmuir 1996, 12, 6410–6418.
466 3 Specific Adsorption
486. Th. Dretschkow, D. Lampner, Th. Wand- 510. Cl. Bues-Herman, L. Gierst, N. Vanlae-
lowski, J. Electroanal. Chem. 1998, 458, them-Meuree, J. Electroanal. Chem. 1981,
121–138. 123, 1–19.
487. Th. Dretschkow, Th. Wandlowski, J. Elec- 511. Th. Wandlowski, G. B. Jameson, R. de
troanal. Chem. 1999, 467, 207–216. Levie, J. Phys. Chem. 1993, 97,
488. Th. Dretschkow, Th. Wandlowski, Elec- 10 119–10 126.
trochim. Acta 1999, 45, 731–740. 512. Th. Wandlowski, R. de Levie, J. Electroanal.
489. H. Noda, T. Minoha, L. J. Wan et al., Chem. 1992, 329, 103–127.
J. Electroanal. Chem. 2000, 481, 62–68. 513. R. Guidelli, M. L. Foresti, J. Electroanal.
490. D. Mayer, K. Ataka, Th. Wandlowski, Chem. 1986, 197, 123–141.
J. Electroanal. Chem. 2002, in press. 514. Th. Wandlowski, P. Chaiyasith, H. Baum-
gärtel, J. Electroanal. Chem. 1992, 346,
491. F. Cunha, Q. Jing, N. J. Tao, Surf. Sci. 1997,
271–279.
389, 19–28.
515. D.M. Mohilner, in Electroanal. Chem. (Ed.:
492. O. Dominguez, L. Echegoyen, F. Cunha
A. J. Bard), Marcel Dekker, New York, 1966,
et al., Langmuir 1998, 14, 821–824.
pp. 241–405, Vol. 1.
493. J. Pan, S. M. Lindsay, N. J. Tao, Langmuir 516. J. Lipkowski, W. Schmickler, D. M. Kolb
1993, 9, 1556–1560. et al., J. Electroanal. Chem. 1998, 452,
494. E. Bunge, R. J. Nichols, H. Baumgärtel 193–197.
et al., Ber. Bunsen-Ges. Phys. Chem. 1995, 517. S. Wu, J. Lipkowski, O. M. Magnussen
99, 1243–1246. et al., J. Electroanal. Chem. 1998, 446, 67–77.
495. E. Bunge, R. J. Nichols, B. Roelfs et al., 518. L. Stolberg, J. Lipkowski, in Adsorption of
Langmuir 1996, 12, 3060–3066. Organic Molecules at Metal Electrodes (Eds.:
496. A. S. Dakkouri, D. M. Kolb, R. Edelstein, J. Lipkowski, P. N. Ross), VCH, New York,
Shima et al., Langmuir 1996, 12, 1992, pp. 171–238.
2849–2852. 519. U. Retter, H. Jehring, V. Vetterl, J. Elec-
497. L. J. Wan, H. Noda, Y. Hara et al., J. Elec- troanal. Chem. 1974, 57, 391–397.
troanal. Chem. 2000, 489, 68–75. 520. B. B. Damaskin, N. K. Akhmetov, Sov. Elec-
498. L. J. Wan, M. Terashima, H. Noda et al., J. trochem. 1979, 15, 478–481.
Phys. Chem. B 2000, 104, 3563–3569. 521. Yu. Kharkats, J. Electroanal. Chem. 1980,
499. N. Batina, M. Kunikate, K. Itaya, J. Elec- 115, 75–88.
troanal. Chem. 1995, 405, 245–250. 522. Yu. Ya. Gurevich, Yu. I. Kharkats, J. Elec-
500. M. Kunikate, N. Batina, K. Itaya, Langmuir troanal. Chem. 1978, 86, 245–258.
1995, 11, 2337–2340. 523. U. Retter, H. Lohses, J. Electroanal. Chem.
501. K. Ogaki, N. Batina, M. Kunikate et al., J. 1982, 134, 243–250.
Phys. Chem. 1996, 100, 7185–7190. 524. P. Nikitas, J. Electroanal. Chem. 1991, 300,
502. S. L. Yau, Y. G. Kim, K. Itaya, J. Am. Chem. 607–628.
Soc. 1996, 118, 7795–7803. 525. P. Nikitas, Electrochim. Acta 1991, 36,
447–457.
503. L. J. Wan, K. Itaya, Langmuir 1997, 13,
526. P. Nikitas, S. Andoniou, J. Electroanal.
7173–7179.
Chem. 1994, 375, 339–356.
504. R. Srinivasan, J. C. Murphy, R. Fainchtein
527. U. Retter, J. Electroanal. Chem. 1984, 165,
et al., J. Electroanal. Chem. 1994, 312,
221–230.
293–300. 528. U. Retter, J. Electroanal. Chem. 1987, 236,
505. R. Srinivasan, R. Gopalan, J. Phys. Chem. 21–30.
1993, 98, 8770–8775. 529. U. Retter, J. Electroanal. Chem. 1993, 349,
506. N. J. Tao, Z. Shi, Surf. Sci. 1994, 321, 41–48.
L149–L156. 530. U. Retter, Electrochim. Acta 1996, 41,
507. N. J. Tao, Z. Shi, J. Phys. Chem. 1994, 98, 2171–2174.
7422–7426. 531. R. Srinivasan, R. de Levie, S. K. Rangarajan,
508. N. J. Tao, Phys. Rev Lett. 1996, 76, Chem. Phys. Lett. 1987, 142, 43–47.
4066–4069. 532. C. Young, Phys. Rev. 1952, 85, 808–816.
509. L. J. Wan, C. Wan, C. Bai et al., J. Phys. 533. Y. I. Kharkats, U. Retter, J. Electroanal.
Chem. B 2001, 105, 8399–8402. Chem. 1990, 287, 363–367.
3.3 Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces 467
534. U. Retter, J. Electroanal. Chem. 1992, 329, 548. P. Steel, Coord. Chem. Rev. 1990, 106,
81–89. 227–265.
535. R. Sridharan, R. de Levie, J. Phys. Chem. 549. E. B. Constable, Adv. Inorg. Chem. 1989, 34,
1982, 86, 4489, 4490. 67–138.
536. R. Philipp, U. Retter, J. Dittrich et al., Elec- 550. L. L. Merritt, E. Schroeder, Acta Crystallogr.
trochim. Acta 1987, 32, 1671–1677. 1956, 9, 801–804.
537. U. Retter, V. Vetterl, J. Jursa, J. Electroanal. 551. E. A. Mambetkaziev, A. M. Shaldybaeva,
Chem. 1989, 274, 1–9. V. N. Statsyuk et al., Sov. Electrochem. 1975,
538. Th. Wandlowski, J. Electroanal. Chem. 1991, 11, 1643–1645.
302, 233–253. 552. N. K. Akhmetov, R. I. Kaganovich, E. A.
539. S. Laushera, E. Stenina, Sov. Electrochem. Mambetkaziev et al., Sov. Electrochem. 1977,
1992, 28, 78–87. 13, 248–250.
540. V. S. Griffiths, J. B. Westermore, J. Chem. 553. N. K. Akhmetov, R. I. Kaganovich, E. A.
Soc. 1963, 4941–4945. Mambetkaziev et al., Sov. Electrochem. 1978,
541. L. K. Partridge, A. C. Tansley, A. C. Porter, 14, 1534–1537.
Electrochim. Acta 1967, 12, 1573–1580. 554. L. Pospisil, J. Kuta, J. Electroanal. Chem.
542. B. B. Damaskin, S. L. Dyatkina, S. I. 1979, 101, 391–398.
Petrochenko, Sov. Electrochem. 1969, 5, 555. M. Kim, K. Itoh, J. Phys. Chem. 1987, 91,
935–941. 126–132.
543. A. A. Moussa, H. A. Ghaly, M. M. Abou- 556. D. Yang, D. Bizotto, J. Lipkowski et al.,
Romia, Electrochim. Acta 1975, 20, 485–497. J. Phys. Chem. 1994, 98, 7083–7089.
544. L. Gierst, C. Franck, G. Quarin et al., 557. M. Hoon-Khosla, W. R. Fawcett, J. D. God-
J. Electroanal. Chem. 1981, 129, 353–363. dard et al., Langmuir 2000, 16, 2356–2369.
545. W. Obretenov, V. Bostanov, V. Popov, 558. J. Trost, J. Wintterlin, G. Ertl, Surf. Sci.
J. Electroanal. Chem. 1982, 132, 273–276. 1995, 329, L583–L587.
546. M. Poelman, C. Buess-Herman, J. P. Badi- 559. G. E. Piorier, Chem. Rev. 1997, 97,
ali, Langmuir 1999, 15, 2194–2201. 1117–1129.
547. A. Popov, N. Dimitrov, T. Vitanov, Elec-
trochim. Acta 1992, 37, 2373–2376.
471
10000
Other
Dental
×10 Glass
8000 Electronics
Chemical
Jewellery
Autocatalyst
[kilo-ounce]
6000
Demand
×10
4000
2000
×10
0
um um um um um
tin di di ni di
a lla o e Iri
Pl Pa Rh th
Ru
Fig. 1 Principal uses of noble metals (platinum, palladium, rhodium, ruthenium and iridium) in the
world in 2000 [2]. The scales for rhodium, ruthenium and iridium are enlarged by 10 times.
of many metals have become possible than that of the metal itself, a three-
now [23, 24]. dimensional (3D) island growth mecha-
In this chapter, we describe the recent in nism, namely, the VW mode is favored in-
situ atomic resolution studies on the elec- dependent of crystallographic mismatch.
trochemical deposition and dissolution of When the binding energy between the
noble metals, which show extremely high metal and the substrate is higher than that
catalytic activity for many chemical reac- of the metal itself, a two-dimensional (2D)
tions [1, 2]. metallic monolayer can be formed in
the underpotential deposition (UPD) re-
4.1.2 gion, and there are two possible growth
Atomically Controlled Deposition of Noble mechanisms for the subsequent growth
Metals of the metallic layers depending on the
crystallographic mismatch. When the crys-
Depending on the binding energy and tallographic mismatch of the metal and
crystallographic mismatch between the the substrate is negligibly small, metallic
deposited metal and the substrate, the layers are formed on the UPD layer on
growth of a metallic layer on the sub- the substrate by a layer-by-layer growth,
strate near equilibrium condition can namely, the FM mode. When the crys-
be classified into three mechanisms, tallographic mismatch is relatively large,
namely, the Vomer-Weber (VW), the the UPD layer contains considerable in-
Stranski-Krastanov (SK), and the Frank- ternal strain, and therefore, the growth of
van der Merwe (FM) modes [25–27]. Gen- the unstrained 3D metal islands on top
erally, when the binding energy between of the strained UPD layers is energetically
the metal and the substrate is lower favored and this mechanism is called the
Tab. 1 Physical properties and electrochemical growth modes of noble metal elements on gold substrate
Au Pt Pd Rh Ru Os Ir Ag
Atomic number 79 78 46 45 44 76 77 47
Atomic weight 196.97 195.08 106.42 102.90 101.07 190.2 192.2 107.87
Lattice structure fcc fcc fcc fcc hcp hcp fcc fcc
Electronic structure 5d10 6s1 5d9 6s1 4d10 5s0 4d8 5s1 4d7 5s1 5d6 6s2 5d7 6s2 4d10 5s1
Atomic diameter rA [nm] 0.288 0.277 0.275 0.269 0.270 0.273 0.271 0.289
Special gravity [g cm−3 ] 19.32 21.45 12.01 12.41 12.37 22.59 22.56 10.50
Melting point [◦ C] 1064 1768 1554 1963 2250 3045 2457 962
Thermal conductivity [Wcm−1 K−1 ] 2.93 0.72 0.76 1.50 1.05 0.87 1.47 4.19
Resistivity [µ cm, 20 ◦ C] 2.2 10.6 10.8 4.6 7.3 9.5 5.1 1.62
Tensile strength [Mpa]a 108 123 170 695 – – 1088 147
Hardness [Vickers]a 22 41 41 101 – 350 220 24
Surface energy νA [Jm−2 ] [28] 1.488 2.691 2.043 2.828 3.409 3.947 3.231 1.302
Geometry parameter rAB b 1 0.962 0.955 0.934 0.937 0.948 0.941 1.00
Lattice mismatch AB c 0 −3.8% −4.6% −6.7% −6.2% −5.2% −5.8% 0.17%
(continued overleaf )
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals
473
474
Tab. 1 (continued)
4 Underpotential Deposition
Au Pt Pd Rh Ru Os Ir Ag
Surface energy mismatch AB d 0 0.575 0.314 0.620 0.802 0.905 0.739 0.133
Electrochemical growth mode on FM SK FM SK VW FM
gold substratee [29–31] [32] [33–36] [37] [38–40] [41–45]
a Annealed;
b Geometric parameter rAB = rB /rA [26] where A and B denote substrate and metal, respectively. Values in the table are calculated with
respect to the gold substrate;
c Lattice mismatches
AB [26] are shown with respect to that of gold;
d The surface-energy mismatches are calculated with respect to a gold substrate following the definition given by Bauer and
mode: Stranski-Krastanov mode, unstrained 3D islands are formed on top of strained 2D layers. See text for details.
Note: fcc: face centered cubic; hcp: hexagonal close packed.
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 475
SK mode. Thus, the metal growth mode on If one examines the parameters shown in
the substrate can be estimated qualitatively Table 1, one can expect that electrochem-
from the properties of both deposited and ical deposition of gold and palladium on
substrate metals. a gold electrode should be the best candi-
Physical properties such as lattice and date for the layer-by-layer growth process
electronic structures, atomic diameters, among these noble metal elements. The
melting point, thermal conductivity, re- observed growth modes shown in the table
sistivity, tensile strength, hardness as well are in qualitative agreement with expecta-
as surface energy (νA ) [28] of the noble tions based on the physical parameters.
metal elements gold, platinum, palladium,
rhodium, ruthenium, osmium, and irid- 4.1.2.1 Platinum
ium are summarized in Table 1. The same Platinum is the most important mate-
properties for silver are also shown in rial in catalysis including electrocatal-
the same table for comparison. Following ysis [46–48]. For example, platinum is
the definition given by Bauer and cowork- widely used as a catalyst for automo-
ers [26], geometry parameter (rAB ), lattice biles [2]. Platinum and its alloys are also
mismatch (AB ) and surface-energy mis- employed as the main catalytic materials
match (AB ) in the table are given with in fuel cells, which are attracting more
respect to gold, which is widely used as a attention due to their high-energy conver-
substrate in studies of electrochemical epi- sion efficiency, flexibility, and low/clean
taxial growth of noble metals. The experi- emission than those of traditional internal-
mentally observed electrochemical growth combustion engines [49, 50]. The pla-
modes of these metals on gold are also tinized platinum electrode [51] has been
listed in the table. These parameters play used for standard reversible hydrogen elec-
significant roles in the growth mechanism trodes (RHE) because of its extremely high
for the electrochemical deposition of these reversibility for hydrogen evolution and
metals on various substrates. As shown in oxidation reactions.
the table, the lattice mismatches (AB ) are Ultrathin platinum layers on a chemi-
observed from −3.8% and −6.7% among cally inactive substrate are very important
these noble metal elements, and platinum systems in the field of catalysis and electro-
shows the smallest mismatch (−3.8%) rel- catalysis. Many studies on epitaxial growth
ative to the gold substrate. The value of of platinum thin layers on various sub-
AB , which is calculated from the sur- strates have been reported in the UHV
face energy of each metal with respect environment [4, 52–55]. The electrochem-
to a gold substrate [28], decreases in the ical growth of platinum layers is achieved
sequence of gold (0), palladium(0.314) < by using a number of commercially avail-
platinum(0.575) < rhodium(0.620) < iri- able electroplating baths [56–58]. These
dium(0.739) < ruthenium(0.802) < osm- studies are mainly limited to the purpose
ium(0.905). As proposed by Bauer and of electroplating where film thickness, pu-
coworkers, the value of AB can be em- rity, film strength, adhesive force to the
ployed to judge the growth mode of metal substrate and growth rate are more im-
deposition on various substrates [26]. AB portant than surface atomic structure [46,
values <0.5 with small lattice mismatch 47, 51, 59–62]. The atomically controlled
(rAB ≤ 1.00) are required for the 2D crys- electrochemical deposition of platinum on
talline growth of the metal on the substrate. a gold single-crystal substrate was realized
476 4 Underpotential Deposition
50 400
End
0 300
Start
Mass change
[µA cm−2]
−50
Current
End 200
[ng]
−100
100
−150
Start 0
−200
500 600 700 800 900 1000 1100
Potential vs RHE
[mV]
Fig. 2 Potential dependence of current (solid line) and the mass
change (dotted line) at a (111)-such as gold/EQCM electrode in
50 mM HClO4 + 0.6 mM H2 PtCl6 . Sweep rate: 20 mV s−1 [32].
(a) 0 20 40 nm (b) 0 20 40 nm
(c) 0 20 40 nm
Fig. 3 STM images (40 × 40 nm2 ) of platinum deposition process on an Au(111) substrate
(a) at +0.95 V, (b) 10 min and (c) 30 min, respectively, after the potential was stepped from
+0.95 V to +0.70 V in 50 mM H2 SO4 + 0.05 mM H2 PtCl6 [32].
into the blank solution√ at the √ same from the top-right portion of the image
potential [64]. The same 7 × 7R19.1◦ (Fig. 3b) and covered the entire area of the
adlattice structure of the PtCl6 2− com- image after prolonged deposition (Fig. 3c).
plex on an Au(111) electrode has been The bottom-left portion of Fig. 3(b) is the
confirmed by Kolb and colleagues re- Au(111)
√ covered with a PtCl6 2− adlayer
√
cently [65]. As will be described later, of 7 × 7R19.1◦ structure (c.f. Fig. 3a).
similar adlattice structures of metal com- The terrace size of the deposited platinum
plexes, such as PdCl4 2− [33], RhCl6 2− [37], layer seemed to be much smaller than that
AuCl4 − [66] and PtCl4 2− [67], have also of the Au(111) surface before the platinum
been observed on the Au(III) electrode deposition. A number of small clusters
surface. were also observed on the platinum layer.
As shown in Figs. 3(b) and (c), the depo- These clusters are higher than the first plat-
sition of a platinum layer of monoatomic inum layer by a monoatomic height of the
height was initiated as soon as the potential platinum layer. The third platinum layer
was stepped to +0.70 V. The growth of the started to grow before the growth of the
first layer seemed to be essentially in a 2D second platinum layer was completed. The
mode. The deposited platinum layer grew morphology of the subsequent platinum
478 4 Underpotential Deposition
m
3n
0.3
m
6n
0.7
m
6n
0.7
√ √
layers became rougher than that of the the adlattice structure of 7 × 7R19.1◦
first layer. This means that the growth of disappeared but a highly ordered surface
the platinum layer on an Au(111) electrode structure of a hexagonal array with a neigh-
is not a perfect 2D growth, namely, the FM bor atomic distance of 0.28 nm (Fig. 4b).
mode. However, Friedrich and colleagues Although it is difficult to distinguish the
observed that platinum deposition on an distance of the unit lattice of Pt(111)
Au(111) surface was mainly observed at (0.277 nm) and Au(111) (0.288 nm) sur-
the step edges of the substrate as isolated faces by the present STM measurement,
platinum particles at the lower coverage the image shown in Fig. 4(b) should cor-
by ex situ STM measurements [68]. Recent respond to the Pt(111) – (1 × 1) structure
investigation on the initial stage of plat- formed on the Au(111) substrate as both
inum deposition on Au(hkl) surfaces by EQCM and XPS measurements confirmed
Kolb and colleagues also showed that the the existence of platinum on the Au(111)
nucleation of platinum occurred preferen- surface [32]. Adsorption of the PtCl6 2−
tially at the surface defects and that the complex is considered to inhibit the 3D
mobility of platinum was apparently in- growth but to promote the initial 2D
creased by the addition of chloride into the growth of platinum [32].
solution [65]. √ On the other hand, X-ray diffraction
√The adlayer with a structure of 7 × (XRD) measurements of the deposited
7R19.1◦ was also observed on the surface platinum films (ca. 250 ML) on the
of deposited platinum, demonstrating that Au(111) substrate demonstrated that a
the PtCl6 2− complex adsorbed not only on (111)-orientated platinum-bulk phase was
the gold substrate but also on the plat- definitely formed on the Au(111) elec-
inum surface during the electrochemical trode surface by electrochemical epitaxial
deposition process (Fig. 4a). In situ STM growth [32].
measurement of the platinum-deposited The aforementioned experimental re-
Au(111) surface in 50 mM HClO4 solution sults suggest that the platinum layer
without the PtCl6 2− complex showed that grows on the Au(111) electrode by the
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 479
SK mechanism. This may be related to the market price of palladium since 1999 may
relatively large AB of platinum with re- affect its applications in the industry [2].
spect to the gold substrate (0.575) and the Since Clavilier proposed a simple way
moderately slower surface diffusion rate of to prepare a noble metal single crystal [9],
platinum atoms [69, 70]. many electrochemical studies have been
Electrochemical characterizations of the carried out on platinum, gold, and irid-
UPD of hydrogen and copper show that ium as well as palladium single-crystal
the surface structure of the electrochem- electrodes. The preparation of a palladium
ically deposited platinum layer on the single-crystal is, however, difficult because
Au(111) substrate was quite different from palladium cannot be subjected to the usual
that of a polycrystalline platinum elec- convenient flame-annealing treatment to
trode, but more similar to that of the prepare the electrode [73]. It is of great in-
Pt(111) single-crystal electrode [32]. The terest if one can grow a palladium epitaxial
difference in the electrochemical behavior layer on other noble metal single-crystal
between the platinum-deposited Au(111) surfaces, which are easily prepared by Clav-
and an ideal Pt(111) electrode is attributed ilier’s method. This provides not only a
to the relatively small (111) domain and convenient method of preparing an or-
the higher step density on the platinum- dered palladium surface but also a way
deposited surface. Recently, Watanabe and of investigating the chemical and physical
colleagues reported that the (111)-like sur- properties of a thin metallic layer, which
face structure of a platinum thin-film may be different from those of the bulk
electrode prepared by a sputtering method material.
can be improved by electrochemical an- The investigation on the electrochem-
nealing at the potential region of hydrogen ical epitaxial growth of palladium on
adsorption in HClO4 solution [71]. gold and platinum single-crystal sub-
strates was started recently. Attard and
coworkers reported the irreversible depo-
4.1.2.2 Palladium sition of submonolayer and monolayer
As shown in Fig. 1, palladium is mainly coverage of palladium on Pt(111) elec-
consumed as an autocatalyst now [1, 2]. trode by the immersion technique for
Palladium is also an important noble metal the first time [74]. Llorca and coworkers
element widely applied in the development investigated the irreversibly adsorbed pal-
of electrocatalytic materials [46, 48, 72]. It ladium on Pt(hkl) in acidic solution [75]
is widely used as a promoter for electroless and reported that electrocatalytic activity
deposition of metals on various substrates. for the oxidation of formic acid on the
Palladium is also known for its extraor- Pt(100) electrodes was improved drasti-
dinary ability to absorb a large amount cally by the palladium-adlayer modifica-
of hydrogen. The growth of palladium tion, while that on the Pt(111) electrode
thin layers on various substrates in UHV was not greatly affected by the Pd-
has been investigated in detail [3–8]. Elec- modification [76]. However, it is difficult
trodeposition of palladium is widely used, to prepare an ultrathin film of palladium
and a number of commercially available with various thickness by the immersion
electroplating baths of palladium for differ- technique, and STM observation of the
ent purposes have been developed [56–58, atomic structure of these surfaces is not
61, 72]. However, the rapid increase in the available yet.
480 4 Underpotential Deposition
30
[µA cm−2]
Current
(a) −30
500
400
300
[ng cm−2]
∆m
200
100
105 nm × 105 nm 23 nm × 23 nm
0 10 nm
(a) (b)
related to deposition and dissolution of in very good agreement with the results
palladium on the Au(111) surface [33]. of coverage estimated from the EQCM
Kolb and colleagues reported recently that measurements [33]. A similar but not iden-
the characteristic features of the current- tical adlattice structure of the PdCl4 2−
potential curves for palladium deposition complex on an Au(111) substrate was ob-
on gold single-crystal electrodes depend served by Kolb and colleagues as shown in
significantly on the surface structure and Fig. 6(b) (105 × 105 nm2 ) [35]. This image
concentration of the PdCl4 2− complex as reveals the existence of an ordered adlayer
well as the chloride ion contained in the with parallel stripes separated by about
solution [35]. 1.8 nm and a distorted hexagonal struc-
Figure 6 shows STM images of the ture (inset, 23 × 23 nm2 ) incommensurate
Au(111) electrodes in dilute H2 SO4 so- with the Au(111) substrate. The approxi-
lution containing 0.1 mM PdCl4 2− at the mate next-neighbor distances were 0.81,
potential where no palladium deposition 0.68 and 0.73 nm. Domain boundaries and
takes place as reported by (a) Uosaki and small domains with three directions are
colleagues [33] and (b) later by Kolb and clearly observed.
colleagues [35]. Figure 6(a) (10 × 10 nm2 ) Figure 7 shows the surface structure on
shows an adlattice in hexagonal symme- the Au(100) electrode in dilute H2 SO4 so-
try with a nearest-neighbor distance of lution containing the PdCl4 2− complex at
0.78 nm [33]. This structure is very similar the potential where no palladium deposi-
to that of the PtCl6 2− adlayer on an Au(111) tion takes place as reported by (a) Uosaki
electrode (Fig. 3a) and is attributed to the and colleagues [34] and (b) later by Kolb
structure of the adsorbed PdCl4 2− complex and colleagues [36]. The two images show
on the Au(111)
√ √ with an adlattice struc- essentially the same structure of an ad-
ture of 7 × 7R19.1◦ . This result is lattice with long and short axis lengths of
482 4 Underpotential Deposition
4.5 nm × 4.5 nm
(001)
(010)
(a) 0 5 nm (b)
(c)
Fig. 7 (a) Lowpass filtered-STM white box [34]; (b) STM image (4.5 × 4.5 nm2 )
image (5 × 5 nm2 ) of an Au(100) surface at of an Au(100) surface at 0.61 V in 0.1 M
+1.05 V in 50 mM H2 SO4 + 0.5 mM PdCl4 2− . H2 SO4 + 0.5 mM PdCl4 2− + 0.6 mM HCl [36];
The inset is the Fourier transformed 2D (c) A ball model of the PdCl4 2− complex
spectrum and a unit lattice determined from the adsorbed on Au(100) surface [34].
spectrum is superimposed on the image with a
1.47 nm and 1.03 nm and can be attributed conditions, suggesting that the adsorption
to the adlattice structure of the adsorbed of PdCl4 2− on Au(100) is weaker than that
PdCl4 2− complex on the Au(100) surface. on Au(111) [34].
An adlattice model is given in Fig. 7(c). The Figure 8 shows STM images (300 ×
STM image of the ordered-adlayer struc- 300 nm2 ) of the Au(111) electrode sur-
ture on the Au(100) surface was hard to face during the palladium deposition
obtain when the concentration of PdCl4 2− in 50 mM H2 SO4 solution containing
species was lower than 0.05 mM, while 0.5 mM PdCl4 2− [33]. Atomically flat and
the ordered PdCl4 2− adlayer was observed large terraces (Terrace-1 and Terrace-2)
on the Au(111) surface under the same with monoatomic steps on the Au(111)
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 483
0.80 V
0.95 V
(a) (b)
1.0 Pd-1
1.0 Pd-1
nm
nm
0 0
Au Au
Substrate Substrate
−1.0 −1.0
1 min 2 min
(c) (d)
1.0 Pd-2 1.0
nm
nm
3 min 28 min
(e) (f)
1.0 1.0 Pd-3
nm
nm
0 Pd-2 0 Pd-2
Pd-1 Pd-2 Pd-1 Pd-1 Pd-2
Pd-1
Au Au
−1.0 −1.0
Substrate Substrate
(e') 0 300 nm (f') 0 300 nm
48 min 90 min
(g) (h)
1.0 1.0 Pd-4
Pd-3 Pd-4
Pd-4
nm
nm
Fig. 8 (Continued)
substrate were observed at +0.95 V (d) 2 min, (e) 3 min, (f) 28 min, (g) 48 min
where no palladium deposition takes and (h) 90 min after the potential of the
place (Fig. 8a). Figures 8(b–h) show se- Au(111) was stepped from +0.95 V to
quentially obtained STM images of +0.80 V as indicated by the thick arrow
the same area at (b) 0 min, (c) 1 min, in Fig. 8(b) where palladium deposition is
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 485
expected to take place. The palladium nu- reactant, namely, PdCl4 2− , between the
clei were being generated on the Au(111) STM tip and the electrode surface, leading
substrate immediately after the potential to a negative shift in the equilibrium
step. The cross section (Fig. 8b ) along the potential for palladium deposition [78].
dotted line shows the formation of palla- The palladium layer, which was originally
dium islands with a monoatomic height first deposited on Terrace 1, was also
on the large terraces of the gold substrate. grown laterally and finally merged with
In the upper part of Fig. 8(b), 2D growth Pd-3 on Terrace 2 (Figs. 8f,g,h). The STM
of the first palladium monolayer (Pd-1) images in Fig. 8 demonstrate that the
progressed not only on the large terrace electrochemical deposition of palladium
but also on a narrow terrace between the on the Au(111) surface is an epitaxial layer-
step lines. This result is in contrast to that by-layer process.
observed by Kolb and colleagues, in which Figure 9 shows STM images (300 ×
the palladium layer started to preferentially 300 nm2 ) of an Au(100) electrode ob-
grow from the step-and-edge sites [35, 77]. tained in 50 mM H2 SO4 solution con-
This discrepancy should be due to the
taining 0.5 mM PdCl4 2− [34]. A number
differences in the deposition potential and
of monoatomic-high gold islands were
solution conditions.
observed in the STM image at +1.0 V
As shown in Figs. 8(c) and (c ), after the
(Fig. 9a) when no palladium deposition
first palladium layer (Pd-1) was completely
took place. These gold islands were
formed on the gold substrate, islands
formed as a result of the reconstruc-
of the second palladium layer (Pd-2)
tion lifting of the Au(100) surface due
were generated on both Terrace-1 and
to the large difference (25%) in sur-
Terrace-2, and the first palladium layer
face densities between the reconstructed
(Pd-1) on the narrow terrace started
to grow laterally further onto the Pd-1 and unreconstructed Au(100) surface [81].
layer on Terrace-2 from the step line. Figures 9(b)–(g) show STM images se-
Both the Pd-2 layer on Terrace-2 and quentially obtained during the palladium
the palladium layer extended from the deposition process. As soon as the poten-
Pd-1 layer on the narrow terrace grew tial of the Au(100) substrate was stepped
two-dimensionally and merged together, from +1.0 to +0.80 V (thick arrow in
resulting in a large flat terrace of a second Fig. 9b), palladium nuclei were generated
palladium layer on Terrace-2 (Figs. 8d, e). both on the wide terrace and on the gold
The cross section (Figs. 8c ,d ,e ) clearly islands (cross section I). The height of the
shows the 2D growth of the palladium palladium nuclei corresponded to that of
monoatomic layer. After the complete a palladium monoatomic step (0.19 nm).
second palladium layer was formed on Kolb and colleagues reported that the pal-
both Terrace-1 and Terrace-2, the third ladium nucleated at the steps and island
palladium layer (Pd-3) started to grow two- rims on the island-covered Au(100) elec-
dimensionally on the terraces and from trode surface and that the first layer of
the step line (Figs. 8f,g). The growth rate palladium two-dimensionally grew from
of the third palladium layer was, however, these nuclei [36].
slower than that of the first and the As shown in the cross section II in
second layers. This should be due to the Fig. 9(b), the growth of the first pal-
decrease in the local concentration of the ladium layer on the Au(100) surface
486 4 Underpotential Deposition
10 s
(II)
0.8 V
(I)
1.0 V
1.0
nm
0.0 Au Au
Au substrate
−1.0
(a) 0 300 nm 1.0
Pd
11 s
nm
0.0 1st Pd
Au Au 1st Pd
Au substrate
Cross section (II)
−1.0
1.0 Pd
Pd
nm
0.0 Au Au
Au substrate
Cross section (I)
56 s −1.0
(b) 0 300 nm
0.5
1st Pd
nm
0.0 2nd Pd
2nd Pd 1st Pd
Au 1st Pd Au
Au substrate
−0.5
(c) 0 300 nm
was completed within 10 s after the po- the second layer of palladium seemed
tential step, and the second palladium to grow laterally from the deposited pal-
layer then started to grow (Fig. 9c). As ladium layer just on top of the gold
shown in the cross section in Fig. 9(c), islands. Growth of the second layer of
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 487
103 s 291 s
148 s 335 s
0.5 0.5
2nd Pd
2nd Pd 3rd Pd 3rd Pd
3rd Pd
nm
nm
635 s 1110 s
680 s 1155 s
0.5 0.5
Fig. 9 (Continued)
palladium on the gold terrace was com- reflecting the difference in atomic di-
pleted within about 30 s. It was noted ameters of gold (0.288 nm) and palla-
that, although the gold terraces were cov- dium (0.275 nm).
ered with two layers of palladium, the The cross section of Fig. 9(d) (103 ∼
gold islands were covered with only one 148 s) clearly shows that the third layer
monolayer of palladium. At this stage, of palladium was also initiated from
the height difference between the two re- the deposited palladium layer on top of
gions is very small but still distinguishable, the gold islands and grew laterally. As
488 4 Underpotential Deposition
500 nm × 500 nm
0.65 V
0.52 V
0.65 V
0.52 V
0.47 V
0.47 V
0.43 V
0.43 V 0.39 V
Fig. 10 Sequence of STM images for palladium deposition on island-free Au(100) in 0.1 M
H2 SO4 + 0.1 mM PdCl2 + 0.6 mM HCl solution [36]. Deposition potentials as indicated in
the figure. IT = 2nA.
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 489
Even after four layers of palladium were islands were much smaller than the ini-
deposited on the Au(100) terrace and three tial gold islands. Kolb and colleagues
layers of palladium on the gold islands, attributed this phenomenon to the for-
the shape of the gold islands was still mation of a palladium-gold surface al-
distinguishable. loy on the Au(100) electrode, although
Recently, Kolb and colleagues com- such an effect was not observed on the
pared the electrochemical deposition be- Au(111) electrode [36]. The surface alloy
havior of palladium on the island-covered was considered to be the main reason
and island-free unreconstructed Au(100) for the deviations in the electrocatalytic
electrode surfaces [36]. By electrochemical behavior of massive Pd(100) and thin
annealing in 0.1 M H2 SO4 + 0.1 mM palladium layers (<3 ML) on Au(100).
PdCl2 + 0.6 mM HCl at +0.80 V (vs. SCE; However, the formation of a surface al-
same for Kolb’s work) where no palladium loy during the dissolution of palladium
deposition takes place, the monoatomic- deposited on an Au(100) electrode has
high gold islands generated by the lift- not been observed by Uosaki and col-
ing of the reconstruction disappeared leagues in their in situ STM [34] and SXS
instantaneously. Figure 10 shows in situ studies [82, 83]. This may be related to
STM images for palladium deposition the differences between the experiments
on an island-free Au(100) surface in the carried out in the two groups, such as
same solution. Palladium deposition com- electrode potential and chloride concen-
menced at +0.52 V, and palladium nu- tration as well as the in situ STM cell
cleated selectively at the monoatomic-high configuration.
steps and grew onto the terraces (Fig. 10b). Because STM is sensitive only to the out-
The second layer of palladium started ermost surface and the spatial resolution
to grow at +0.47 V by creation of small of STM is limited to about ±0.1 nm, the
palladium islands at the steps (Fig. 10c). crystallographic relationship between the
The fourth monolayer started to grow be- palladium layers and the gold substrate
fore the third monolayer was completed, has remained unclear. Recently, Uosaki
where palladium also nucleated on the and colleagues carried out the in situ SXS
terraces (Figs. 10e–f). Square-type struc- measurements to elucidate the detailed
tures of palladium deposits were generated structure of palladium electrochemically
with the third monolayer, revealing the deposited on Au(111) and Au(100) elec-
symmetry of the Au(100) substrate and trode surfaces [82, 83]. The pseudomor-
showing a layer-by-layer epitaxial growth phic layer structure was found only on
of palladium (Figs. 10e–f). The behav- the first monolayer of palladium on the
ior is markedly different from those Au(111) substrate, while the pseudomor-
observed on the island-covered Au(100) phic layer-by-layer structure was observed
electrode surface (Fig. 9) [34, 36]. Fur- on the Au(100) substrate up to 14 ML. The
thermore, when the palladium deposited relaxation of the palladium layer on the
on both island-covered and island-free Au(111) electrode was found between the
Au(100) surfaces was stripped anodically, second and fourth layer, and the palla-
many small monoatomic-high islands and dium then started to grow from the fifth
holes were created at those places where layer with (111) orientation. However, the
palladium had been deposited. These relaxation of palladium on the Au(100)
490 4 Underpotential Deposition
Rh (111)
200
Rh/Pt (111)
200
[µA cm−2]
[µA cm−2]
j
j
0.2 0.6 1.0 0.2 0.7 1.0
E RHE E RHE
[V] [V]
−400 −400
(a) (b)
200
0.2 0.9
[µA cm−2]
E RHE
[V]
j
−400
(c)
Fig. 11 CVs in 0.5 M H2 SO4 solution for (a) a prepared by electrochemical deposition onto a
massive Rh(111) electrode, (b) Pt(111) electrode monolayer Rh-modified Pt(111) electrode in
modified by one monolayer of rhodium using 0.06 mM Rh3+ at +0.6 V for 620 s [89].
electroless deposition and (c) rhodium overlayer
−10
−20
30 nm × 30 nm 10 nm × 10 nm
550 mV
480 mV
(a) (b)
Fig. 13 STM images for Au(111) √ in 0.1√M H2 SO4 + 0.1 mM RhCl3 + 1 mM HCl.
(a) E = +0.65 V, showing the ( 13 × 13) R13.9◦ -superlattice pf RhCl6 3− and (b) the same
structure at +0.55 V with the atomically resolved Au(111) surface after a potential step to
+0.48 V at the moment shown by the arrow [37].
of (a) well-defined (solid line) and (b) 5◦ observed at the stepped-Au(111) surface,
stepped Au(111) electrodes (dashed line) suggesting that the rhodium deposition
in 0.1 M H2 SO4 + 0.1 mM RhCl3 + 1 mM proceeds easier at surface defects. The
HCl. A pair of spikes was only observed rhodium deposited on a gold-electrode sur-
on the well-defined Au(111) surface at face cannot be dissolved again because
+0.52 V (vs. SCE). The deposition of a passivating oxide is formed on the
rhodium at the well-defined Au(111) sur- rhodium layer when the potential is swept
face was started around +0.20 V with a to the positive region [37].
large (+0.10 V) and a small cathodic peak Figure 13 shows in situ STM images at
(+0.18 V). The peak position of the small the potential region around the current
peak coincided with the reduction peak spikes. An STM image with a hexagonal
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 493
structure can be observed in the potential to the adsorption of the RhCl6 3− com-
region more positive than that of the spike plex. When the potential was stepped to
(Fig. 13a).
√ This√ structure is determined +0.48 V, that is, negative relative to the
to be 13 × 13R13.9◦ and is attributed spikes, this adlattice structure disappeared
Rh
bilayer
1 ML Rh
(a) (b)
(c) (d)
(e)
Fig. 14 Sequence of STM images for the deposition of rhodium on Au(111) surface at
+0.19 V (400 × 400 nm2 ) in. 1 M H2 SO4 + 0.1 mM RhCl3 + 1 mM HCl. (a) Bare Au(111)
surface; (b–d) growth of a rhodium bilayer on Au(111) and (e) rhodium cluster formation
on the bilayer [37].
494 4 Underpotential Deposition
and the Au(111) substrate could be imaged a direct methanol fuel cell (DMFC) [49,
atomically (Fig. 13b). 50], which is a key candidate for use
Figure 14 shows a sequence of STM as a pollution-free power supply in the
images (400 × 400 nm2 ) during electro- near future. In fact, the most efficient
chemical deposition of rhodium on an catalyst for the practical DMFC is the
Au(111) electrode in the same solution Pt/Ru based electrode, and ruthenium is
as Fig. 13 [37]. When the potential was known to promote the methanol oxidation
stepped to +0.19 V, a number of rhodium on a platinum surface by providing OHad
islands were formed at the terrace of the species on its surface.
gold substrate (Fig. 14b). Most of the small The electrochemical deposition of a
rhodium islands are immediately covered metallic-ruthenium film is very difficult
by a rhodium bilayer with a total height compared with that of other platinum-
of 0.3 nm. The adsorption of the RhCl6 3− group elements [46, 61, 91–93]. One of the
complex was also observed on the surface reasons may be related to the complicated
of the rhodium bilayer. A rhodium film electrochemistry of ruthenium deposition
with a two-monolayer height covered the and the stability of the Ru-chloro com-
gold surface almost completely (Fig. 14d). plex [92]. For example, it has been reported
The formation of the rhodium islands that the RuCl3 species in HClO4 solution is
shows a fractal growth, indicating a low decomposed partly into RuO2+ , in which
mobility of rhodium atoms on the Au(111) Ru(IV) is present [94].
surface. A number of small rhodium The controlled deposition of ruthe-
clusters were observed on the flat rhodium nium on well-defined surfaces, such
bilayer (Fig. 14e). These results suggest as Pt(hkl) [95–103] and Au(hkl) [38–40],
that the rhodium deposition proceeds has been characterized by electrochemi-
through the SK mode [37]. This behav- cal measurements, Fourier transform in-
ior contrasts significantly with palladium frared reflection-absorption spectroscopy
on Au(111) and Au(100) electrodes where (FT-IRRAS), XPS and STM measure-
a layer-by-layer growth (Frank-van der ments. The interest in these studies is
Merwe growth mode) of palladium was mainly concentrated on the ruthenium
observed clearly. modification of a platinum surface be-
Electrochemical characterization also cause of its extreme importance in
showed that the rhodium layers on Au(111) electrocatalysis. It has been demon-
with a thickness of two monolayers or strated that a ruthenium-deposited Pt(111)
less behave similarly to a well-ordered substrate showed an extremely high
Rh(111) single-crystal electrode, while activity in methanol oxidation com-
thicker rhodium layers are similar to elec- pared to ruthenium-deposited Pt(hkl)
trochemically disordered Rh(111), where electrodes with other crystallographic
the surface has a higher defect density [37]. orientations [98, 99].
The studies of ruthenium deposition
4.1.2.4 Ruthenium are mainly carried out by electrochemi-
The principle application of ruthenium cal and spontaneous (electroless) deposi-
is in the electronics industry for the tion procedures. Friedrich and colleagues
manufacture of resistors [2]. Ruthenium found that the well-reproduced current-
is attracting more attention because of its potential curves of ruthenium-modified
crucial role in electrooxidation catalysis in Pt(111) electrodes were obtained when
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 495
ruthenium was deposited electrochem- worse when 0.1 M HClO4 was used as
ically in the potential region between supporting electrolyte during the electro-
+0.8 V and +0.3 V (vs. RHE) from a chemical deposition [95–97]. On the other
freshly prepared 0.1 M H2 SO4 containing hand, Wieckowski and colleague deposited
5 mM RuCl3 . The reproducibility became ruthenium from 0.1 M HClO4 containing
30
20
10
Current density
[µA cm−2]
−10
−20
Pt(111)
Ru/Pt(111) Edep = 0.6 V
−30
Ru/Pt(111) Edep = 0.3 V
Fig. 15 CVs of Pt(111) modified by
ruthenium in 0.1 M H2 SO4 + 5 mM −40
RuCl3 at +0.6 and +0.3 V in 0.1 M 0 200 400 600 800 1000
HClO4 in comparison to the bare
Pt(111) voltammogram in 0.1 M HClO4 . E (RHE)
Sweep rate: 20 mV s−1 [97]. [mV]
25 nm 25 nm
(a) (b)
Fig. 16 STM images of ruthenium-modified Pt(111) electrodes recorded in 0.1 M HClO4 at
+0.5 V. Deposition was performed prior to imaging in a standard glass cell from 0.1 M
H2 SO4 + 5 mM RuCl3 at +0.6 V for (a) 5 min and (b) 30 min, respectively [97].
496 4 Underpotential Deposition
0.2 nm 0.2 nm
0 0
0 25.0 50.0 0 25.0 50.0
(a) nm (b) nm
0.5 nm 75 0.5 nm
25
0 0
0 25.0 50.0 0 25 50 75 100
(c) nm (d) nm
Fig. 17 STM images of the Pt(111) after spontaneous deposition of ruthenium for (a) 10 s, (b) 20 s,
(c) 40 s (50 × 50 nm2 ) and (d) 90 s (100 × 100 nm2 ). See Table 2 for details [100].
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 497
was estimated to be 0.25 and 0.6 from the the inner monoatomic layer has not yet
STM images and was in agreement with been completed. The bilayer islands, cor-
the calibration by XPS measurement. The responding to 10% of the overall islands,
density of the ruthenium islands depends were observed clearly at coverage of 0.14 to
on the deposition potential and deposition 0.19 ML (Fig. 17c,d). Similar bilayer island
time [97]. structures of vapor-deposited ruthenium
Figure 17 shows STM images of a on a Pt(111) surface have been observed
Pt(111) electrode obtained by Wieckowski by Iwasita and coworkers [103]. Extend-
and colleagues in air after immersion in ing the immersion time more than 90 s
0.1 M HClO4 solution containing 0.5 mM does not result in higher ruthenium cov-
RuCl3 for (a) 10 s, (b) 20 s, (c) 40 s and erage. The limit of ruthenium coverage on
(d) 90 s [100]. In order to keep the Pt(111) the Pt(111) surface was approximately 0.2,
substrate clean for longer STM-imaging much less than that from electrochemical
times, after ruthenium deposition, the deposition. When an Ru-modified elec-
platinum surface was covered with an io- trode was polarized in hydrogen-evolution
dide monolayer to keep the surface free region, the amount of the ruthenium de-
from contamination since the iodide layer posited also decreased, indicating that a
has no effect in the STM topography of ruthenium oxide-reduction process in that
the Ru-modified surface. Ruthenium is- potential region may be accompanied by
lands of nanometer size were observed partial ruthenium dissolution from the
on the Pt(111) surface and their number surface [100].
increased with an increase in immer- Recently, Behm and colleagues investi-
sion time. Table 2 summarizes the results gated the electrodeposition of ruthenium
obtained from these STM images. The is- on reconstructed [38] and unreconstructed
lands are not uniformly distributed over Au(111) electrodes in detail [39]. It was
the surface when the coverage is low found that the nucleation of ruthenium
(Fig. 17a and b). There is no selective islands during the electrodeposition on
growth of islands at the steps, indicat- a reconstructed Au(111) electrode occurs
ing that the electroless deposition is not exclusively at the elbows of the Au re-
nucleated by the crystallographic defects construction and that the fcc area of the
at the surface. When the coverage is above reconstructed Au(111) surface is decorated
0.14 ML, a second layer deposit on top at higher coverage [38]. However, a ran-
of the inner layer was observed, although dom nucleation process was observed on
10 0.01 0.22
20 0.08 0.22
40 0.14 0.22 and 0.45
90 0.19 022 and 0.45
120 0.19 0.22 and 0.45
498 4 Underpotential Deposition
0 2.00 4.00 µm
an unreconstructed Au(111) surface [39] STM-tip effect during the ruthenium de-
as was observed on a Pt(111) electrode position.
(cf. Figs. 16 and 17). The ruthenium is As described in the preceding text, al-
deposited in the entire Au double-layer though it became possible to grow electro-
potential region and a multilayer of ruthe- chemically the ultrathin ruthenium films,
nium can be deposited when the potential which can be applied in electrocatalysis,
becomes more negative than – 0.1 V (vs. the epitaxial growth of a metallic ruthe-
Ag/AgCl). nium film seems to be difficult at the
The similar random nucleation pro- present stage.
cess of ruthenium deposition on an It should be mentioned here that the
unreconstructed Au(111) electrode was composition of the electrolyte solution
observed by Uosaki and colleague in- used in these in situ STM studies of the
dependently [40]. Figure 18 shows STM electrochemical deposition of noble metals
images (4000 × 4000 nm2 ) at +300 mV (platinum, palladium, rhodium and ruthe-
nium) are quite different from those used
(vs. Ag/AgCl) in 0.05 M H2 SO4 solution
in the real electroplating industry [56–58].
containing 1.6 mM RuNO (NO3 )x (OH)y
Although some experimental conditions
(x + y = 3). It has been reported that the
(temperature, concentration) may be dif-
RuNO complex is more stable than that
ficult for the in situ STM measurements,
of the RuCl3 species in solution [104].
electrodeposition in a practical electrolyte
Before this image was obtained, STM
bath should provide more information
imaging was carried out at the nar-
both in application and in fundamental
row center portion of the image (shown fields.
by the dotted white box) for approxi-
mately 30 min, and no clear ruthenium- 4.1.3
nucleation island was observed. When Atomically Controlled Dissolution of Noble
the tip scan region was enlarged, how- Metals
ever, many ruthenium-nucleation islands
were randomly formed outside of the re- Knowledge about the metal-dissolut-
gion originally scanned, indicating a large ion mechanism is very important for
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 499
500
400
I-Pd(100)
300
100
Pd(100)
−100
controlling the metal surface structure. electrodes in sulfuric solution was cat-
Studies of anodic dissolution of metal alyzed by a chemisorbed iodine mono-
at atomic resolution have been reviewed layer for the first time [113–116]. The
elsewhere [23, 24, 105]. In this section, electrochemical dissolution processes of
the discussion will be concentrated on palladium modified by iodine have been
the anisotropic layer-by-layer dissolution investigated in detail by a combination of
behavior of noble metal electrodes, ultrahigh vacuum electrochemistry (UHV-
palladium and gold, catalyzed by an EC) and Auger as well as low-energy
adlayer of adsorbate on the substrate. electron diffraction (LEED) [117, 118].
Similar anisotropic anodic dissolution Figure 19 compares the CVs of a bare
processes catalyzed by the ordered Pd(100) and an I-modified Pd(100) elec-
adsorbed layer have also been observed trode in 0.05 M H2 SO4 solution [117]. A
on S-Ni(100), [106] I-Ni(111), [107] I- small anodic peak due to the oxidation
Ag(100) [108] and Cl-Cu(hkl) [109–112] as of the Pd(100) surface was observed at
well as Cl-Au(hkl) [29–31], which will be +0.90 V (vs. RHE) on the bare Pd(100)
discussed in the next section in detail. electrode (Reactions 1 and 2):
Pd + H2 O −−−→ Pd − OHs + H+ + e−
4.1.3.1 Palladium
(1)
Soriaga and colleagues reported that the
dissolution of palladium single-crystal Pd − OH(s) −−−→ PdO(s) + H+ + e− (2)
500 4 Underpotential Deposition
600
I-Pd (111)
500
I-Pd (100)
I-Pd (110)
400
[µA cm−2]
300
I
200
100
−100
0.5 1.0 1.5
E vs RHE
[V]
Fig. 20 CVs of I-modified Pd(111), Pd(100) and Pd(110) electrodes
in 0.05 M H2 SO4 solution [117].
This anodic peak was suppressed com- coverage remains constant throughout the
pletely on the I-modified Pd(100) electrode, dissolution process [117].
and a new anodic peak with an order-of- Figure 20 compares the structure depen-
magnitude larger peak current than that of dence of the I-catalyzed electrochemical
the I-free Pd(100) electrode was observed dissolution of Pd(hkl) electrodes in 0.05 M
at +1.25 V. This new anodic peak has been H2 SO4 solution. The dissolution rate
attributed to the two-electron dissolution slightly depends on the crystallographic
process of palladium as: orientation and decreases in the order of
I-Pd(110) > I-Pd(111) > I-Pd(100) [117].
Pd(s) −−−→ Pd(aq) 2+ + 2e− (3) Figure 21 shows a sequence of in situ
STM images of an I-modified Pd(100)
Therefore, the anodic dissolution of pal- electrode after the potential was stepped
ladium was enhanced significantly when to +1.05 V for (a) 5, (b) 6, and (c) 7 min,
the iodine adlayer was adsorbed on the where the palladium started to dissolved
palladium surface. The rate of the dissolu- anodically (Fig. 19) [117]. An atomically
tion is directly proportional to the coverage flat Pd(100) surface with a monoatomic
of iodine on the surface, and the iodine step was observed, and the I-catalyzed
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 501
T2
T3 T2 T3
[001]
[010]
T1
T1
(a) (b)
T2
0 40 80 nm
(c)
Fig. 21 STM images (80 × 80 nm2 ) of Pd(100)-c(2 × 2)-I obtained after (a) 5, (b) 6 and
(c) 7 min of dissolution at +1.05 V in 0.05 M H2 SO4 [117].
dissolution take places only at the step in the corner of the top terrace can be re-
edges in a layer-by-layer mode, while garded as the most reactive ones for the
etching from the pit formation was not dissolution. Removal of atom A (dissolu-
observed at the surface. As shown in tion along [001] direction) easily permits
Fig. 21, terraces T1 , T2 and T3 became the iodine atom to slide down from a 4-
narrow along [001]-directed steps but only fold site at the upper terrace onto another
a minimal change was observed in the hollow site (H) at the lower terrace. On
direction perpendicular to it, that is, an the other hand, iodine drop onto site H
anisotropic dissolution of palladium took cannot be accomplished upon removal of
place. Itaya and colleagues interpret the atom B (dissolution along [010] direction)
dissolution behavior using a schematic because of steric hindrance by atom A.
model (Fig. 22) where iodine formed a Therefore, iodine is easier to diffuse from
c(2 × 2) adlattice on the Pd(100) sur- a high-symmetry site from the upper to
face [117]. The palladium atoms A and B a lower terrace and is considered to be
502 4 Underpotential Deposition
4.1.3.2 Gold
Gold is known to have excellent chemical
stability and is widely used in industry
and as an accessory material [119, 120].
I on the upper terrace
It becomes unstable and dissolves in
[010]
[011] the positive-potential region, especially in
I on the lower terrace
solutions containing Cl− .
[001] Figure 23 shows CVs of (a) Au (111) and
Pd at upper terrace
(b) Au(100) electrodes in 0.1 M HClO4
Pd at lower terrace solution containing 1 mM Cl− in the
potential region between +0.2 and +1.7 V.
A significant increase in the anodic current
the major driving force in the anisotropic started at +1.25 V, and two anodic peaks
dissolution [117]. were found at +1.40 V and +1.51 V at
The step-selective layer-by-layer dissolu- Au(111) electrode. When the potential
tion behavior was also observed on an I- became more positive than +1.55 V, the
Pd(111) surface. However, the anisotropic- anodic current quickly decreased. The
dissolution features are not as obvious pronounced anodic current in the positive-
as that observed on an I-Pd(100) sur- going potential sweep was attributed to
face. The pit formation on the I-modified the 3e− oxidative dissolution of gold
Pd(110) precludes layer-by-layer dissolu- (Reaction 4, forward) [29].
tion and leads to progressive disorder [117,
118]. Au + 4Cl− ←−→ AuCl4 − + 3e− (4)
The ex situ LEED [113–116] and in + −
Au + H2 O ←−→ AuO + 2H + 2e (5)
situ STM measurements with high res-
olution [105, 117] demonstrated that the In the more positive potential region, the
ordered iodine adlayers were formed √ dissolution reaction competed with oxide
on the palladium surfaces as ( 3 ×
√ formation (Reaction 5, forward) on the
3R30◦ )-I-Pd(111), c(2 × 2)-I-Pd(100) and gold surface and passivation was com-
pseudohexagonal-I-Pd-(110). It is notewor- pleted around +1.7 V. The cathodic peaks
thy that the same adlattice structure of observed at +1.21 and +1.10 V in the
iodine was observed on the palladium ter- negative-going potential sweep were re-
race after anodic dissolution. The ordered lated to the simultaneous oxide reduction
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 503
10 µA cm−2 10 µA cm−2
200 200
Peak I Peak II Peak I Peak II
0 0
0.5 1.0
0.5 1.0
Current density
100
100
[µA cm−2]
0
0
−100
(a) (b)
Fig. 24 A sequence of in situ STM images (1 × 1 µm2 ) of the Au(111) surface in 0.1 M HClO4
containing 1 mM Cl− . (a) At +0.8 V, (b) +0.99 V ∼ +1.09 V during a potential
sweep (2 mV s−1 ) with rastering the tip downwards, (c) 0 min, (d) 6 min and (e) 40 min after
the potential was swept to +1.27 V, (f) potential was stepped to +1.36 V at the moment shown
by the arrow with rastering the tip downwards, (g) next scan after (f) with rastering the tip
upwards, (h) 3 min later from (f) with rastering tip downwards. Etip = +1.2 V. iT = 5 nA. A
compass was drawn to indicate the orientation of the surface. See the text for details [30].
take place in this potential region. The was etched mainly from its step edge
dissolution proceeded mainly at the step in prolonged etching at +1.27 V (6 and
sites, and the originally observed straight 40 min., Figs. 24d–e). At the beginning of
step lines were etched like the teeth the dissolution, only steps along the [110]
of a saw at more positive potential direction were observed (Figs. 24b,c). As
(+1.27 V, Fig. 24c). The dissolution also shown by the arrows in Fig. 24d, some
proceeded on the terrace (Fig. 24c) but of the step lines were rotated about
only from spots where small pits or 30 degrees, and new step lines running
defects were found at more negative along the [211] direction of the Au(111)
potentials (Figs. 24a,b) as indicated by substrate were observed. The step edges
the black circles. The Au(111) surface along the [211] direction, however, became
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 505
(e) (f)
[110]
[211]
Fig. 24 (Continued)
dominant after the electrode was kept at part of the image, rectangular-shaped pits
+1.27 V for a long time (Fig. 24e). were clearly observed. In the next STM
As the potential was stepped to a more scan (Figs. 24f,g), rectangular pits were
positive values (+1.36 V, Figs. 24f–h), clearly observed on all terraces. It is
where gold still dissolves without oxide interesting to note that the rectangular
formation, at the moment shown by an pits were parallel to each other within the
arrow in Fig. 24(f), acceleration of the same domain. Furthermore, the direction
dissolution rate, especially on the terraces, of the long axis of the rectangular pits
was observed. Many small pits were was parallel to the [211] direction of the
formed on the terraces. In the bottom substrate, that is, the new step direction
506 4 Underpotential Deposition
after dissolution (Fig. 24e). The pits grew for longer periods (Figs. 25e,f). These
and then merged. Figure 24(h) shows an results suggest that anodic dissolution of
STM image captured about 2 min after the Au(100) electrode surface is also an
Fig. 24(g). The top layer of the Au(111) anisotropic layer-by-layer etching process
was almost etched away, and the next in which the step edges preferentially
layer of the gold surface started to dissolve. retreated along the [100] direction of the
Nearly all of the step lines were orientated Au(100) substrate [31].
along the [211] direction, and the step As described in the preceding text,
line along the [110] direction was absent. the anisotropic-dissolution behavior was
These results demonstrated that gold was observed both on Au(111) and Au(100)
dissolved layer by layer in an anisotropic electrode surfaces only in acidic solu-
way in which the step edges preferentially tion containing Cl− . The structure of
retreated along the [211] direction of the the chloride adlayer on the gold surface
Au(111) substrate. is considered to be one of the impor-
STM observations were also carried out tant origins of the anisotropic-dissolution
in solutions of various Cl− concentrations. process. Suggs and Bard observed that cop-
It was found that the critical potential of the per dissolved in aqueous-chloride solution
dissolution depended on the Cl− concen- from the step edges respectively retreat-
tration. The higher the Cl− concentration, ing along the [211] and [100] directions on
the more negative the critical potential. Cu(111) [109] and Cu(100) electrode sur-
This potential dependence is consistent faces [110]. Vogt and coworkers observed
with that of Peak II. In solutions contain- a similar anisotropic dissolution on a
ing Cl− at less than 0.1 mM, the dissolu- Cu(100) electrode surface in HCl solution
tion rate was very low, and both the [211] and proposed that the preferential forma-
and [110] step lines were observed after the tion of the [100]-step edge resulted from
electrode was kept at +1.36 V for 20 min. its higher stability because the [100]-step
In a solution of higher Cl− concentration, edge of the Cu(100) was more stable than
that is, more than 10 mM, the anisotropic that of the [110]-step edges by 3 to 4 orders
characteristics were hard to observe be- of magnitude [111, 112].
cause of the very high dissolution rate. Figure 26 illustrates a real-space model
The anisotropic dissolution behavior was of the ordered chloride adlayer and the un-
also observed on the Au(100) electrode derlying Au(111) at the potential of Peak II,
in the similar solutions (Fig. 25). The based on the SXS measurements [121]. A
dissolution takes place only on the step chloride adlayer with a close-packed hexag-
edges in the potential region around onal structure is not commensurate with
Peak II (Fig. 25a). At the beginning of the gold substrate, and the interatomic
the dissolution, steps only along the [110] distance was close to the van der Waals
direction were observed (Figs. 25b,c). As diameter of chloride. The highest density
shown by the arrows in Fig. 25(d), many of the chloride adatom was found in the
step lines rotated 45 degrees from the step lines along the [211] direction, and
original [110] direction, that is, the [100] the adlayer was compressed more with
direction of the Au(100) substrate, after increasing potential. As the STM results
10 min etching at +1.30 V. Almost all of demonstrated in Fig. 24, the original step
the step lines were running along the [100] lines along the [110] direction disappeared
direction after being polarized at +1.30 V and new [211] step lines were observed
(a) (b) (c)
[110]
[100]
Fig. 25 A sequence of in situ STM images of the Au(100) surface in 0.1 M HClO4 containing 1 mM Cl− . (a) At +0.8 V (1 × 1 µm2 ), (b) At +1.1 V,
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals
(c) 0 min, (d) 10 min, (e) 20 min, and (f) 40 min after the potential was swept to +1.30 V (0.5 × 0.5 µm2 ). Etip = +1.40 V. iT = 3 nA. A compass is
drawn to indicate the orientation [31].
507
508 4 Underpotential Deposition
[211]
[110]
on the surface after the anodic dissolu- chloride adlayer structure has also been
tion. The lateral interactions between the observed using in situ STM by Kolb and
adsorbed chloride and gold were consid- colleagues [123]. The chloride adlayer with
ered to stabilize the atom rows along a higher Cl− adatom density on the step
the [211] direction more than those in edges of the [100] direction will make these
the other directions because of its highest step edges more stable than those of [110]
chloride-adatom density. In other words, a and other directions on the Au(100) elec-
much lower etching rate on the [211] step trode surface in the positive potential
lines was expected, and as a result, the region. Because of the stability, the [110]-
anisotropic dissolution of Au(111) in the step lines, which were originally observed
Cl− containing solution was observed. before dissolution, disappeared and only
In the case of an Au(100) electrode, the [100]-step lines can be observed after
Wang and coworkers observed using SXS the anodic dissolution.
measurements that chloride√formed uni-
axially incommensurate c( 2 × p)R45◦ 4.1.4
adlayer structure on the Au(100) elec- Summary
trode surface and was compressed along
the [100] direction when the potential Electrochemical deposition of noble metal
became more positive [122]. A similar elements (platinum, palladium, rhodium,
4.1 Atomically Controlled Electrochemical Deposition and Dissolution of Noble Metals 509
and ruthenium) on substrates (mainly gold and Technology Foundation, the Japan Se-
and platinum single-crystal electrode sub- curities Scholarship Foundation and the
strates) has been investigated by many Tokuyama Science and Technology Foun-
modern instrumental analysis methods, dation.
such as STM, EQCM, and SXS measure-
ments at an atomic level. As summarized References
in Table 1, electrochemical deposition of
palladium on a gold substrate takes places 1. F. R. Hartley, (Ed.), Chemistry of the Plat-
inum Group Metals, Elsevier Scientific Pub-
in the FM mode, similar to that of silver.
lishers B. V., Amsterdam, 1991.
Platinum and rhodium grow electrochem- 2. A. Cowley, M. Steel, Platinum 2001, John-
ically on a gold substrate in the SK mode. son Matthey, London, 2001.
The electrochemical growth of ruthenium 3. G. A. Somorjai, Surface Chemistry and Catal-
on a gold substrate prefers the SK mode, ysis, John Wiley & Sons, New York, 1994.
4. C. Argile, G. E. Rhead, Surf. Sci. Rep. 1989,
and a metallic-ruthenium film is still diffi- 10, 277–356.
cult to grow at the moment. These types of 5. J. A. Rodriguez, Surf. Sci. Rep. 1996, 24,
behavior can be understood qualitatively 223–287.
using the mismatch parameter of lattice 6. H. Brune, Surf. Sci. Rep. 1998, 31, 121–229.
7. T. T. Kodas, M. J. Hampden-Smith, The
and surface energy between the deposited Chemistry of Metal CVD, VCH, Weinheim,
metal and the gold substrate. 1994.
The anistropic layer-by-layer dissolution 8. M. K. Hampden-Smith, T. T. Kodas, Chem.
of palladium and gold are enhanced by Vapor Deposition 1995, 1, 8–23.
9. J. Clavilier, R. Faure, G. Guinet et al.,
an adlayer of iodine and chloride on the
J. Electroanal. Chem. 1980, 107, 205.
substrate surfaces. 10. G. Binnig, H. Rohrer, C. Gerber et al., Appl.
The present information is very useful Phys. Lett. 1982, 40, 178.
for creating electrocatalytic materials with 11. G. Binning, C. F. Quate, C. Gerber, Phys.
high activity and selectivity. This is also Rev. Lett. 1986, 12, 930.
12. O. M. Magnussen, J. Hagebock, J. Hotlos
very important for understanding the et al., Faraday Discuss. 1992, 94, 329–338.
reaction mechanism of the metal epitaxy in 13. A. Cuesta, M. Kleinert, D. M. Kolb, Phys.
the field of surface electrochemistry. Very Chem. Chem. Phys. (PCCP) 2000, 2,
rapid progress is expected in this field. 5684–5690.
14. A. M. Funtikov, U. Stimming, R. Vogel,
Surf. Sci. 1995, 324, L343–L348.
Acknowledgment 15. A. M. Funtikov, U. Stimming, R. Vogel,
J. Electroanal. Chem. 1997, 428, 147–153.
16. L. J. Wan, T. Suzuki, K. Sashikata et al.,
This work was partially supported by J. Electroanal. Chem. 2000, 484, 189–193.
Grants-in-Aid for Scientific Research on 17. L. J. Wan, S. L. Yau, K. Itaya, J. Phys. Chem.
Priority Area of ‘‘Electrochemistry of Or- B 1995, 99, 9507–9513.
dered Interfaces’’ (No. 09237101) and for 18. L. J. Wan, M. Hara, J. Inukai et al., J. Phys.
Chem. B 1999, 103, 6978–6983.
Encouragement of Young Scientists (No. 19. M. F. Toney, J. McBreen, Interface 1993, 1,
12750723, 10740314) from the Ministry 22–31.
of Education, Science, Sports and Cul- 20. J. X. Wang, R. R. Adzic, B. M. Ocko in Inter-
ture, Japan. SY acknowledges the support facial Electrochemistry (Ed.: A. Wieckowski),
Marcel Dekker, New York, 1999,
from PRESTO, Japan Science and Tech- pp. 175–186.
nology Corporation (JST). SY acknowl- 21. D. A. Buttry, M. D. Ward, Chem. Rev. 1992,
edges grants from the Iketani Science 92, 1355–1379.
510 4 Underpotential Deposition
66. W. Haiss, J. K. Sass, J. Electroanal. Chem. 90. G. A. Attard, R. Price, A. Al-Akl, Surf. Sci.
1997, 431, 15–17. 1995, 335, 52–62.
67. T. Katsuzaki, K. Uosaki, unpublished. 91. F. H. Reid, J. C. Blake, Trans. Inst. Metal
68. K. A. Friedrich, A. Marmann, U. Stimming Finishing 1961, 38, 45–51.
et al., Fresenius’ J. Anal. Chem. 1997, 358, 92. J. F. Llopis, I. M. Tordesillas in Encyclopedia
163–165. of Electrochemistry of the Elements (Ed.:
69. E. G. Seebauera, C. E. Allen, Prog. Surf. Sci. A. J. Bard), Marcel Dekker, New York, 1976,
1995, 49, 265–330. pp. 277–298, Vol. VI.
70. N. Kimizuka, T. Abe, K. Itaya, Denki Kagaku 93. K. R. Seddon, Platinum Met. Rev. 1996, 40,
1993, 61, 796–799. 128–134.
71. S.-L. Yau, T. Moriyama, H. Uchida, M. Wat- 94. F. P. Gortsema, J. W. Cobble, J. Am. Chem.
anabe, Chem. Commum. 2000, 2279–2280. Soc. 1961, 83, 4317–4321.
72. J. F. Llopis, F. Colom in Encyclopedia 95. K. A. Friedrich, K.-P. Geyzers, U. Linkeb
of Electrochemistry of Elements (Ed.: et al., J. Electroanal. Chem. 1996, 402,
A. J. Bard), Marcel Dekker, New York, 1976, 123–128.
pp. 253–275, Vol. VI. 96. S. Cramm, K. A. Friedrich, K.-P. Geyzers
73. A. Cuesta, L. A. Kibler, D. M. Kolb, J. Elec- et al., Fresenius’ J. Anal. Chem. 1997, 358,
troanal. Chem. 1999, 466, 165–168. 189–192.
74. G. A. Attard, A. Bannister, J. Electroanal. 97. K. A. Friedrich, K. P. Geyzers, A. Marmann
Chem. 1991, 300, 467–485. et al., Zeitschrift fur Physikalische Chemie
75. M. J. Llorca, J. M. Feliu, A. Aldaz et al., 1999, 208, 137–150.
J. Electroanal. Chem. 1993, 351, 299–319. 98. W. Chrzanowski, A. Wieckowski, Langmuir
76. M. J. Llorca, J. M. Feliu, A. Aldaz et al.,
1997, 13, 5974–5978.
J. Electroanal. Chem. 1994, 376, 151–160.
99. W. Chrzanowski, A. Wieckowski in Interfa-
77. M. Baldauf, D. M. Kolb, Electrochim. Acta
cial Electrochemistry (Ed.: A. Wieckowski),
1993, 38, 2145–2153.
Marcel Dekker, New York, 1999,
78. H. Naohara, S. Ye, K. Uosaki, Colloids Surf.,
pp. 937–954.
A 1999, 154, 201–208.
100. E. Herrero, J. M. Feliu, A. Wieckowski,
79. H. Naohara, S. Ye, K. Uosaki, Electrochim.
Langmuir 1999, 15, 4944–4948.
Acta 2000, 45, 3305–3309.
101. H. Kim, I. R. de Moraes, G. Tremiliosi
80. H. Naohara, S. Ye, K. Uosaki, J. Electroanal.
Chem. 2001, 500, 435–445. et al., Surf. Sci. 2001, 474, L203–L212.
81. A. S. Dakkouri, D. M. Kolb in Interfacial 102. W. F. Lin, M. S. Zei, M. Eiswirth et al., J.
Electrochemistry (Ed.: A. Wieckowski), Mar- Phys. Chem. B 1999, 103, 6968–6977.
cel Dekker, New York, 1999, pp. 151–173. 103. T. Iwasita, H. Hoster, A. John-Anacker
82. M. Takahashi, Y. Hayashi, J. Mizuki et al., et al., Langmuir 2000, 16, 522–529.
Surf. Sci. 2000, 461, 213–218. 104. G. Tremiliosi-Filho, H. Kim, W. Chrzan-
83. T. Kondo, K. Tamura, M. Takahashi, owski et al., J. Electroanal. Chem. 1999, 467,
J. Mizuki, K. Vosaki, Electrochim. Acta, in 143–156.
press (2002). 105. K. Itaya in Interfacial Electrochemistry (Ed.:
84. M. Baldauf, D. M. Kolb, J. Phys. Chem. 1996, A. Wieckowski), Marcel Dekker, New York,
100, 11 375–11 381. 1999, pp. 187–210.
85. J. F. Llopis, I. M. Tordesillas in Encyclopedia 106. T. Suzuki, T. Yamada, K. Itaya, J. Phys.
of Electrochemistry of the Elements (Ed.: Chem. 1996, 100, 8954–8961.
A. J. Bard), Marcel Dekker, New York, 1976, 107. P. Muller, S. Ando, T. Yamada et al.,
pp. 299–322, Vol. VI. J. Electroanal. Chem. 1999, 467, 282–290.
86. J. Inukai, M. Ito, J. Electroanal. Chem. 1993, 108. T. Teshima, K. Ogaki, K. Itaya, J. Phys.
358, 307–315. Chem. 1997, 101, 2046–2053.
87. R. Gomez, A. Rodes, J. M. Perez et al., Surf. 109. D. W. Suggs, A. J. Bard, J. Am. Chem. Soc.
Sci. 1995, 327, 202–215. 1994, 116, 10 725.
88. R. Gomez, A. Rodes, J. M. Perez et al., Surf. 110. D. W. Suggs, A. J. Bard, J. Phys. Chem.
Sci. 1995, 344, 85–97. 1995, 99, 8349.
89. R. Gomez, J. M. Feliu, Electrochim. Acta 111. O. M. Magnussen, M. R. Vogt, F. A. Moller
1998, 44, 1191–1205. et al., Surf. Sci. 1996, 367, L33–L41.
512 4 Underpotential Deposition
aluminum anodization), but will not be referred to as UPD [34–39]. In the deposi-
discussed here. The electrodeposition of tion of one element on a second, frequently
II to VI compounds has been extensively the first element will form an atomic layer
studied and is well reviewed in a number of at a potential under, or prior to, that needed
articles [18–23]. The most prominent com- to deposit the element on itself. One way
pound electrodeposition methods include of looking at UPD is that a surface com-
codeposition, precipitation, and various pound, or alloy, is formed, and the shift
two-stage techniques. in potential results from the favorable free
The focus of the work described here energy of formation of the surface com-
is on understanding the mechanisms of pound.
compound electrodeposition and how to EC-ALE is the combination of UPD
control structure, morphology, and com- and ALE. Atomic layers of a compound’s
position. The primary tool for understand- component elements are deposited at
ing compound electrodeposition and for underpotentials in a cycle, to directly
improving control over the process has form a compound. It is generally a
been the methodology of EC-ALE [23–26]. more complex procedure than most of
Historically, EC-ALE has been developed the compound electrodeposition methods
by analogy with ALE [27–33]. ALE is a described earlier, requiring a cycle to form
methodology used initially to improve each ML of the compound. However,
epitaxy in the growth of thin films by it is layer-by-layer growth, avoiding 3D
MBE and VPE. The principle of ALE is nucleation, and offering increased degrees
to use surface-limited reactions to form of freedom, atomic level control, and
each atomic layer of a deposit. If no more promoting of epitaxy.
than an atomic layer is ever deposited, In addition, EC-ALE offers a way of better
the growth will be 2D, layer-by-layer, understanding compound electrodeposi-
and epitaxial. Surface-limited reactions tion, essentially a way of breaking it down
are developed for deposition of each into its component pieces. It allows com-
component element, and a cycle is formed pound electrodeposition to be deconvolved
with them. With each cycle, a compound into a series of individually controllable
ML is formed, and the deposit thickness is steps, resulting in an opportunity to learn
controlled by the number of cycles. more about the mechanisms, and gain
As noted, in techniques such as MBE a series of new control points for elec-
and VPE, surface-limited reactions are trodeposition. The main problem with
generally controlled by the temperatures compound electrodeposition by codeposi-
of the reactants and substrate. The tem- tion is that the only control points are the
perature is kept high enough so that solution composition and the deposition
deposition over a ML sublimes, leaving potential, or current density, in most cases.
only the atomic layer, forming the com- In an EC-ALE process, each reactant has its
pound. Problems are encountered when own solution and deposition potential, and
the temperatures needed to form atomic there are generally rinse solutions as well.
layers of different elements are not the Each solution can be separately optimized,
same, as changing the temperature be- so that the pH, electrolyte, and additives
tween layers is difficult. or complexing agents are tailored to fit
Surface-limited reactions are well known the precursor. On the other hand, code-
in electrochemistry, and are generally position uses only one solution, which is
516 4 Underpotential Deposition
Waste
Valve Block
Computer
control
Deposition
solution
Pump
Parmer). The main requirement for the coworkers have used pressurized bottles,
pumps is that they are clean. If smoother without pumps, to deliver solution [44, 45].
pumping is required, pulse dampening or In Fig. 2, there is one pump for each line,
syringe pumps could be used. Foresti and to push solution through the cell. Villegas
(a)
Fig. 3 Diagrams of electrochemical cells used in flow systems for
thin-film deposition by EC-ALE. (a) First small thin-layer flow cell
(modeled after electrochemical liquid chromatography detectors). A
gasket defined the area where the deposition was performed, and
solutions were pumped in and out through the top plate. (Reproduced
by permission from Ref. [41].) (b) H-cell design where the samples were
suspended in the solutions, and solutions were filled and drained from
below. (Reproduced by permission from Ref. [42].) (c) Larger thin-layer
flow cell. This is very similar to that shown in (a), except that the
deposition area is larger and laminar flow is easier to develop because
of the solution-inlet designs. In addition, the opposite wall of the cell is
a piece of ITO, used as the auxiliary electrode. It is transparent, so the
deposit can be monitored visually, and it provides excellent current
distribution. The reference electrode is incorporated into the cell as well.
(Adapted from Ref. [40].)
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 519
(b)
Teflon nut
Gasket
Inlet Outlet
Fig. 3 (Continued)
and coworkers used a single pump on the sufficient to prevent problems with oxy-
outlet to suck each solution into the cell, gen, as most tubing has some oxygen
in an elegant simplification [43]. permeability. To better avoid this problem,
There are a number of vendors that sell the solution delivery tubes were threaded
solenoid-actuated Teflon valves, which are through larger ID tubes and into the Plexi-
easily interfaced to a computer. Care must glas box (Fig. 2) that houses the pumps and
be taken to choose a design in which the valves. The sparging N2 was made to flow
internal volume at the valve outlet can be out of the solution reservoirs, through the
flushed easily between steps, however [41]. large ID tubes (around the outside of the
Rotary selection valves have been used as solution-delivery tubes) and into the box,
well, but given the number of rotations greatly decreasing oxygen exposure. The
needed for a 200-cycle deposit, various measured oxygen content of the N2 leav-
failure modes revealed themselves. ing the Plexiglas box was 10 to 30 ppm, as
As deposition of most of the relevant measured with a glove box oxygen analyzer
atomic layers involves reduction at rela- (Illinois Instruments, model 2550).
tively low potentials, oxygen has proven to The majority of deposits formed in this
be a major problem. It has been shown group have been on Au electrodes, as they
repeatedly that oxygen not rigorously ex- are robust, easy to clean, have a well-
cluded results in thinner deposits, if they characterized electrochemical behavior,
are formed at all. For this reason, exten- and reasonable quality films can be formed
sive sparging of the solution reservoirs is by a number of methodologies. However,
critical. Sparging alone is generally not Au it is not well lattice-matched to
520 4 Underpotential Deposition
5 nm
200 300
200
100
100
0 0
0 100 200 0 100 200 300
nM nM
Z range: 15 nm Z range: 2 nm
Fig. 5 (a) AFM image of Au vapor deposited on Si(100) at room temperature and (b) AFM image of
Au on glass, annealed at 550 C for 12 h, in a tube furnace, in a flow of N2 .
given a brief flame anneal, in the dark used as a substrate. The problems in-
with a H2 flame to a dull orange glow, volve adequately preparing the substrate
prior to use. surfaces and understanding the electro-
Some Cu substrates have been used, chemistry of a compound semiconductor
including Cu foils, etched foils, and vapor- substrate. Work is progressing in this di-
deposited Cu on glass. There does not rection. Good quality deposits of CdSe
appear to be a significant difference in the have been formed on InP and GaAs
quality of deposits formed on Cu versus substrates using codeposition, by Mau-
Au, beyond that expected to result from rin and coworkers [55, 56]. They used
considerations of lattice matching. reflection high-energy electron diffraction
Single-crystal silver substrates have been (RHEED) and TEM to follow the habit of
used exclusively by Foresti and cowork- CdSe growth. Their work clearly shows
ers [44, 45, 52, 53]. They use macroscopic the applicability of high-quality com-
Ag single crystals, formed in-house. They mercial compoundsemiconductor wafers
have formed a number of II to IV com- as substrates for compound electro-
pounds using EC-ALE, including ZnSe, deposition.
CdS, and ZnS.
Semiconductors such as polycrystalline 4.2.3
ITO on glass have been used to form Deposition Programs
deposits of ZnS [54], CdS, and CdTe,
by this group, with no obvious prob- 4.2.3.1 Cycle Steps
lems. Ideally, lattice-matched semicon- As described in the introduction, UPD is
ductor substrates could also be used the formation of an atomic layer of one
to form deposit. For instance, InSb is element on a second element at a potential
lattice-matched with CdTe, and could be under that required to deposit the element
522 4 Underpotential Deposition
0.0
[µA]
Cd UPD
−20.0
Alloy dep
(c) (b)
3′ Te
Te pH 2.2
pH 9.2
1′ 1′
80 3′
2′
Current
2′
[µA]
0
1 0
2 2
1
3
−80
3
Bulk Se removal
Se UPD removal
10 µA
Further bulk Se Se UPD
Limited bulk Se
20 µA
Fig. 6 (Continued)
stabilization by bonding with previously Equations (4) and (5) together have essen-
deposited Cd, forming CdTe: tially the same result as Eq. (3), oxidative
UPD or formation of a Te atomic layer,
Teo + Te(UPD) + 2H+ + 2e− ←−→
along with conversion of some tellurite
Te(UPD) + H2 Te (5) to telluride. The advantage is that F and G
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 525
can be accomplished using the more stable Reactants can also be generated electro-
tellurite solutions, instead of the unstable chemically, prior to pumping the solution
telluride. into the cell, or generator electrodes can be
incorporated into the cell, so a precursor
could be generated on the opposite wall of
4.2.3.2 Cycles
the cell, and allowed to diffuse across the
A cycle consists of the steps needed to
thin solution layer to deposit. This method
form a ML of the compound. Figure 7 is a
would have advantages in that species with
cycle diagram for CdTe and describes the
limited stability could be used, as they need
deposition, stripping, and rinsing steps, as
only last long enough to diffuse across, and
well as times and potentials.
only enough would be generated to form
In principle, each solution can be inde-
the atomic layer. This has been suggested
pendent, composed of different solvents, as a way of obtaining H3 As, H2 Te, and so
reactants, electrolytes, buffers, and addi- on as precursors for oxidative UPD.
tives. In general, aqueous solutions have For some elements, there are a variety
been used, as their electrochemical be- of possible precursors, sulfur for instance
havior is better understood, and ultrapure (Fig. 8). Metal-organic precursors, such as
water can be obtained by a number of used in metallo organic molecular beam
methods. epitaxy (MOMBE) or metallo organic va-
The choice of reactant has to do with por phase epitaxy (MOVPE), are possible
whether you want to perform oxidative or precursors that can be used, if they can be
reductive UPD, oxidation state, solubility, made soluble in water or if a nonaqueous
availability, purity, and price. In general, solvent were used. Mixed aqueous-organic
salts are used, simplifying solution prepa- solvents could improve solubility. Overall,
ration. Gases can be used by saturating the there would be an increased probability of
solution, prior to pumping it into the cell. carbon contamination. One of the benefits
.
cs .
se cs
2 se
V, 2
.7
0 V,
−0.8 0
−0 0.
7
@ − .
fill @ cs cs
.
n e se se
tio ns 2
Potential (vs Ag/AgCl)
lu ri V, 2
so nk 60 V,
−0.7 a 0. 0
Te Bl − .6
@ −0
fill @
n se
Te deposition utio rin
20 secs. @ −0.70 V s ol nk
−0.6 a
quiescent Cd Bl
Cd deposition
20 secs. @ −0.60 V
quiescent
−0.5
0 10 20 30 40 50 60
Time
[s]
Fig. 7 Diagram of an EC-ALE cycle for the formation of CdTe. (Adapted from Ref. [60].)
526 4 Underpotential Deposition
2.5
2
Coverage
[ML]
Sulfide
1.5
Thiosulfate
1
0.5
0
−1.5 −1 −0.5 0 0.5
Potential vs Ag/AgCl
[V]
Fig. 8 Graphs of the Auger signals resulting from exposure to various
sulfur-containing precursors as a function of potentials. (Adapted from Ref. [61].)
using a rinse solution at all. That is, the be used. In general, the same elec-
reactant solutions were exchanged by each trolyte concentrations have been used
other, underpotential control, suggesting for rinsing and deposition solutions
that some small amount of codeposition so far.
probably did occur. The starting potentials for most atomic
The amount of electrolyte needed in a layers, in this group, were obtained by
rinse solution depends on current flow studies of the voltammetry for an element
during the rinse. Rinsing can be per- on an Au electrode [57, 66–68], usually
formed at open circuit in some cases, using a thin-layer electrochemical cell
so that no electrolyte is needed. If the (TLEC) (Fig. 9) [69, 70]. UPD potentials
amount of current during the rinse on Au are not expected to be optimal for
under controlled potential is very low, growth of a compound; however, they are
low-electrolyte concentrations could again generally a good start.
Solution
gap
0.001′
Electrode
+
+ +
+
+
+
+
+ + + +
+ + + + Pin hole
+ + + solution
+ + + Thin layer
+ + in and out
+ electrode
10 mm ID
teflon
thermometer
adaptor
1 mm capillary
tubing
24 / 40
joint
Male
ER Varibor Reference
teflon electrode
stopcock compartment
2 mm
10 mm
fine
frit
The simplest model for EC-ALE is that a Reasonable quality deposits have been
set of conditions is chosen for a cycle, and obtained for 200-cycle runs, using these
each cycle produces one compound ML. more negative ‘‘steady state’’ potentials
Ideally, the same potentials and solutions from the beginning. However, that more
are used for each cycle of a deposition. than an ML/cycle is formed during the
Recently, it has become clear that this is first few cycles is a significant problem, as
not the optimum case for EC-ALE growth it suggests bulk deposition, non–layer-by-
of many compounds on Au, as the initial layer, or 3D growth. That high growth rates
conditions do not appear optimal from were observed initially is understandable,
start to finish. in that the potentials were close to bulk
Graphs of the charges for Cd and Te growth conditions for Cd and Te.
versus the cycle # (Fig. 10) show that if a Using these more negative potentials
single set of potentials is used for each did result in reasonable deposits. Overall
cycle, the first cycles are much larger they were stoichiometric, displayed the
than the succeeding cycles. If potentials expected X-ray diffraction (XRD) patterns
determined from using a TLEC are used, without annealing, and the graphs of
the first cycle looks good, the charges growth versus the number of cycles
correspond to the formation of an ML were linear (after the first few). Deposit
of the compound, but the charges quickly morphologies appeared reasonable but
die away to nothing. On the other hand, probably not optimal, given some 3D
if potentials are used that will result in growth during the first few cycles.
something close to an ML of the compound It has become clear that the potentials
for each cycle once steady state has been needed to form atomic layers shift neg-
reached, the first few cycles will result in a atively as the semiconductor films grow,
great excess of deposition (Fig. 10). especially over the first 25 cycles. The
4.00 To
14.5 ML
3.50 Monolayers Te deposited
Monolayers deposited
Monolayers Cd deposited
3.00 To
9.5 ML
2.50
2.00
1.50
1.00
0.50
0.00
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7
Cycle number
Fig. 10 Bar graphs showing the charges, in terms of monolayers, where 1 ML would
be optimum, as a function of the cycle, for the first 7 cycles of CdTe deposition. The
first half cycle was Te deposition, and the second Cd. A single set of potentials was
used for all the cycles, and during the first cycle, the amounts deposited were excessive.
(Adapted from Ref. [60].)
530 4 Underpotential Deposition
most logical reason for this shift is that of deposition agreed closely with the de-
a junction potential is developing between posited amounts, suggesting that proton
the Au substrate and the depositing com- and oxygen reduction were not significant.
pound semiconductor. When a program Finally, no elemental Cd and Te were ob-
is used where the potentials are initially served in XRD patterns; only diffraction
shifted for each cycle, steady state poten- peaks for CdTe and the substrate, for 200-
tials are generally achieved after about 25 cycle deposits, were observed.
cycles, which can be maintained through One explanation for the observed charge
the rest of the deposit without the de- disparity is that excess Cd is being
posited amounts changing for each cycle. deposited, and when the TeO3 2− is
Similar procedures were used by Ville- introduced, some Te is exchanged for Cd.
gas and coworkers [43] in the formation The TeO3 2− may react chemically with the
of CdTe. A reasonable potential ramp can cadmium, or the current observed may be
be determined by inspection, and the po- the sum of that for oxidation of part of the
tential changes programmed through the previously deposited excess Cd and that for
first 25 cycles or so. Some form of feed- reductive deposition of the Te atomic layer.
back within the cycle would be better, to In general, the currents observed during a
regulate deposition on the fly. The most cycle in the formation of a compound have
not been easy to understand, and they
obvious feedback would be the deposition
are presently a major topic of study. For
currents (or charges).
instance, in the formation of CdSe, a cycle
Figure 10 displays the coverages for Cd
similar to that for CdTe is used; however,
and Te deposited in each of the first few
it is the charge during Se deposition that is
CdTe cycles, using a program in which
in excess and that for Cd, which is almost
the potentials remained constant. The cov-
nonexistent. Again, the deposit is nearly
erages are the result of coulometry, two
stoichiometric, within a percent of 1 : 1,
electrons for Cd and four electrons for Te.
using EPMA. In summary, the relative
Note the excessive deposition for the first charges for Cd and Te deposition are
few cycles, significantly more than a ML, not easily interpreted in terms of deposit
and the disparity between Cd and Te cover- stoichiometry. They are, however, a key to
ages. The relative coverages for Cd and Te incorporation of feedback in the EC-ALE
should be 1 : 1 in Fig. 10, but seldom are cycle.
in such graphs. It is not yet clear why the Ideally, each component element can be
stoichiometry is not 1 : 1 from coulometry, deposited at a potential optimized sepa-
while the stoichiometries from EPMA for rately, independent of the conditions used
the resulting deposits are 1 : 1. Some pos- to deposit subsequent atomic layers in the
sible reasons for this discrepancy (Fig. 10) cycle. In some cases this independence
that have been considered include solvent is nearly realized, as in the deposition of
or proton reduction, oxygen reduction, and Te atomic layers (Fig. 11) discussed pre-
charging currents. However, blank solu- viously. A broad potential region (0.7 V)
tions provide a gauge of proton, solvent, is evident in Fig. 11, where the thick-
and oxygen reduction, as well as charg- nesses of 200-cycle CdTe deposits do not
ing currents, and they do not account change appreciably as the Te deposition
directly for the asymmetry. In addition, potential is adjusted. However, Te depo-
coulometric stripping after 5 to 10 cycles sition is a relatively irreversible process
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 531
240
220
210
200
180
160
Thickness
140
[nm]
120
100
80 78
68 70 68
60 62 58 57
40
20
0 2 3
−0.9 −0.8 −0.7 −0.6 −0.5 −0.4 −0.3 −0.2 −0.1 0 0.1 0.2
Te deposition potential (vs Ag/AgCl)
Fig. 11 Graph of CdTe deposit thickness as a function of the potential used to deposit Te. Each
point represents a deposit formed with 200 cycles. (Adapted from Ref. [60].)
at pH 9 (Fig. 6c), suggesting that once it formed by stripping should still be close
is deposited, the potential can be shifted to the deposit surface, in the barrier layer,
significantly more positively (Fig. 6c) with- where they may redeposit as the second
out its stripping from the surface. On the element deposits. On the other hand, given
other hand, a more reversible species, such that the second element is reversible, the
as Cd, may strip from the surface if the potential can be stepped positively after its
potential is shifted positively in a subse- introduction to the cell. Even if some bulk
quent cycle step. This is exaggerated after deposit is formed at the more negative
a rinse, where the activity for Cd2+ has potential, it should strip reversibly when
been removed. the potential is stepped positively.
In a system where both elements The inverse process, in which the
deposit reversibly, at significantly different potential is shifted negatively, is not
potentials, the rinsing procedures must generally a problem. The potential cannot
be carefully thought through. If, for be shifted prior to rinsing, or else bulk
instance, a reversible species is reductively deposition of the first element would occur
deposited, and a more positive potential before the solution is rinsed from the cell.
is to be used to deposit the next element, It is best to shift the potential after the
the rinse step can be performed at the blank rinse, just prior to introduction of
more negative potential, and then shifted the second element. In fact, the potential
positively upon introduction of the second can be shifted after introduction of the
reactant, hopefully avoiding material loss second element, as it should not deposit at
during rinsing and minimizing losses the more positive potential, but must wait
during introduction. If the second element until the potential is shifted negatively.
is pumped in rapidly, and with a minimum Keeping the potentials close can be bene-
of solution, ions of the first element ficial, using complexing agents or the pH.
532 4 Underpotential Deposition
on flow rate is underway. Most of the include the chalcogenides: S, Se, and Te;
compounds, described in the next section, the pnictides: As and Sb; the group III
form homogeneously in the deposit area. metals: Ga and In; the group II metals:
CdTe, on the other hand, does not deposit Zn, Cd, and Hg; as well as Cu and Co. The
well at the inlet and outlet, suggesting range of compounds accessible by EC-ALE
turbulence may affect the deposit. Slower is not clear, although the majority of work
pumps appear to help, but a detailed study has been performed on II to VI compounds
is needed to determine what step in the (Table 1). The III to V compounds InAs
cycle is causing the problem, and why. and InSb have recently been formed [40,
Besides the flow rate, rinse times are 62]. In addition, Shannon and coworkers
very important. As discussed previously, have begun studies of CoSb3 [73] with the
if rinsing is insufficient to achieve the intent of forming thermoelectric materials.
desired pH, deleterious effects can result. Initially, EC-ALE was developed on the
The example was given that if Te solutions principle that reductive UPD of a metal and
are too acidic, bulk Te will form. oxidative UPD of a main group element
Generally, this group has used stop flow were required to form a working cycle. This
to provide adequate deposition times and would then limit compounds that could be
to limit the volumes of solution needed. formed to those containing a chalcogenide
Villegas and coworkers have used con- or a pnictide, as reduced forms of some
tinuous pumping, with good results [43], of these elements were reasonably stable
which may be an important alternative. in aqueous solutions. Recently, it has
Which element should be deposited first been shown that reductive UPD of both
is an ongoing question. In the forma- elements is possible [40], suggesting a
tion of II to VI compounds by this group, much larger range of compounds or
the chalcogenides have generally been de- possibly alloys might be formed.
posited first. The reasons for this have Table 1 is a listing of compounds formed
more to do with the rich surface chemistry using EC-ALE. The first EC-ALE studies
of those atomic layers than a feeling that focused on CdTe [24, 25] for the historical
better deposits will result. Results from reason that its electrodeposition had been
two groups using Raman spectroscopy, studied the most [98–109]. The genesis
Shannon and coworkers [71] and Weaver for most of that work has been the desire
and coworkers [72], have suggested that to electrodeposit CdTe photovoltaics. The
higher-quality deposits of CdS are formed majority of the CdTe work is codeposition,
when the Cd is deposited first. Recent re- after the classic paper by Kroger and
sults by this group [63] indicate that no coworkers [99].
mater which is deposited first, Cd ends The first studies of EC-ALE were stimu-
up being the atomic layer next to the Au lated by Mike Norton [66], who suggested
substrate, suggesting it is probably better that an electrochemical form of ALE might
to deposit the Cd first. be possible, which prompted studies of
the UPD of Cd and Te on Cu, Au, and
4.2.4 Pt substrates, by this group. Those studies
Compound Formation involved the use of a TLEC [69, 70] (Fig. 9),
which consisted of a polycrystalline rod
At present, the elements used in the of the substrate metal, inserted into a
formation of compounds by EC-ALE vacuum-shrunk glass compartment. The
534 4 Underpotential Deposition
Tab. 1 (continued)
Note: STM: scanning tunneling microscopy; EC-ALE: electrochemical atomic layer epitaxy; UHV-EC:
ultrahigh vacuum electrochemistry; SERS: surface-enhanced Raman scattering; RRDE: rotating
ring-disk electrode.
compartment was designed to hold the the deposit to air. The TLEC is used to
electrode in the center of a roughly cylin- determine the potentials needed to form
drical glass enclosure, about 25 µm away atomic layers of the elements on the sub-
from the glass. Two pinholes were ground strate.
into the bottom of the cell, to allow so- Following the codeposition work of
lutions to flow in and out, and for ionic Kroger and coworkers [99] and many
conductivity. The total volume of the cell others, acidic solutions of HTeO2 + were
was 3.0 µL, with a 1-cm2 electrode area. initially used [24, 25]. At that time, it was
The TLEC provided a defined envi- assumed that if Cd was deposited by
ronment for studies of surface-limited reductive UPD, Te should be deposited
reactions. The large surface area to volume using oxidative UPD, from a solution
ratio greatly limits background reactions of H2 Te or a related telluride species.
from traces of oxygen and other contami- It was not felt that reductive UPD
nants. Limited and predictable background of both elements could be performed.
currents facilitate coulometry and allow Solutions of telluride proved unstable,
for more detailed coverage measurements. however, oxidizing and turning purple
The reactant amounts are also limited, with flecks of elemental Te, with even
allowing studies with only slightly more traces of oxygen in solution. This led
than that needed to form atomic layers, to the use of the process described
limiting bulk-deposition and preventing by Eqs. (4 and 5) [24, 25], where more
bulk-stripping currents from swamping than an atomic layer of Te was first
out UPD stripping. In addition, the TLEC deposited, from a HTeO2 + solution, and
facilitated solution exchange, which is very the excess Te was stripped reductively
important for studies of EC-ALE. One solu- at a more negative potential using a
tion can be flushed out and another rinsed blank solution in a separate step. The
in within a few seconds, without exposing only analysis performed on those initial
536 4 Underpotential Deposition
deposits was coulometry, as the substrates the bubbles floated to the top. Sub-
were not easily removed from the Pyrex strates used in that study were Au on
cells (Fig. 9), and the substrate’s cylindrical Si(100), which looked like a mirror but
geometry did not lend itself to analysis by were composed of 40 nm hemispheres
many techniques. (Fig. 5a). The importance of oxygen ex-
A flow-deposition system was subse- clusion was discovered while draining
quently developed to grow thicker films solutions, as some oxygen was sucked
on flat disposable substrates [41]. The into the cell, unless an N2 gas blanket
first edition was a small thin-layer flow was kept above the solution. This helped
cell (Fig. 3a). The majority of the sys- account for one of the major problems
tem was very similar to that used to- early on, reproducibility. Deposits were
day: pumps, electrochemical cell, valves, formed one day, but not the next, using
solution reservoirs, and potentiostat, all the same conditions. Progress was exceed-
controlled by computer. Those first de- ingly slow until the oxygen problem was
posits were excessively heterogeneous, better understood.
however [41]. The deposits suffered from a The deposit quality was greatly im-
lack of adequate oxygen exclusion, gasket proved with the H-cell [42, 58], and Fig. 12
effects, and bubble problems. In addition, shows XRD patterns for deposits of CdTe,
CdSe, and CdS made using the system.
those deposits were grown with a sin-
In each case, the deposits have a [110] pre-
gle set of unoptimized deposition poten-
ferred orientation, and crystallize in the
tials. Scanning electron microscopy (SEM)
zinc blende (cubic) structure. Graphs of
images revealed extensive 3D growth.
deposit thickness versus the number of
Interestingly, many deposits were com-
cycles were linear, a good indication of a
posed of smooth bumps, suggesting they
surface-limited growth. The best deposits
might result from layer-by-layer growth
of CdTe were a little thin, however, cor-
on a series of nucleation sites. The de-
responding to the growth of only about
posits were stoichiometric by EPMA, but 0.4 ML/cycle. In hindsight, oxygen exclu-
graphs of coverage versus cycle num- sion was probably still a problem, the
ber were not linear, as expected, but samples were being overrinsed each cycle,
increased exponentially, clearly demon- and a constant set of deposition potentials
strating increasing surface roughness or were used.
3D growth. As noted, the most important advance
There were a number of excuses made, resulting from changing to the H-cells
and a number of solutions suggested was that CdTe deposits were more re-
for those first deposits, such as prob- producible. Studies of the dependence of
lems with valves, with gaskets, and with deposit thickness on various cycle variables
substrates [41]. The next set of CdTe thin- were possible using the H-cell design [58],
film growth studies were made using a including the dependency on the poten-
H-cell (Fig. 3b) [42, 58]. The substrates tials used for Te deposition, Te stripping,
were simply suspended in the H-cells, and Cd deposition. Plateaus in coverage
and the solutions were pumped in from as a function of those potentials were
the bottom and then drained out. This observed, confirming the surface-limited
eliminated problems with gaskets and nature of the growth process [58]. That
bubbles, as there were no gaskets, and is, the deposit thickness was zero if the
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 537
Omega optimized
CdTe
(a)
Relative intensity
CdSe
(b)
CdS
(c)
5 25 45 65
2θ
Fig. 12 XRD patterns for (a) CdTe, (b) CdSe, (c) CdS. (Reproduced with
permission from Ref. [42].)
Cd deposition potential was too far pos- inside an N2 -purged Plexiglas box, greatly
itive, and if the potential was too far improving oxygen exclusion, and an
negative, some bulk Cd was formed for oxygen analyzer (Illinois Instruments,
each cycle, resulting in a thick rough de- model 2550) was used to measure the oxy-
posit. In between was a range of potentials gen levels in the box, 10 to 30 ppm. Use
where the deposit thickness was constant, of the thin-layer flow cell dropped the so-
close to 0.4 ML/cycle, suggesting the de- lution volumes used from 100 mL/cycle
position was controlled by surface-limited to about 1 mL/cycle, and allowed potential
reactions. control to be maintained throughout the
There were some major problems with deposition.
the H-cells, however, such as the need At that time, the cycle program was also
for 50-gallon drums to hold the re- changed, so that cathodic UPD was used
sulting chemical waste, as 100 mL of for deposition of both elements. Use of a
solution was used for each cycle. In pH 10 tellurite solution, TeO3 2− , instead
addition, potential control was lost for of pH 2, shifted the Te UPD potential to
the deposits each time the solution was better coincide with that for Cd UPD. A
drained, which can result in some de- program similar to that presently used
posit loss. for depositing CdTe is shown in Fig. 7.
Subsequently, the H-cell was replaced A comparison of the programs in Fig. 13
with a larger thin-layer flow cell (Fig. 3c), clearly shows the relative simplicity of the
with a 1 × 3 cm deposition area. The new program and hardware, compared to
pumps, valves, and cell were placed the older.
538 4 Underpotential Deposition
Cd-16 = 16 ml of Cd solution
Cd, B-16 = 16 ml of Cd solution
Cd, B-24 = 24 ml of Cd solution
−1.20 Te-16 = 16 ml of Te solution
Te Te, B-24 Te, B-24 = 24 ml of Te blank solution
strip O. C. O. C. = Open circuit
Potential
−1.00
Te, B-24 Te-16 Te Te, B-24
[V]
O. C. O. C. dep O. C.
−0.80
Cd
Cd, B-16 Cd-16 Cd, B-24
dep
−0.60
1 2 3 4 5
(a) min
B-1 = 1 ml of Cd blank
Cd-1 = 1 ml of Cd solution
Te-1 = 1 ml of Te solution
−0.75
Potential
Te deposition B-1
[V]
−0.70
Cd-1
Te-1
Te-1
B-1
B-1
Cd deposition
−0.65
30 60
(b) s
Fig. 13 Diagrams of the cycle program for use with (a) the H-cell (Fig. 3b) and (b) the large
thin-layer flow cell (Fig. 3c).
As noted above, the deposits made with Figure 11 is a study of the potential
the H-cell design were thin, only about dependence of deposit thickness, using
0.4 ML/cycle. For a 200-cycle deposit, 200-cycle CdTe deposits, formed with the
they appeared deep blue, the result of large thin-layer flow cell, as a function of
interference effects in the 30-nm-thick the Te deposition potential. A 0.7 V plateau
films. With the new cycle program and in deposit thickness suggests broad flexi-
large thin-layer cell, the best deposits bility in the choice of the Te deposition
appeared gold in color, and ellipsometric potential. Those results were very encour-
measurement indicated they were very aging. At positive potentials, the coverage
close to 1 ML/cycle. A study of the dropped dramatically, as no Te was de-
thickness as a function of the number posited, so there was nothing for Cd to
of cycles is shown in Fig. 14, where the deposit on. At potentials below −0.7 V,
line is straight, indicating surface-limited some bulk Te appears to have deposited
control of growth. Figure 15 is a graph of for each cycle, as the thickness greatly ex-
the square of absorptivity X photon energy ceeded that expected for ML/cycle growth,
versus energy, from which a band gap of and the morphology became ‘‘sandy’’
1.55 eV has been extrapolated, equivalent (Fig. 16). Sandy is a term used in this group
to literature values for CdTe. to describe deposits where 3D growth is
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 539
200 188
180
160
140
Film thickness
120
[nm]
100
80
72.6
60
40
35.5
20 14.4
0
0 50 100 150 200 250 300 350 400 450 500
Deposition cycles
Fig. 14 Thickness of CdTe deposits formed with the thin-layer flow cell, as a function of the number
of cycles. (Adapted from Ref. [60].)
3.0
CdTe 190-nm thickness
Gold on silicon
2.5 Experimental band gap EG = 1.55 eV
2.0
(α h ν)2
1.5
1.0
A
0.5
0.0
1.2 1.4 1.6 1.8
hν
[eV]
Fig. 15 Graph showing extrapolation to the band gap for CdTe films. (Adapted from
Ref. [60].)
evident in an optical microscope, appear- This is a strong indicator that the growth
ing as dark or multicolored dots, sand, mode is no longer layer by layer, and
when a 200-cycle deposit is examined at the deposit is roughening. Ellipsometric
1000 X with a metalographic microscope. thickness measurements are only accurate
540 4 Underpotential Deposition
Good Sandy
Fig. 16 Optical micrograph, 1000 X, taken with a
metalographic microscope of CdTe deposits. The good deposit
is on the left, and was produced using reasonable potentials.
The bad deposit is on the right, and was produced with a
program in which the potential for Te was excessively negative,
leading to roughening of the deposit, and what is referred to
here as sand, owing to its appearance in the microscope.
for smooth films [111], and became mis- described above, and Fig. 4 is a TEM of
leading for the films formed at such one of their deposits. Their cell was a wall
negative Te potentials. In between these jet design, and they used a single pump
potential extremes, the 0.7-V wide plateau to suck the solution from a distribution
was evident (Fig. 11), where the deposit valve, through their cell. The deposit was
thickness was consistent with ML/cycle formed in a continuous-flow mode, not
growth. The presence of this plateau is the stop flow used by this group. Their
consistent with surface-limited control of design greatly simplifies the hardware and
the deposition. Activation-limited growth appears to work just as well as the thin-
would have resulted in increased deposi- layer flow system. In addition, Villegas
tion as the potential was decreased. Mass was the first to notice the need to adjust
transfer–limited growth would have re- potentials as the deposits grow. They used
sulted in a plateau as well, but at a a cycle incorporating the reactions shown
much higher growth rate, and would not in Eqs. (4 and 5).
have been expected to increase again at Films of CdSe have been grown with
still more negative potentials. In addi- the automated flow-cell systems, both us-
tion, the deposit thickness for the plateau ing the H-cell [42] (Fig. 3b), and recently
was not a function of the precursor con- with the large thin-layer cell (Fig. 3c) [86].
centration, consistent with surface-limited Comparisons of XRD patterns have sug-
control. gested that in both cases the quality of
Thin films of CdTe have also been the CdSe deposits were better than equiv-
produced using an automated flow-cell alent CdTe films formed under similar
system by Villegas and coworkers [43]. conditions (Fig. 12b). Figure 17 shows an
Their films were very similar to those XRD pattern for a 200-cycle CdSe deposit,
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 541
10 000 Au
(111)
CdSe
8000 (111)
Intensity
6000
[counts]
Au
(200)
4000
CdSe Au
(220) CdSe (220)
2000 (311)
0
20 30 40 50 60 70
2θ
[degrees]
Fig. 17 XRD of CdSe formed using the large thin-layer flow cell, and 200 cycles. (Adapted from
Ref. [86].)
formed using cathodic UPD for both el- in growth and measurement at that time,
ements. The probable reason for CdSe rather than any significant variation in
forming higher quality deposits than CdTe coverage as a function of potential. The
is the lattice-match with the Au substrate. plateau and trends in the graph are as
That is, from atomic level studies, it is expected. At potentials above −0.55 V,
known that an ML and a half of CdTe [63, the deposits die off, while at potentials
78, 79], or CdSe [83, 84], will form a (3 × 3) below −0.68 V, some bulk deposition of
surface unit cell on Au(111). The struc- Se results, the growth rate exceeds an
tures proposed to account for these (3 × 3) ML/cycle, and the surface roughens (sand
unit cells involve superposition of an ML is evident).
and a half of these compounds, with the Typical 200-cycle CdSe deposits appear
zinc blende structure, on the Au surface, gold in color, as did the CdTe deposits.
where three Au lattice constants match up Given that the band gaps of CdSe and
with two times the Cd–Cd distance in the CdTe are both direct and about 1.6 eV,
(111) plane of CdTe or CdSe (Fig. 18). In similarities in appearance are expected.
the case of CdTe, this results in a 6% lat- Visual inspection of the 1 cm × 3 cm
tice mismatch, while it is less than 1% for CdSe deposits shows them to be more
CdSe on the same Au(111) surface. This homogeneous than corresponding CdTe
suggests there should be fewer defects, or deposits, in general.
a larger critical thickness for CdSe deposits Weaver and coworkers formed superlat-
compared with CdTe [112]. tices with CdSe and CdS, using EC-ALE,
Studies of the potential dependence of without an automated system [85]. They
CdSe deposit thicknesses show a much studied their relatively thin deposits by
shorter plateau region with CdSe than surface-enhanced Raman (SERS), examin-
CdTe (Figs. 11 and 19). The variability in ing stress build-up in the deposits.
thickness evident in the plateau (Fig. 19) CdS growth, by EC-ALE, has been
probably represents the standard deviation studied by more groups than any other
542 4 Underpotential Deposition
(a) (b)
√ √
Fig. 18 Structures proposed to account for the (a) ( 7 × 7)R19.1◦ and (b) (3 × 3) structures
observed in the formation of monolayer of CdTe. (Adapted from Ref. [61].)
200.0 190.6
155.6
150.0
Thickness
[nm]
compound (Table 1) [42, 52, 65, 67, 75, 85, as a function of the numbers of cycles
87, 89–91, 94, 113]. Initial EC-ALE studies performed. The dependence of thickness
in this group of CdS were performed with on the Cd deposition potential, for CdS
a TLEC (Fig. 9), to determine potentials deposits, revealed a plateau between −0.3
for a cycle [67]. Cd and S coverages were and −0.55 V, with the best deposits formed
determined coulometrically for deposits at −0.5 V, using a pH 5.9, 10 mM, CdSO4
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 543
deposits formed with a greater number of time, however. A paper looking at the pho-
cycles. toelectrochemistry of ZnS films as a func-
A ZnS cycle was automated using the tion of the layer thickness was published
H-cell design, with the idea of form- by Yoneyama and coworkers in 2000 [76].
ing phosphor screens for field-emission Their study involved the growth of films
displays (FED) [54]. Doping studies were of increasing thickness with an increasing
performed with Ag, Cu, and Mn, and will number of EC-ALE cycles. The deposits
be discussed in a subsequent section. The were studied with photoelectrochemistry
films showed evidence of 3D growth and (photocurrent curves) and clearly showed
roughening, and were rich in Zn (a Zn/S the dependence of the band gap on the
ratio of 1.35 from EPMA). XRD showed thickness of the ZnS films. The band gaps
the deposits to be zinc blende ZnS, of of deposits formed with more than 4 cy-
variable quality, depending on the doping cles, about 1.5 nm, showed the expected
scheme. Deposits formed with as many dependence on thickness. The thinner
as 200 cycles were produced. A plot of deposits were blueshifted from the bulk
coverage versus cycle number was expo- value of 3.6 eV, to about 4 eV. However,
nential; the coverage per cycle increased deposits formed with less than 4 cycles
as more cycles were performed, consistent did not display the theoretical exponen-
tial dependence expected. This work was
with some 3D growth and roughening.
extended by the formation of CdS/ZnS su-
SEM of the deposits grown on ITO re-
perlattices [75], discussed in a subsequent
vealed small smooth bumps, suggesting
section.
that the deposits could be layer-by-layer
InAs was the first III to V compound
growth on 3D nucleation sites.
grown using EC-ALE, although there was
From the exponential graph of coverage
some early work concerned with the for-
versus cycle number, and the presence
mation of GaAs [68, 95]. In those early
of excess Zn, it appears that the Zn
GaAs studies, no more than a single
potential was pushed too far negative, ML was ever formed. The major prob-
causing some bulk Zn deposition and 3D lems at that time were the reactivity of
growth. These ZnS deposits, formed five Ga, and the hardware problems previ-
years ago, were deposited using the same ously discussed: lack of oxygen exclu-
cycle potentials throughout the deposition. sion and loss of potential control during
Graphs of Zn/S ratio, as a function of rinsing.
the number of cycles, indicated that after Indium, on the other hand, is less re-
50 cycles the deposits were very close active than Ga, making InAs easier to
to stoichiometric, possibly a little S rich. form than GaAs. Deposits of InAs of good
Excess Zn began to show up at the 100- quality have been formed using the au-
cycle point. It is probable that the ZnS tomated thin-layer flow-deposition system
deposits would be greatly improved, using (Fig. 3c) [40, 62]. Homogeneous deposits
the present hardware and shifting the have been formed with close to 1 : 1 stoi-
potentials through the first 30 cycles. chiometry. EPMA suggested the deposits
In 1999, Foresti and coworkers pub- were somewhat rich in As, as much
lished a paper describing the growth ZnS as 20%, but attempts to better optimize
and CdS, using their automated deposition the deposition potentials did not greatly
system [45]. Few details were given at that improve the EPMA results. Elemental
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 545
analysis, using inductively coupled plasma the (αhν)2 versus energy, the band gap was
mass spectroscopy (ICP-MS), suggested estimated to be 0.36 eV, in agreement with
these deposits were closer to 5% rich in literature values. Band gaps for the InAs
As and that there may have been some In deposits appear to be sensitive functions of
loss during EPMA analysis. Figure 20 is an a number of cycle variables. Several sam-
XRD pattern for zinc blende InAs, formed ples resulted in band gaps closer to 0.44 eV.
with 500 cycles. There are no indications These blueshifts appear to result from
of As in the XRD. From Fig. 21, a plot of smaller crystallites, nanoclusters, formed
Au
6000 (111)
5000
4000
Intensity
[counts]
InAs Au
3000 (111) (200)
2000 InAs Au
(220) InAs (220)
1000 (311)
0
20 30 40 50 60 70
2θ
[degrees]
Fig. 20 Glancing-angle XRD pattern of a 270-cycle deposit of InAs. (Adapted from Ref. [62].)
InAs, 0.36 eV
[arb. units]
(αhν )2
when the deposition conditions were less potential is shown in Fig. 24. At posi-
than optimal. tive potentials, above −0.6 V, little de-
AFM images of the InAs films show posit is formed, as would be expected.
them to be relatively conformal with the Au Below −0.6 V, a relative plateau is ob-
substrate, with some texture, but little 3D served, but which gradually increases be-
growth for a 250-cycle deposit. Figure 22(a) tween −0.625 and −0.775 V. Below about
is an image of an Au on glass substrate, −0.7 V, the deposits correspond to more
annealed at 550 ◦ C for 12 h. Figure 22(b) is than 1 ML/cycle, and some roughening
of an equivalent substrate onto which 250 is evident with optical microscopy. Be-
cycles of InAs have been grown. low −0.775 V, the coverages measured
The InAs program involved slowly with ellipsometry drop to the 1-ML/cycle
shifting the potentials for In and As level, but microscopy shows the deposits
deposition negatively for the first 25 cycles, to be roughened and sandy. Coverage
and then holding them constant. Figure 23 measurements with EPMA also indicate
is a graph of cycle potentials as a function that these deposits are rich in As, and
of the number of cycles. The progression there is a significant increase in cov-
is essentially exponential, asymptotically erage, as expected, demonstrating that
approaching the steady state values, as the the ellipsometry readings were faulty at
junction potential is built up. these negative potentials. The As poten-
The thickness dependence of the InAs tial was driven too far negative, and some
deposits as a function of the As deposition bulk As was formed for each cycle. The
0 1.00 0 1.00
µM µM
Data type Height Data type Height
Z range 20.0 nM Z range 20.0 nM
(a) (b)
Fig. 22 AFM images of (a) the Au on glass substrate, annealed at 550 ◦ C for 12 h and
(b) equivalent substrate onto which an InAs deposit has been formed with 200 cycles of
deposition.
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 547
0 20 40
− 0.400
Potential
[V]
− 0.600
− 0.800
Cycle no.
Fig. 23 A graph of the deposition potentials for In and As as a function of
the cycle number in the growth of InAs. (Adapted from Ref. [62].)
100
80
60
40
20
0
− 0.900 − 0.850 − 0.800 − 0.750 − 0.700 − 0.650 − 0.600 − 0.550
As deposition potential vs Ag/AgCl
[V]
Fig. 24 A graph of deposit thickness as a function of the potential used to
form As atomic layers. Each deposit was made with 200 cycles of
deposition. (Adapted from Ref. [62].)
best deposits were those where the As Again, the thickness measurements at
potential was between −0.6 and −0.7 V: these negative potentials, determined with
where the deposits did not appear rough, ellipsometry, appear faulty, consistent
had close to 1 : 1 stoichiometry, and IR with the roughening observed. The best
absorption indicated a band gap close deposits were grown in the potential range
to 0.36 eV. between −0.725 and −0.8 V.
The dependence of the InAs film The importance of the substrate struc-
thickness on the In potential used is ture on the quality of the resulting de-
shown in Fig. 25. Again, little deposition posit cannot be overemphasized. Figure 26
occurs for potentials above −0.7 V. There shows two reflection transmission IR spec-
is a plateau between −0.725 and −0.8 V, tra for InAs. The InAs was grown on an Au
where deposits grow at close to 1 ML/cycle, on glass substrate that had been carefully
and the deposit thickness increased below annealed. Then, half the substrate was cov-
−0.8 V, as some bulk In began to form ered, and more Au was vapor-deposited
for each cycle, creating a rough deposit. on one half at room temperature. The
548 4 Underpotential Deposition
2
Monolayers/cycle
1.5
0.5
0
− 0.9 − 0.85 − 0.8 −0.75 −0.7
In deposition potential vs Ag/AgCl
[V]
Fig. 25 Graph of the deposit thickness as a function of the In potential
used. (Adapted from Ref. [62].)
1.0 Unannealed
Annealed
Transmission
Substrate
[arb. units]
Annealed Unannealed
0.8
Deposit
0.6
2000 4000 6000 8000
Frequency
[cm −1]
Fig. 26 IR reflection transmission spectra for InAs films formed on smooth
annealed Au surface and roughened Au surface. Substrate was formed by taking
a well-annealed, smooth Au on glass substrate and vapor-depositing more Au
on one half, at room temperature, so that a series of 40 bumps were formed
over one half of the surface. The two deposits were subsequently formed in the
same electrodeposition run. (Adapted from Ref. [114].)
0 1.00 0 1.00
µM µM
Data type Height Data type Height
Z range 25.0 nM Z range 25.0 nM
10061844.001 10061753.001
Au/glass, flame annealed Au/glass, unannealed
(a) (b)
Fig. 27 AFM images of an Au on glass substrate that had been well annealed, and then half
of it was covered. More Au was vapor-deposited on the other half at room temperature,
forming a surface composed of 40-nm Au hemispheres. (Adapted from Ref. [114].)
Given positive results with InAs, it was 115–119]; however, reduction of Sb, under
felt that InSb could be grown using EC- the conditions used by this group, was very
ALE, since the voltammetry of Sb is difficult. The stability of elemental Sb over
more straightforward than that for As. a large potential range suggested that InSb
That is, Sb+3 solutions have well-defined, formation would be tractable.
relatively reversible UPD features (Fig. 28), The potentials used to form InSb
while As+3 solutions show significant were applied in a roughly exponential
irreversibility (Fig. 6e). In addition, it has progression, as with InAs, through the
been shown that As is easily reduced first 30 cycles. Figure 29 shows the (111)
to arsine species at modestly negative XRD peak for zinc blende InSb, from
potentials [68, 95], making it difficult to a film formed with 200 cycles on a Cu
keep in the deposit if other steps in the substrate. The poor quality of the XRD
cycle are performed at overly negative appears to result from the small grain size
potentials, one of the basic problems in the deposit, as seen in the AFM image
with GaAs formation by EC-ALE. The in Fig. 30(b). The surface is composed of
potentials for Ga deposition were too smaller crystallites than the corresponding
similar to those where As is reduced to InAs deposits (Fig. 30a). The origin of
arsine. There are reports in the literature this difference appears to be the lattice
describing the reductive conversion of Sb constants, again. If we assume that the
to stibine, H3 Sb, or related species [110, interface is similar to that for CdTe (zinc
550 4 Underpotential Deposition
20.0
Current
0.0
[ µA]
−20.0
InSb
2500 (111)
Intensity
[counts]
1500
0
24 26 28 30 32
2θ
[degrees]
Fig. 29 XRD pattern of the (111) peak for zinc blende InSb, formed with 200
cycles on a Cu substrate. (This figure was adapted from Ref. [62].)
blende) on Au(111), then there should formation of CuInSe2 , CIS, over the last
again be the 2 : 3 match between the 17 years [120–132]. A number of electrode-
compound and the Au surface. The lattice position methodologies have been used,
mismatch for InSb would be 6%, as it including codeposition and a number of
was for CdTe and Au, as CdTe and InSb two-stage methodologies. The impetus of
have the same lattice constant (0.648 nm), that work has been the excellent photo-
while InAs and CdSe have the same lattice voltaic properties of CIS.
constant (0.606 nm), resulting in a much Initial TLEC studies developing an EC-
better lattice mismatch, only 1% with Au. ALE cycle to form CIS were published
There have been a number of stud- in 1996 [96]; however, a complete cycle
ies directed toward the electrochemical was not accomplished at that time. Steps
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 551
0 500 0 500
nM nM
Data type Height Data type Height
Z range 10.0 nM Z range 10.0 nM
inas 05051659.001
InAs 7-23-99, 200 cycles InSb 12-15-99, 200 cycles on annealed Au
(a) (b)
Fig. 30 AFM images of (a) 200 cycles of InAs and (b) 200 cycle of InSb, both on Au on glass
substrates.
involving Se UPD, followed by Cu, Se, In, are more problematic. Deposits have been
and then Se again were successful. How- made, which visually look very good, but no
ever, keeping In on the electrode while the peaks have as yet been observed with XRD,
next atomic layer of the significantly more besides those due to the Au substrate. Pos-
noble Cu was deposited, has proven prob- sible reasons for the absence of XRD would
lematic. Presently, these studies are being be the fact that the films are probably less
revisited using the automated flow-cell de- than 100-nm thick, have small crystallites,
position system, and Cu complexation is and the pattern for In2 Se3 has many peaks.
being used to shift its deposition potential Figure 31 is an AFM image of one of the
to more closely match that of In. However, deposits. EPMA indicates that the deposits
problems persist in keeping the In in the are close to stoichiometric. Reflection ab-
deposit. To better understand the mech- sorption measurements give a band gap of
anism for CIS formation, studies of the
1.6 eV, while there are various reports of
EC-ALE formation of In2 Se3 and Cu2 Se
band gaps between 1.5 and 1.8 eV in the
have begun.
literature [97].
It appears that some Cu2 Se has been
formed, but not a pure phase. Work is con-
tinuing on the deposition conditions to 4.2.4.1Toward Growing Device Structures
see if different phases can be formed, and There are a number of ways to intro-
some very interesting deposit morpholo- duce dopants into an EC-ALE deposit.
gies have been produced. The In2 Se3 films For instance, they can be introduced
552 4 Underpotential Deposition
500
250
0
0 250 500
nM
Fig. 31 An AFM image of a 200-cycle deposit of In2 Se3 . (Adapted
from Ref. [97].)
Intensity
resulted in the emission spectrum shown growth of CdS on Au, using EC-ALE,
in Fig. 32, which is characteristic of Mn. which was then capped with a single ML
of HgS [94]. The HgS capping layer was
4.2.4.2 Diodes deposited in two ways, one using EC-ALE
There are a number of papers in the litera- and the other by chemical exchange. The
ture concerning formation of compound chemical step involved exchange of the last
semiconductor diodes by electrodeposi- layer of Cd in a solution containing Hg
tion, the most popular structure being a ions. The deposits were studied with STM,
CdS-CdTe-based photovoltaic. In general, electrochemistry, and photoluminescence.
CdS was deposited first on an ITO on glass Photoluminescence showed excellent cou-
substrate, followed by a layer of CdTe, pling between the electrodeposited HgS
usually by codeposition [133–143]. layer and the underlying CdS, while photo-
So far, there has been little work done luminescence from the chemically formed
on the formation of heterojunctions using layer was not nearly as good.
EC-ALE. One study, however, performed Sailor and Martin and coworkers grew
by Shannon and coworkers, involved the an array of CdSe-CdTe nanodiodes [144]
554 4 Underpotential Deposition
InSb
Repeat
unit
InAs
compound layer had to be adjusted with Superlattices should display satellite peaks
each cycle. at angles corresponding to the period
Figure 34 is a graph of the potential of the lattice [152]. Given that the in-
program used to grow the first 100 cycles terplaner spacings, in the [110] direction,
of an InAs/InSb superlattice. Experience are 0.35 nm for InAs and 0.374 nm for
showed that the potentials had to be InSb, ideally, a period composed of 10 ML
changed for both elements for each of each should be 7.24-nm thick. Using
cycle. Those potentials and changes were Bragg’s law, with 0.1789-nm X rays, the
determined experimentally, and were not angle between the primary peak and its
extensively optimized. Figure 34 shows satellites should be 0.71◦ . Figure 35 sug-
that the potentials for As and Sb were each gests the presence of two shoulders, each
shifted to lower potentials with successive about 0.93◦ from the central peak. The
cycles in each half period. On the other quality of this preliminary XRD pattern
hand, the In potential was increased is bad, but the symmetric shoulders are
during the InSb half period, and decreased evident. On the basis of 0.93◦ , the pe-
during the InAs half period. At present, riod for the superlattice appears to be
these potential changes were arrived at closer to 5.5 nm, suggesting that about
by observation. It is hoped that feedback 3/4 ML/cycle was actually deposited.
can be incorporated to account for and Elemental analysis with ICP-MS was
optimize the progressions or that the performed on a dissolved piece of the
potential shifts can at least be modeled. superlattice, and indicated a deposit com-
Given the lattice mismatch of InAs position of In0.49 As0.37 Sb0.14 . It was grat-
(a = 0.606 nm) with InSb (a = 0.648 nm), ifying to find the deposit close to 50%
about 6.5%, defects are expected. At best In, as expected. However, there appears
a strained-layer superlattice would result. to be significantly more As than Sb. The
Figure 35 is an XRD pattern for a 41- deposit was started with 30 cycles of an
period InAs/InSb deposit, where each InAs buffer layer, so the deposit should
period was 10 cycles of InAs followed by be about 10% higher in As than Sb, but
10 cycles of InSb. The central [110] re- there was significantly more than 10% ex-
flection is near 28◦ and is quite broad. cess As over Sb. The absolute coverages
0 50 100
−0.35
Deposition potentials
−0.45
In potential
−0.55
As potential
−0.65 Sb potential
−0.75
−0.85
Cycle no.
Fig. 34 Deposition potential versus cycle number for an InAs10/InSb10-41X
superlattice: In (solid line), As (dot dash), Sb (dash). (Adapted from Ref. [62].)
556 4 Underpotential Deposition
105 27.887
90 26.876 28.779
[counts/second]
75
Intensity
60
45
30
25 27 29 31 33
2θ
[degrees]
Fig. 35 Grazing incidence angle X-ray diffraction (XRD) pattern of an
InAs10/InSb10-41X superlattice. The incident angle was 0.50◦ . (Adapted
from Ref. [62].)
1.0
0.8
Signal
0.6
0.4
0.2
0.0
1000 1500 2000 2500 3000 3500
Frequency
[cm−1]
Fig. 36 Reflection IR of an InAs10/InSb10-41X superlattice. (Adapted from
Ref. [62].)
4.2 Electrodeposition of Compound Semiconductors by Electrochemical Atomic Layer Epitaxy (EC-ALE) 557
from ICP-MS were about 10% less than redshifted from pure InSb, it is probably a
would be expected if an ML was deposited Type II superlattice [154].
for each cycle, suggesting somewhat more
than the 3/4 ML/cycle, as indicated by the Acknowledgment
XRD satellite peaks, was deposited. The
extra material is easily understood if the Acknowledgment is made to the National
substrate is considered slightly rough. A Science Foundation, Divisions of Chem-
roughness factor (RF) of 1.15 would ratio- istry and Materials for support of this work.
nalize the ICP-MS and XRD results, where
an RF of 1.0 corresponds to an atomically References
flat surface. One explanation of the dispar-
ity between As and Sb is that close to a full 1. A. Y. Cho, J. Vac. Sci. Technol. 1971, 8, s31.
ML of InAs was deposited for each cycle, 2. A. Y. Cho, J. R. Arthur, Prog. Solid State
while the coverage of InSb was closer to Chem. 1975, 10, 157.
3. J. R. Arthur, J. Appl. Phys. 1968, 39, 4032.
a 1/2 ML/cycle. This conclusion is consis- 4. J. Y. Tsao, Materials Fundamentals of Molec-
tent with the fact that the InSb cycle had ular Beam Epitaxy, Academic Press, Boston,
not been well optimized. 1993.
Figure 36 is an IR reflection- 5. M. B. Panish, H. Temkin, Annu. Rev. Mater.
Sci. 1989, 19, 209.
transmission spectrum taken from
6. M. A. Herman, H. Sitter, Molecular Beam
the superlattice deposit. Very strong Epitaxy: Fundamentals and Current Status,
adsorption is evident at higher wave Springer-Verlag, Berlin, 1989.
number, nearly 0% transmission, for the 7. H. O. Pierson, Handbook of Chemical Vapor
250-nm-thick superlattice deposit. The Deposition, Noyes Publications, Park Ridge,
1992.
deposit starts to absorb at very low 8. F. S. Galasso, Chemical Vapor Deposited
energies, near 1000 cm−1 . Problems with Materials, CRC Press, Boco Raton, 1991.
the spectrometer’s beam splitter make it 9. W. Kern in Microelectronic Materials and Pro-
hard to get a clear measure of the leading cesses (Ed.: R. A. Levy), Kluwer Academic,
Dordrecht, 1989.
edge of the absorption, and thus a band gap
10. K. K. Schuegraf, Handbook of Thin-Film
measurement. However, a comparison Deposition Processes and Techniques, Noyes,
was made between the absorptivity of the Park Ridge, 1988.
superlattice and that of a single crystal of 11. P. C. Andricacos, C. Uzoh, J. O. Dukovic
InSb, the superlattice component with the et al., IBM J. Res. Dev. 1998, 42, 567.
12. P. C. Andricacos, Interface 1999, 8, 32.
lower band gap, around 1300 cm−1 . The 13. M. Fleischmann, H. R. Thirsk, (Eds.), Metal
absorptivity of the superlattice was about Deposition and Electrocrystallization, Ad-
20 times greater than that for single crystal vances in Electrochemistry and Electro-
InSb, indicating that the superlattice is chemical Engineering, Interscience Pub-
lishers, John Wiley & Sons, New York, 1963,
redshifted from InSb. If the deposit was
Vol. 3.
an alloy, InAsx Sb1−x , the band gap should 14. P. E. Light, D. Shanefield, J. Appl. Phys.
be blueshifted from InSb, assuming it is 1963, 34, 2233.
a linear function of the mole fractions of 15. K. R. Lawless, J. Vac. Sci. Technol. 1965, 2,
As and Sb, and the band gaps for the two 24.
16. H. J. Choi, R. Weil in Electrochemical So-
compounds, or about 0.3 eV [153]. These ciety National Meeting (Eds.: R. Weil,
results strongly indicate that the deposit R. G. Barradas), The Electrochemical Soci-
is not an alloy, and since the band gap is ety, 1981, p. 169, Vol. 81-6.
558 4 Underpotential Deposition
1.2
1.0
Normalized absorption
0.8
0.6
0.4
0.2
0.0
−20 0 20 40
Energy (Relative to Pb L3 edge 13055 eV)
[eV]
Fig. 1 Normalized Pb XANES spectra for Pb on Pt at various potentials:
−0.24 V (– – –), 0.25 V (positive sweep)( ), 0.25 V (negative sweep)
(·– · –·), 0.6 V (- - - - ), and 1.15 V ( . . . ). (Reproduced with permission
from Ref. [7].)
to be slightly increased by adlayers of in the double layer. This effect could ac-
several metals [10]. The energy levels in- count for the small increase of the redox
volved in the redox reactions are those reactions by metal adlayers. Even in the ab-
of the adsorbates, not of the substrates, sence of trace amounts of chloride, small
and these systems seemed ideal to provide enhancements in the Fe2+ /Fe3+ reaction
insights into the unresolved question of rates caused by Bi on Pt and Ag on Au
the role of the nature of the metal elec- were found. No effect was observed for a
trode in the kinetics of the outer-sphere Cu adlayer on the reaction on Au. Rhodes
redox reaction. The charge transfer in these and coworkers [12] found an enhanced rate
‘‘noncatalytic’’ reactions takes place in the of a Cr3+ /Cr2+ reaction in HCl on Au by
outer Helmholtz plane. Consequently, it Bi, Pb, Tl, and Sn adlayers, and attributed
should not be significantly affected by this to the enhanced adsorption of Cl− in
the metal adlayer. The data reported so the presence of adatoms. Chlorides make
far are in accord with this statement (cf. bridges that are necessary for this complex,
Refs. [1] and [2]), thus confirming a small inner-sphere, and redox couple reaction to
role that the nature of the metal elec- occur.
trode has these reactions. In a recent work, Several mechanisms have been pro-
Nagy and coworkers [11] took special care posed to explain the small enhancement
to remove traces of chlorides that can be effects of metal adatoms. A change in PZC
coadsorbed with metal adatoms and that caused by metal adlayers should induce a
can increase the concentration of cations change in the ϕ2 potential and thus affect
564 4 Underpotential Deposition
3.0
0
i
1
−1.5 2
3
4
−3.0
0.2 0.6 1.0
E SHE
[V]
Fig. 2 Current–potential curves for a mixture of pyrocatachol (1 mM) and
o-benzoquinone (1 mM) on Pt and Bi/Pt rotating disk electrodes in 0.5 M
HClO4 with 1 mM Bi(ClO4 )3 . Sweep rate 10 mV s−1 . Rotation frequency
(1) 12.5, (2) 25, (3) 50 and (4) 75 Hz. (Reproduced with permission from
Ref. [2].)
used for the oxidation of adsorbate [18, 2. Bifunctional mechanism: The bifunc-
19], differential electrochemical mass spec- tional mechanism, often used in catal-
trometry (DEMS) [20], and radiotracer ysis, has been considered in electro-
measurements [25]. Bewick and cowork- catalysis by Cathro [33] and Shibata and
ers [26] employed electrochemically mod- Motto [34]. The surface is considered
ulated IR spectroscopy (EMIRS), which to have two types of sites that have
clearly identified CO as the poisoning distinct roles in the reaction. The sub-
species. This was strongly supported by strate, such as Pt, breaks the bonds
further work [27]. Two e.p.s. were found in organic molecules upon adsorption,
for several single-crystal surfaces, which while the adatom can adsorb oxygen-
also suggest that CO is a poison [24, 28]. containing species that can oxidize
More details on this topic can be found in strongly adsorbed intermediates such
reviews in Refs. [29–31]. as CO. Oxygen adsorption on adatoms
The mechanism of the catalytic ac- is expected to occur at potentials nega-
tion of metal adlayers in the oxida- tive of oxygen adsorption on Pt. This
tion of organic molecules has been in- facilitates CO oxidation at potentials
terpreted by using several models in- more negative than on bare Pt, thus
freeing the surface for the main reac-
cluding a ‘‘third-body’’ effect, an elec-
tion.
tronic effect, and the bifunctional catalyst
3. Electronic effects: Electronic effects can
mechanism.
result from the modifications of the
1. Third-body effect: A third-body effect, properties of uncovered near-neighbor
which is equivalent to an ‘‘ensem- sites and next near-neighbor sites. The
ble’’ effect in heterogeneous catalysis, adatoms modified in the interaction
is based on the role of metal adatoms in with substrates, may, per se, take part
blocking the surface sites for a side re- in reactions.
action that generates poisoning species,
Striking catalytic effects observed in the
or in blocking the adsorption of the in- oxidation of formic acid on Pt modified by
hibiting species, which requires more Pb and Bi adlayers are shown in Fig. 3 [35].
than one surface site for adsorption. Currents in anodic sweeps are enhanced
The reaction in the main pathway can by almost two orders of magnitude, which
proceed at the unoccupied sites that is also observed in steady state mea-
now form smaller ensembles. In this surements [35–37]. Figure 4 shows quasi-
model, adatoms do not enhance the stationary curves for the Pb/Pt system as
rate on the uncovered substrate sites. In a function of Pb coverage. (Metal adatom
electrocatalysis, the effect was proposed coverages are defined in two ways through-
by Conway and coworkers [32] to inter- out the literature reviewed in this text.
pret the effect of acetonitrile on formic According to one definition, the cover-
acid oxidation on Pt. The third-body age is defined as the ratio between the
effect is operative for any adsorbate number of covered and total number of
statistically distributed on the surface substrate atoms. The coverage is also de-
that is not desorbed upon adsorption of fined as a number of atoms of adsorbate
reacting species and intermediates of per one atom of substrate. The latter defi-
reactions. nition is being used more often. For large
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials 567
80
60 Pt/Pbad 60
Pt/Biad
[mA cm−2]
[mA cm−2]
40 40
i
20 20
0 0
0 0.4 0.8 1.2 1.6 0 0.4 0.8 1.2 1.6
E E
[V] [V]
Fig. 3 Oxidation of HCOOH (0.26 M) on Pt with Pb and Bi adlayers (full
lines) and on bare Pt (dashed line) in 1 M HClO4 . Lead concentration 1 mM;
sweep rate 50 mV s−1 . (Reproduced with permission from Ref. [35].)
600 θ Pb
:0
500 : 0.12
: 0.52
• : 0.62
400
: 0.73
[mV]
E
300
200
100
0
−6 −5 −4 −3 −2 −1 0
log j
[A cm−2]
Fig. 4 Quasi-stationary current–potential plots for HCOOH (0.5 M)
oxidation on sputtered Pt modified by Pb adlayer and HCOOH adsorbates in
0.1 M HClO4 . Current taken 2 min after stepping the potential. The current
density is given with respect to the free-Pt surface (16% of the whole surface).
(Reproduced with permission from Ref. [37].)
3.2
2.4
θ Pb = 0.36 θ Pb = 0
1.6
[mA]
I
0.8
0
(a)
m /z = 44
8
θ Pb = 0.36
θ Pb = 0
MI × 1010
[A]
(b) 0
m /z = 45
2
MI × 1010
θ Pb = 0
[A]
1
θ Pb = 0.36
0 0.5 1 1.5
E vs RHE
(c) [V]
Fig. 5 Current (a) and mass signal (b,c) voltammetry curves in a
0.01 M HClO4 solution containing 5 mM H12 COOH: ( ) for
sputtered Pt previously with modified θPb = 0.36 and
θH13 COOH = 0.49; ( . . . ) only H13 COOH was adsorbed
(θHCOOH = 0.84); sweep rate 100 mV s−1 . (Reproduced with
permission from Ref. [37].)
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials 571
almost the same magnitude for the cat- and is similar to that observed with the
alytic effects of UPD Pb and preadsorbed UPD Bi adlayer [35].
Pb on Pt(111). The maximum effect was Two roles were assumed for Bi on
observed for a Pb coverage of 0.29, which Pt(100) [48, 51]. First, Bi inhibits formation
almost completely prevented the adsorp- of poison through a third-body-type effect.
tion of CO. These authors also indicate The poison coverage decreases linearly
that a third-body effect and a bifunctional and becomes zero at θBi = 0.45(0.22).
mechanism cannot fully explain the ob- Second, the highest HCOOH oxidation
served effects. To summarize, for Pb, a rates are observed at high Bi coverage,
part of the observed catalytic effect is due ∼θBi = 0.9(0.45), which indicates a direct
to a third-body effect, but there is a grow- role of Bi in the oxidation of HCOOH,
ing evidence for the electronic effect to be that is, an electronic effect is operative
partly responsible for the catalysis. in this system. The maximum rate is,
however, still lower than the intrinsic rate
on Pt(100). Bi forms on Pt(100) a c(2 × 2)
Bismuth Large effects of Bi have been
adlayer with a coverage of θBi = 0.5, with
found in several studies [35, 44]. Clav-
adatoms residing at bridge positions [71,
ilier and coworkers [48, 51] showed that
72]. Therefore, at ∼θBi = 0.45, a small
Bi preadsorbed on Pt(111) enhances the
number of Pt sites apparently remains
kinetics of HCOOH oxidation in a large uncovered and a direct participation of Bi
interval of coverages from 0 to 0.8(0.27) at in the reaction appears likely. Weaver and
potentials below 0.6 V. The accumulation coworkers [70] showed by IR spectroscopy
of strongly bound intermediates is sup- a complete cessation of CO formation at
pressed at very small Bi coverages. This θBi = 0.2, which is in good agreement with
suggests that the next-nearest neighbor’s Refs. [48, 51]. The catalytic effect of Bi was
Pt atoms must be affected by Bi and that ascribed to the attenuation of the degree
the third-body effect cannot account for the of CO poison formation on Pt(100), which
inhibition of the poison formation. At high the authors interpreted in terms of a third-
Bi coverages, large currents are measured body effect.
at potentials where Bi adsorbs OH [55]. Recent work involved stepped Pt sur-
Therefore, in that potential region, it could faces with controlled adsorption of Bi on
act through the bifunctional mechanism. step edges [73]. For the stepped single-
Figure 6 shows voltammetry curves for crystal surfaces Pt(554), Pt(332), and
Pt(111) with θBi = 0.82(0.27) and the oxi- Pt(221), with nine-, five-, and three-atom
dation of HCOOH on that surface [48, 51]. wide terraces, respectively, the enhance-
Weaver and coworkers [70] reported an IR ment factor for HCOOH oxidation in-
study of this system that shows small CO creases as the terrace width decreases. Bi
coverages on Pt(111), even in the absence appears to block adsorption of poison on
of Bi. This surface is the least poisoned reactive (110) oriented step sites and de-
surface from the low-index planes. The op- creases the reaction ensemble size, which
timal coverage of Bi was 0.2, which was increases the rate on narrow (111) terraces.
interpreted in terms of an ensemble, that Leiva and coworkers [74] reported the
is, a third-body effect. An enhancement simulation of the catalytic effects of several
factor of approximately two orders of mag- adatoms by using the model in which both
nitude is cited in both works [48, 51, 70] adatoms and poisoning species block one
572 4 Underpotential Deposition
100
[µA cm−2]
0
I
0.1
E vs RHE
[V]
(a) 40
[mA cm−2]
30 c
I
20
10 d b
a
0
[mA cm−2]
0.5 1
θ Bi
I
10
0
0.1 0.5
E vs RHE
(b) [V]
Fig. 6 (a) Voltammetry curve for Pt(111) with adsorbed θBi = 0.82(0.41) in 0.5 M
H2 SO4 . Current multiplied by 5 for short sweep. (b) First sweep for the oxidation of
0.25 M HCOOH. Insert: Peak currents at 0.5 to 0.55 V versus Bi coverage, (•) anodic
and (×) cathodic sweeps. (Reproduced with permission from Refs. [48, 51].)
surface site. Eight and six near neighbors at 50 × 50 lattices up to the coverage of
covered by adatoms protect one site from θM , and then the poison is adsorbed up
poison adsorption on square and hexag- to θp . The oxidation currents are calcu-
onal lattices, respectively. The adatoms lated for such a configuration for several
are assumed to be adsorbed randomly catalytic cases. In the Pt(111)–Bi system,
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials 573
Bi–Pt pairs were assumed to be responsi- the oxidation of HCOOH through an elec-
ble for the catalytic effect, without poison tronic effect.
formation. Figure 7 shows a comparison
between the experiment and the calcu- Antimony The effect of Sb adlayers on
lated curve. The agreement is surprisingly Pt(100) [60] were ascribed to the sup-
good considering the crude assumptions pression of poison formation, and the
that adatoms cover one surface site and maximum oxidation current is obtained
that they are randomly distributed. Bi is at high coverage, θSb = 0.9(0.3), when no
known to inhibit three H adsorption sites poison is detected at the surface. Sub-
on Pt(111) [75]. Nevertheless, the simula- stantial currents are observed for this
tion data seem to corroborate the electronic surface. The maximum activity, however,
effect for the Bi–Pt system. The data for does not surpass the intrinsic activity of
Pt(111) and Pt(100) revealed the role of Pt(100). A third-body effect was found op-
substrate in determining the effect of Bi erative for this system. Kizhakevariam and
in suppressing poison adsorption (the elec- Weaver [76] found that Sb inhibits the ad-
tronic effect for Pt(111), but a third body for sorption of CO on Pt (100) and Pt (111)
Pt(100)). In addition, on the same surface by decreasing the twofold binding geome-
(Pt(100)), Bi can suppress poison forma- tries deduced from the relative νCO band
tion by a third-body effect and can catalyze intensities. The optimum catalysis for
50
40
30
[mA cm−2]
j
20
10
Pt(100) is observed at θSb = 0.35. Sb causes and its reaction has been thoroughly in-
significant up-shifting of the CO band vestigated on surfaces modified by metal
formed from HCOOH oxidation, which adatoms. The overall reaction involves the
suggests that dissociative chemisorption exchange of six electrons per molecule:
is triggered at adjacent sites, electronically
modified by Sb. Figure 8 shows IR spectra CH3 OH + H2 O −−−→ CO2 + 6H+ + 6e−
for Pt(100) and Pt(100) with an Sb adlayer (2)
(θSb = 0.25) and the corresponding oxida- This is a multistep reaction, which prob-
tion current versus θSb and θCO curves. The ably proceeds through a sequential de-
enhancement has been interpreted in-line hydrogenation that can produce several
with third-body and electronic effects [76]. intermediates, as indicated in the general
Good agreement was found between the scheme proposed by Bagotzky and cowork-
simulated effects and experimental data ers [86]. Details of the dehydrogenation of
for the Sb–Pt(100) system in which Sb was methanol on Pt have not been established.
assumed to obey the third-body effect [74]. From kinetic isotope measurements of
CD3 OH and CH3 OH oxidation, it was
shown that the first removal of H in the
Noble metal monolayers The systems
reaction
involving a deposition of ultrathin layers
of noble metals on noble metal substrates,
CH3 OH −−−→ CH2 OH + H+ + e− (3)
such as Ru on Pt [77] or Pd on Pt [78]
or on Au [79], exhibit enhancements of is the rate-determining step [87, 88].
the oxidation of organic molecules. These For methanol, more than for formic acid,
systems are outside the scope of this evidence points out that besides CO there
review, and will be only mentioned here. In is at least one additional adsorbate – the
general, deposition of such layers occurs COH species. Three e.p.s. values are found
at overpotentials, and they, unlike the often for adsorbate oxidation and show
UPD adlayers, cannot be dissolved except that it has to contain an H atom [89].
at extreme positive potentials. Usually, Beden and coworkers [90, 91] reported
such deposition produces 2D or 3D from IR measurements the linear, bridge-
clusters, but pseudomorfic deposits can bonded CO and CHO-like species or
also be formed, as in the case of Pd on species with a carbonyl bond. By using sub-
Au(111) [80, 81] and Pt on Au(111) [82]. tractively normalized Fourier transform
Ru can be ‘‘spontaneously’’ deposited in infrared spectroscopy (SNFTIRS), Iwasita
an electroless process on Pt [83, 84]. Ru and coworkers [92] have determined abso-
layers catalyze oxidation of methanol on lute absorption bands for linear-, bridge-,
Pt [83, 84], and Pd enhances the oxidation multibonded CO and COH adsorbates.
of HCOOH on Au [79] and Pt [85] single- Nichols and Bewick [93], by using a flow
crystal electrodes. The activity of the IR-electrochemical cell, identified COH
Pd-modified surfaces depended markedly and CHx OH species as reactive interme-
on the Pd layer thickness and on their diates. Recent results with DEMS indicate
crystallographic orientation. predominantly a mixture of CO and COH,
or only CO as adsorbed intermediates [89,
4.3.3.2.2Methanol Methanol is consid- 94, 95]. Model calculations of methanol ox-
ered the most attractive fuel for fuel cells idation on Pt(111) that provide a good fit to
E E
[V] [V]
0.36
0.36
2056 0.26
2071 6 0.6
0.22 0.15
0.11 0.04
0
i
−0.06 4 0.4
−0.09
θ CO
[mA]
−0.20 −0.17
−0.25 −0.25
1850
2 × 10−3 a.u. 2 × 10−3 a.u. 2 0.2
0
2000 1900 2000 1900 0.0 0.2 0.4
ν ν θ Sb
(a) [cm−1] (b) [cm−1] (c)
Fig. 8 (a) Single potential alteration infrared (SPAIR) spectra for 0.1 M HCOOH in 0.1 M HClO4 on Pt(100); (b) Sb-modified (θSb = 0.25) Pt(100),
spectra obtained during the 2 mV s−1 anodic sweep and (c) θCO (solid circles) at about 0.25 to 0.5 V and peak currents (open squares) and currents at
0.35 V (solid squares) versus θSb plot. (Reproduced with permission from Ref. [76].)
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials
575
576 4 Underpotential Deposition
the experimental data without adjustable Pt(111) and Pt(100) with Sn adlayers [105],
parameters involve a reactive intermediate and for Pt3 Sn alloys [107]. Janssen and
with the stoichiometry H : C : O [96]. Moolhuysen [53] concluded that Pt is mod-
Negligible catalytic [2, 97, 98] and pre- ified through a ligand, that is, through
dominantly inhibition effects [99] are ob- an electronic effect, while Motoo and
served for methanol oxidation on Pt in coworkers [111] proposed that a bifunc-
acid solutions. The exceptions are the ef- tional mechanism is operative for Sn.
fects of Sn [53, 100–102] and Ge [101, 102]. Bittins-Cattaneo and Iwasita [110] found
Significant catalytic effects have been ob- that Sn(II) species are responsible for
served for Pb, and Bi on Pt in alkaline the catalytic effect, which can form a
solutions [103, 104]. However, carboniza- hydroxy-complex such as Sn(OH)+ that
tion of the alkaline electrolyte with CO2 can transfer O to the organic residue. The
makes these systems unsuitable for practi- activity was ascribed to a suitably oriented
cal applications. For single-crystal Pt(100) adsorbed H2 O on a Pt/Sn surface. Figure 9
and Pt(111) surfaces, the reaction has been shows a linear sweep voltammetry and
found to be inhibited by the Bi adlayer [49, mass signal of CO2 for Pt and Pt with ad-
105]. sorbed Sn. The observed enhancement of
Some agreement exists on the effects methanol oxidation is in the middle range
of Sn, although the results are not re- of those reported so far. Molecular orbital
producible. The UPD of tin differs from studies demonstrated that Sn atoms on
the majority of the UPD systems [106]. Pt(111) can weaken CO adsorption [112],
Its oxidation state is extremely potential- but subsequent work showed that Sn in
dependent. The effects of Sn on the the Pt surface does not attract H2 O or ac-
oxidation of CH3 OH on Pt are attractive tivate the formation of the OHads species
because they occur at very low potentials. that can oxidize CO intermediates [113].
The system Pt–Sn has been investigated The data on Sn indicate that its state on
in three forms: Sn irreversibly adsorbed,
Pt surfaces is highly dependent on sur-
Sn adsorbed by UPD, and ordered single-
face structure and on the method of Sn
crystal Pt3 Sn alloy [107]. Vassiliev and
deposition, which determine its catalytic
coworkers [100] found a Temkin adsorp-
properties for CH3 OH oxidation. Catalytic
tion behavior of neutral Sn at Pt in the
effects are often observed with dispersed
potential region from 0 V to 0.2. Above
Pt, while inhibition is usually observed
0.2 V, tin undergoes a partial oxidation to
with flat surfaces. While the potential for
Sn+ , or SnOH at ∼0.5 V. A tenfold in-
practical application of Sn-modified Pt for
crease of the rate in the presence of Sn
methanol oxidation is small, this system
has been found and has been ascribed to
provides a strong support for the impor-
the bifunctional mechanism [97]. Frelink
tance of bifunctional catalysts for methanol
and coworkers [108, 109] demonstrated by
oxidation.
ellipsometry that a submonolayer of Sn
oxide disappears from Pt surfaces in the
presence of methanol. The reports of con- 4.3.3.2.3 Carbon Monoxide Carbon mo-
siderable catalytic effects of Sn [53, 101, noxide is a model adsorbate for study-
102] have not been reproduced in re- ing molecular adsorption in catalysis and
cent studies [24, 105, 110]. Inhibition was electrocatalysis. It is a major poisoning
observed for the methanol oxidation on species in the oxidation of small organic
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials 577
Pt + Sn
Current 4
[mA]
Pt
(a) 0
Pt + Sn
2
1012 Mass signal
m/e = 44 (12CO2)
[A]
Pt
0
0.3 0.5 0.7
Potential vs RHE
(b) [V]
Fig. 9 Oxidation of 1 M CH3 OH in 0.5 H2 SO4 on Pt and Pt with Sn adlayer
(θSn = 0.34(0.17)). (a) Current and (b) mass intensity during a potential
scan of 10 mV s−1 . (Reproduced with permission from Ref. [102].)
15
10
[µA]
5
I
−5
(a)
0.2 0.4 0.6 0.8
6
[*10−12 A]
I
Most of the work in this area was done by hydrogenation of some unsaturated hy-
Kokkinidis and coworkers [2] who demon- drocarbons [2]. While the reduction of the
strated catalytic effects of metal adlayers nitro group is catalyzed by metal adlayers
for the reduction of nitro and nitroso com- on Pt, the reduction of nitrobenzene is par-
pounds [131–134], and for the catalytic tially inhibited on Ag by Pb adatoms [135].
580 4 Underpotential Deposition
A number of publications exist in the Rus- (Pt): R−NO2 + 4Pt −−−→ R−NO2
sian literature about the effect of cation 2Had
addition on various electroorganic syn- −−−−→ R−N−O
−H2 O
thetic reactions that are not discussed in
terms of the UPD effects, although most of −−−→ R−N−O
the systems probably involve the metal ad- mHad
layer formation (see Ref. [1] and reference −−−→ Products (4)
r.d.s
therein).
2e− + 2H+
Electroreduction of nitro compounds is (Pt/M ): R−NO2 −−−−−−−→ R−NO
ad
of considerable importance for electroor- −H2 O
ganic synthesis. Interesting catalytic ef- 2e + 2H+
−
1 2 34
−1.6
[mA cm−2]
i
−3.2
−4.8
0.3
Pt/Pbad, high coverage
1.2 V 65
30 4
sec 3
0.05 V
0.2
−E vs SHE
XV
2 m sec
[V]
Pt
0.0
−1
−2
0.013 0.00
0.27
−3
[A cm−2]
log j
0.36
−4
−5
−6
−0.08 −0.06 −0.04 −0.02 0 0.02 0.04 0.06 0.08
E vs RHE
[V]
Fig. 13 H2 evolution on a Bi/Pt(111) electrode in 0.5 M H2 SO4 . Tafel plots for
different Bi coverages: θBi = 0, 0.13 (0.06), 0.27(0.13), and 0.36 (0.18).
(Reproduced with permission from Refs. [150, 151].)
the adlayer structure during the course hollow sites in the structure of the Tl ad-
of H2 evolution has been obtained by layer. This prevents further compression
SXS for Pt(111) with a Tl adlayer [154]. with decreasing potential. These data seem
Figure 14 shows data for Tl adlayers on to corroborate the assumption for the H2
Pt(111) in solutions of two different pH evolution through the hollow sites in the
values. A significant H2 evolution current Bi adlayer on Pt(100).
flows at Pt(111) despite the close-packed Several approaches were used to quan-
configuration of the Tl adlayer. The elec- tify inhibition effects of metal adlayers.
trocompression of the Tl–Tl distance with These involved calculations of currents
decreasing potential is observed with this of H2 evolution based on the order-
system. H2 evolution has no effect on the disorder theory of alloys [141, 144, 145],
structure of the Tl adlayer, but the Tl–Tl and simulations based on geometric [150,
distance ceases to contract at its onset 151, 153] and long-range electronic ef-
(Fig. 14). It is likely that in addition to fects [150, 151]. Verification of the mod-
defect sites, H2 evolution occurs through els used in some simulations seems
584 4 Underpotential Deposition
(a) (b)
3.60
3.55
pH = 0 E 2H+/H2 = 0 V
a TI
pH = 3 E 2H+/H2 = −0.18 V
[Å]
3.50
3.45
3.40
−0.4 −0.3 −0.2 −0.1 0 0.1
E NHE
(c) [V]
Fig. 14 (a) In-plane diffraction pattern from close-packed hexagonal Tl
adlayer (open circles) on Pt(111) in 0.05 M H2 SO4 solution with 1 mM Tl+ .
Solid circles diffractions from Pt(111). (b) Real space model for Tl adlayer. (c)
Tl–Tl interatomic distance determined from the rocking θ scans at diffraction
positions as a function of potential. The electrocompression increases in
solution with pH = 3 beyond that observed at pH = 0, but it ceases to
change when H2 evolution starts (see text). (Adapted from Ref. [154];
reproduced with permission.)
necessary in order to obtain a more metals. Similar effects were observed for
complete understanding of the inhibition Pb, Tl [158, 159] and for Ge, Ga [160]. In
effects. deaerated solutions, the inhibition effect
Related to the inhibition of H2 evolution can be quite pronounced. In the presence
is the effect cations on the corrosion of of O2 , however, an enhanced O2 reduction
iron. Dražić and Vorkapić [157] interpreted can increase corrosion rates.
a decrease of the corrosion of Fe in
sulfuric acid in the presence of Mn2+ , 4.3.3.5 Oxygen Reduction
Cd2+ , and Zn2+ in terms of the inhibition Oxygen reduction is a multielectron re-
of H2 evolution by the UPD of these action that involves several elementary
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials 585
steps in the reaction mechanisms. In with UPD metal adlayers. Catalytic effects
aqueous solutions, oxygen reduction ap- were observed also for Pt [166], glassy
pears to occur by two overall pathways: carbon and graphite [2], and Ru [167]
a ‘‘direct’’ four-electron reduction, and a surfaces with Pb and Tl adsorbates. The
‘‘peroxide’’ pathway that involves H2 O2 as latter three cases are somewhat different
the intermediate [161]. For acid solutions, from the UPD adlayers on metal surfaces.
the reaction in a direct four-electron For carbons, it appears that the UPD
pathway is occurs through an interaction of metal
with carbon surface groups. For oxidized
O2 + 4H+ + 4e− −−−→ 2H2 O Ru surfaces, cations interact with surface
E0 = 1.229 V vs NHE (6) oxides.
Figure 15 shows rotating disk-ring mea-
Two electrons are exchanged in a ‘‘perox- surements of O2 reduction on Au and
ide’’ pathway: Au/Pbad . The upper panel shows the UPD
of Pb on Au in NaOH solution. The disk
O2 + 2H+ + 2e− −−−→ H2 O2
current shows doubling of the diffusion-
E0 = 0.67 V (7) limited current, which means that four
electrons are exchanged in the reaction
Peroxide can undergo further reduction, on Pb/Au. Consequently, the ring current
shows that peroxide generation is almost
H2 O2 + 2H+ + 2e− −−−→ 2H2 O
completely suppressed. A series reaction
E0 = 1.77 V (8) mechanism, involving a two-electron re-
duction to HO2 − , followed by exchange of
or it can catalytically decompose into O2 another two electrons in the reduction of
and H2 O. Peroxide is the final reduction HO2 − to OH− , has been found from the
product on some less active surfaces. analysis of the rotating disk-ring electrode
On some surfaces, the two pathways can measurements, namely,
occur in parallel. The same pathways
are operative in alkaline solutions but Au: O2 + H2 O + 2e−
at different set of potentials involving
−−−→ HO2 − + OH− (9)
intermediates and products determined by
−
the solution pH [161]. Pb/Au: O2 + H2 O + 2e
−−−→ HO2 − + OH− (10)
4.3.3.5.1 Alkaline Solutions Foreign me-
tal adatoms cause remarkable catalytic HO2 + H2 O + 2e −−−→ 3OH−
− −
0.2
0.1 Au/Pb
Au
[mA cm−2]
0.0
i
900 rpm
−0.1
Au /Pbad
0.0
2
Au
[mA cm−2]
−I D
Au /Pbad
Au, indicates a rate-controlling first elec- and high surface area Pt in alkaline
tron exchange. The slope of −55 mV was solutions. These effects shift the half-
also explained by a slow first electron ex- wave potential to more positive val-
change. A low electrosorption valency of ues than those on bare Pt (Fig. 16).
Tl (γTl = 0.6) had to be assumed to satisfy Rotating disk-ring electrode measure-
this slope [165]. ments indicate a series mechanism for
Yeager and coworkers [162] reported O2 reduction, instead of a direct path-
interesting catalytic effects of Pb and way with the O−O bond scission on
Tl on the O2 reduction on smooth bare Pt.
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials 587
−1.0
−0.8
Current Pt I
[mA] Pt II
Pt / Tl
−0.4
Pt / Pb
Powercat 2000
Powercat 2000/Tl
0.0
with adatoms. A somewhat larger inhibi- cause of the generation of H2 O2 and the
tion was observed by Kokkinidis [172] in decrease of the current by 1/2. This conclu-
more concentrated solutions. These data sion should be verified by rotating disk ring
strongly suggest that proton transfer in electrode (RDRE) measurements. A pseu-
O2 reduction, involving any kind of H in- domorphic Ag monolayer and a hexagonal
teraction with the surface, is not likely to aligned bilayer on Pt(111) cause a com-
occur [178]. plete inhibition of O2 reduction [183]. This
Further understanding of the catalysis inhibition as a function of Ag coverage was
of O2 reduction by metal adlayers has used to determine a bridge adsorption of
been achieved by investigation of reac- O2 during reduction on Pt(111).
tions on single-crystal surfaces and by
structure-sensitive techniques. The cat-
4.3.3.5.3 Hydrogen Peroxide Reduction
alytic effects observed with single-crystal
H2 O2 , which is inert on most Au sur-
electrodes are in agreement with those
faces, except on Au(100) and its vicinals
obtained with polycrystalline surfaces [1,
in alkaline solutions, is particularly ef-
164]. SXS studies during the course of O2
fectively catalyzed by several Pb, Tl, and
reduction helped to resolve the question
Bi adlayers [1, 2, 165, 168]. Gewirth and
of the coverage-dependent catalytic effect
coworkers [184, 185] identified by atomic
of Tl. The close-packed, rotated-hexagonal
force microscopy (AFM) the (2 × 2)-Bi
Tl phase on Au(111) was found to sup-
structure in the potential region at which
port predominantly two-electron reduc-
a maximum activity for H2 O2 reduction
tion, while the aligned hexagonal phase in
on Au(111) is observed. The rectangu-
alkaline solution and patches of the (2 × 2) √
lar (p × 3) Bi high-coverage adlayer on
phase in acid solution, which gradu-
Au(111) is not catalytically active for H2 O2
ally vanish during the reaction (Fig. 17),
reduction (Fig. 18), which is very similar
are conducive to a four-electron reduc-
to the inactivity of a close-packed Tl ad-
tion [170].
layer on Au(111) [170]. H2 O2 reduction is
A total inhibition of O2 reduction on
catalyzed on Au(111) with a low-coverage
Au(111) by a Cu adlayer in the poten-
Tl phases, while it is almost completely
tial region where Cu forms a honeycomb
√ √ suppressed by the rotated close-packed
( 3 × 3)R30 structure in the absence of
hexagonal phase [179]. For the Au(111)/Pb
O2 was determined by scanning tunnelling
surface, however, the highest activity was
microscopy (STM) by Itaya and cowork-
found to occur when the surface is maxi-
ers [180]. Sulfate/bisulfate is adsorbed in
mally covered by Pb islands [186].
the center of a Cu honeycomb [181, 182],
which could be the cause of the inhi-
bition. A Cu submonolayer on Pt(111) 4.3.3.5.4 Mechanisms of Catalytic Effects
causes inhibition by decreasing the cur- Several mechanisms have been proposed
rent to approximately one half, while a to explain the catalytic effects of metal ad-
Cu monolayer causes complete inhibition. layers. These include (1) specific electronic
The reduction of O2 to H2 O2 was assumed properties of the metal/adlayer surface that
in the presence of either the simple or can increase the adsorption energy of O2
√ √
the honeycomb ( 3 × 3)R30◦ submono- and intermediates [1], (2) redox properties
layer structure. A change from bridge to of metal adlayers [164], and (3) formation
end-on adsorption of O2 is viewed as a of the bridge bond between O2 and the Au
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials 589
400
Au(111)
300
[µA cm−2]
200
100
I 0
−100
−200
0
Au(111)
−0.5 2e −
[mA cm−2]
Au(111)/TI
i
−1 4e −
rot-hex TI
(2 × 2)TI
−0.8 −0.6 −0.4 −0.2 0 −0.2 −0.4 −0.6 −0.8 1
E SCE
[V]
Fig. 17 Voltammetry curve for the UPD of T on Au(111) in 0.1 M
HClO4 solution containing 5 mM Tl+ (upper panel). Polarization curve
for O2 reduction on Au(111) (dashed line) and Au(111) with
the (2 × 2)Tl, and rotated-hexagonal Tl adlayers. Sweep rate 10 mV s−1 .
(Adapted from Ref. [179]; reproduced with permission.)
adatom sites, which facilitates the weak- on Au with metal adlayers. This favors the
ening of the O–O bond [168, 179, 184, complete reduction of O2 by weakening
185]. Yeager and coworkers [166] assumed the O−O bond. Alvarez and Jüttner [188]
a destabilization of anodic films on Au and reported a simulation of the catalytic ef-
Pt in the interaction with metal adlayers. fects of Pb on H2 O2 reduction on Au(111)
McIntyre and Peck [164] proposed a redox and Au(100) assuming essentially a lattice
cycle in which Tl adatoms are oxidized by gas model (LGM) for the Pb adlayer at low
O2 and HO2 − and are subsequently rede- coverages and well-defined structures at
posited. Schriffrin [187] assumed a redox high coverages. Good agreement between
couple in which Pb2+ is oxidized by HO2 − the simulation and experiment is consid-
to Pb4+ . Jüttner and coworkers [168] as- ered as a support for the bridge adsorption
sumed a change from an end-on adsorp- model. However, the assumptions regard-
tion of O2 on Au to a bridge adsorption ing the Pb adlayer are not in agreement
590 4 Underpotential Deposition
60
A1
40
A2
20
Current
[µA]
−20 C2
−40 C1
(a) −60
−1
Current
[mA]
−2
−3
with the results of SXS studies of the Adžić and Wang [191] studied the UPD
Pb/Au(111) system by Toney and cowork- of Tl on Au(111) in the presence of Br− ,
ers [189]. Additionally, the large coverage which has three potential-dependent, com-
by Pb islands, not a specific structure, was mensurate Tl–Br phases that affect O2 re-
found to produce a maximum activity for duction in different ways. The catalytically
√
H2 O2 reduction on Au(111) [190]. No or- active (3 × 3)-2TlBr phase on Au(111)
dered structure has been reported so far was monitored by SXS during the course of
for Pb on Au(100). Gewirth and cowork- oxygen reduction by measuring the struc-
ers [184, 185] proposed a heterobimetalic ture factors for its two low-order in-plane
bridge model for H2 O2 interaction with the diffractions as a function of potential. The
Au/Bi surface as a cause of the enhanced IR voltage drop in the thin-layer cell did
catalytic activity. not allow drawing definitive conclusions
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials 591
regarding the stability of the active ad- X-ray scattering. Under oxygen reduction
layer phase and the origin of the catalytic conditions, both the (2 × 2)-Bi and a high-
√
effects [192]. coverage (p × 3)-2Bi adlayer structures
Most recently, Tamura and cowork- are stable (Fig. 19). The electrochemical
ers [193] have confirmed a stability of the drop cell configuration utilized for these
low-coverage Bi(2 × 2) phase on Au(111) measurements minimizes the IR potential
during O2 reduction with in situ surface drop relative to that in commonly used
150
(2 × 2)Bi
100 2
Current density
2
50
[µA cm−2]
−50
(p × √3 )2Bi
−100 1 M HCIO4 + 5 mM Bi3+;
10 mV s−1
√3
−150
(a) p
100
0
0.18
O2 (a) N2
Current density
Intensity
−100 0.12 O2
[a.u.]
[µA cm−2]
Bi(2 ×2)
0.06
−200 0
0.18
(b)
Intensity
−300
[a.u.]
0.12
0.06
−400
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7
−500 E vs Ag/AgCl
[V]
E vs Ag/AgCl
(b) [V]
Fig. 19 Voltammetry curves for the UPD of Bi on a Au(111) electrode in 1 M HClO4
with 5 mM Bi3+ (a) and oxygen reduction on the Au(111) surface with Bi adlayers in
the same solution (b); −1
√ sweep rate 10 mV s . The potential regions of the existence
of (2 × 2) and (p × 3) Bi phases are indicated in the graph. Inset: X-ray diffraction
(XRD) intensities at the (1/2, 1/2, 0.2) position (characteristic for the (2 × 2) Bi phase)
as a function of potential in the presence of O2 and N2 obtained in a thin-layer and
drop cells. λ = 0.386 Å; sweep rate 2 mV s−1 .
592 4 Underpotential Deposition
thin-layer cells. Together with rotating disk than by incorporating Au atoms into the
electrode measurements, the results indi- bulk lattice at the surface defects and dis-
cate distinctly different catalytic properties locations, or by 3D growth. Davidović and
for the two different ordered bismuth Adžić [195, 196], however, have shown that
adlayer phases. The potential region cor- Sb and Tl cause a decrease in the nucle-
responding to the (2 × 2)-Bi phase, the O2 ation rate of Au deposition. The nucleation
reduction is promoted to a four-electron changes from instantaneous in Au(CN)2 −
reaction, albeit with relatively slow kinet- solutions in the absence of Sb or Tl ad-
√
ics. The close-packed (p × 3)-2Bi phase, layers into a progressive process in their
formed at more negative potentials, ap- presence (Fig. 20). For progressive nucle-
pears to have a limited number of sites ation, the rate can be described by the
available for the four-electron reduction, equation
but the reaction kinetics on this adlayer
is enhanced by the increase of overpo- t 3/2
tential. These results rule out the redox I (t) = 2zF A N∝ π(2Dc)3/2 M1/2
3ρ 1/2
mechanism as the explanation of the (12)
catalytic effect, which leaves the elec- where I is the current at time t, A is the
tronic/adsorption effects and the bridge area of the electrode, N is the number
bond as possible origins. of nuclei that can be obtained at a given
condition, and ρ is the density of the metal.
4.3.3.6 Electrodeposition of Metals As expected for progressive nucleation, the
Metal adatoms have profound effects on linear I versus t 3/2 plots based on Eq. (13)
the processes of electrodeposition of met- have been obtained at constant E from the
als. The effect of the addition of trace curves in Fig. 20(b).
quantities of various cations to the plating Sb, a well-known ‘‘wetting’’ agent in
solutions on the morphology of deposits gas phase metal deposition [197], causes
has been known for quite some time from remarkable morphological effects by pro-
the plating patent literature. Usually, an ducing very bright Au deposits at current
increased brightness is observed in the densities at which rough deposits were
presence of certain cations in plating so- obtained in its absence. Considerable cat-
lutions. In addition, catalytic effects are alytic effects of Sb adatoms were ascribed
observed with some systems that shift the to a change in the CN− adsorption in the
reactions to positive potentials. McIntyre presence of Sb adlayers. No Sb or Tl incor-
and Peck [194] rationalized these observa- poration was found with AES, or by atomic
tions in terms of the UPD adlayer effects absorption spectroscopy of the dissolved
with the example of the electrodeposi- deposit.
tion of Au from Au(CN)2 − solutions. The Sieradzky and coworkers [198, 199] have
adatoms of Pb and Tl induce a marked recently demonstrated that the UPD of
positive shift of the polarization curve and the Pb and Cu adlayers can mediate
extend the current density range in which growth of ordered multilayer deposits (up
bright, fine-grained deposits are formed. to 200 layers) of Ag on Au. Atomically
It has been proposed that at sufficiently flat epitaxial multilayers of Ag on Au(111)
high θM , these adatoms can act as 2D have been produced in the presence of the
nucleation centers, with crystal growth UPD of Pb and Cu. Ag is codeposited
proceeding around these centers rather with reversibly deposited UPD metal.
20 Au/Au(CN)2−
E
[mV]
1.0
−340
15
−320
E
i
i
[mV] −300
10 −550 −280
[mA cm−2]
[mA cm−2]
−500 0.50 −260
−450
5
−
−400 Au/Sbad/Au(CN)2
0 0
2 4 6 0 10 20 30 40 50
τ τ
(a) [s] (b) [s]
Fig. 20 (a) Potential step transients for Au deposition on Au from 0.1 M Na2 SO4 containing 50 mM KAu(CN)2 , pH = 8.5; (b) Potential step transients
for Au deposition on Au with Sb adlayer. Solution as in (a) with addition of 10 mM Sb2+ . (Adapted from Ref. [195]; reproduced with permission.)
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials
593
594 4 Underpotential Deposition
The latter is periodically deposited and onto which the deposited Ag atoms
stripped from the surface, and this serves undergo interlayer place-exchange with Pb
to significantly increase the density of atoms, forming 2D islands below the Pb
2D islands of Ag atoms, promoting a adlayer. Upon stripping of the Pb adlayer,
layer-by-layer growth mode. The UPD Ag continues to deposit with island
of the Pb monolayer covers the surface growth. Figure 21 shows a scheme of the
Ag-Pb interlayer
exchange
(a)
(b)
Fig. 21 Proposed mechanism of the UPD of Pb-mediated deposition of Ag. The Ag and Pb
atoms are light and dark shaded spheres, respectively. Arrows indicate depositing metal
cations. (a) Pb mediation: at potentials close to reversible Pb deposition, a Pb monolayer
covers the surface. The deposited Ag adatoms undergo interlayer place-exchange with Pb
adatoms (light green sphere) forming 2D islands below the Pb adlayer. (b) On the reverse
cycle, the Pb adlayer is stripped from the surface as Ag continues to deposit resulting in
island growth. (Reproduced with permission from Ref. [198].)
4.3 Electrocatalysis on Surfaces Modified by Metal Monolayers Deposited at Underpotentials 595
20 nm
0 nm
proposed mechanism for the Pb-mediated and in this context, the effects of UPD
deposition of Ag on Au. An ordered metal adlayers discussed above are very
multilayer Ag deposit can also be obtained interesting. The data illustrate remarkable
in a constant potential electrodeposition, possibilities for further exploration of the
as discussed in Refs. [198, 199]. Figure 22 effects of UPD adatoms in electrodepo-
shows the Ag deposits obtained as a sition of metals. It would be interesting
function of time at a potential of 0.125 V to see whether this method can produce
that maintains θPb = 0.8 [199]. a layer-by-layer growth in systems with
Atomic layer control of growth is an larger lattice misfits between substrate
ultimate goal of the electrodeposition of and deposit and with different chemical
metals and nanofabrication technology, properties.
596 4 Underpotential Deposition
43. S. Motoo, M. Watanabe, J. Electroanal. 69. J. Clavilier, J. Electroanal. Chem. 1987, 236,
Chem. 1976, 69, 429. 87.
44. M. Shibata, O. Takahashi, S. Motoo, J. Elec- 70. S.-C. Chang, Y. Ho, M. J. Weaver, Surf. Sci.
troanal. Chem. 1988, 249, 252–264. 1992, 265, 81–94.
45. S. Motoo, M. Watanabe, J. Electroanal. 71. C. A. Lukas, N. M. Marković, B. N. Grgur
Chem. 1979, 98, 203. et al., Surf. Sci. 2000, 448, 65–76.
46. R. R. Adžić, A. V. Tripković, N. M. Marko- 72. C. A. Lukas, N. M. Marković, P. N. Ross,
vić, J. Electroanal. Chem. 1983, 150, 79–88. Surf. Sci. 2000, 448, 77–86.
47. T. Iwasita, X. Xia, E. Herrero et al., Lang- 73. S. P. E. Smith, H. D. Abrun̆a, J. Electroanal.
muir 1966, 12, 4260–4265. Chem. 1999, 467, 43–49.
48. J. Clavilier, A. Fernandez-Vega, J. M. Feliu 74. E. Leiva, T. Iwasita, E. Herrero et al., Lang-
et al., J. Electroanal. Chem. 1989, 258, muir 1997, 13, 6287–6293.
89–100. 75. J. M. Feliu, A. Fernandez-Vega, J. M. Orts
49. S.-C. Chang, Y. Ho, M. J. Weaver, Surf. Sci. et al., J. Chim. Phys. 1991, 88, 1493–1518.
1992, 265, 81–94. 76. N. Kizhakevariam, M. J. Weaver, Surf. Sci.
50. H.-W. Kei, H. Hattori, H. Kita, Electrochim. 1994, 310, 183–197.
Acta 1996, 41, 1619–1628. 77. S. Motto, M. Watanabe, J. Electroanal.
51. J. Clavilier, A. Fernandez-Vega, J. M. Feliu Chem. 1979, 98, 203.
et al., J. Electroanal. Chem. 1989, 261, 78. G. A. Attard, A. Banister, J. Electroanal.
113–125. Chem. 1991, 300, 467.
52. R. R. Adžić, J. X. Wang, unpublished. 79. M. Baldauf, D. M. Kolb, J. Phys. Chem. 1996,
53. M. M. P. Jansenn, J. Moolhuysen, Elec- 100, 11 375–11 381.
trochim. Acta 1976, 21, 861–869. 80. L. A. Kilber, M. Kleinert. R. Randler et al.,
54. J. Clavilier, J. M. Feliu, A. Aldaz, J. Elec- Surf. Sci. 1999, 443, 19–30.
troanal. Chem. 1988, 243, 419. 81. H. Naohara, S. Ye, K. Uosaki, J. Phys. Chem.
55. U. W. Humm, D. Kramer, R. S. Zhai et al., B 1998, 102, 4366.
Electrochim. Acta 1998, 43, 2969–2978. 82. K. Uosaki, S. Ye, H. Naohara et al., J. Phys.
56. J. Clavilier, J. M. Feliu, A. Aldaz, J. Elec- Chem. B 1997, 101, 7566.
troanal. Chem. 1988, 243, 419. 83. E. Herrero, K. Franszczuk, A. Wieckowski,
57. A. Kozawa, J. Inorg. Chem. 1961, 21, 315. J. Electroanal. Chem. 1993, 361, 269.
58. V. E. Kazarinov, N. A. Balashova, M. I. 84. W. Chrzarnowski, H. Kim, A. Wieckowski,
Kulezneva, Elektrokhimiya 1965, 8, 975. Catal. Lett. 1998, 50, 69.
59. R. R. Adžić, N. M. Marković, Electrochim. 85. M. J. Llorca, A. J. Feliu, A. Aldaz et al., J.
Acta 1985, 30, 1473. Electroanal. Chem. 1994, 376, 151–160.
60. A. Fernandez-Vega, J. M. Feliu, A. Aldaz 86. Yu. B. Vassilyev, V. S. Bagotzky, O. A.
et al., J. Electroanal. Chem. 1989, 258, Khazova, Elektrokhimiya 1975, 11, 1505.
101–113. 87. K. Franaszczuk, E. Herrero, P. Zelenay
61. A. Fernandez-Vega, J. M. Feliu, A. Aldaz et al., J. Phys. Chem. 1992, 96, 8509.
et al., J. Electroanal. Chem. 1991, 305, 88. E. Herrero, P. Zelenay, A. Wieckowski, J.
229–240. Phys. Chem. 1994, 98, 5074.
62. X. H. Xia, Electrochim. Acta 1999, 45, 89. S. Wilhelm, T. Iwasita, W. Vielstich, J. Elec-
1057–1066. troanal. Chem. 1987, 238, 383–391.
63. M. J. Llorca, E. Herrero, J. M. Feliu et al., J. 90. B. Beden, F. Hahn, S. Juanto et al., J. Elec-
Electroanal. Chem. 1994, 373, 217–225. troanal. Chem. 1987, 225, 215–225.
64. E. Herrero, M. J. Llorca, J. M. Feliu et al., J. 91. S. Juanto, B. Beden, F. Hahn et al., J. Elec-
Electroanal. Chem. 1995, 394, 161–167. troanal. Chem. 1987, 237, 119–129.
65. J. Clavilier, R. Parsons, R. Durand et al., J. 92. T. Iwasita, F. C. Nart, B. Lopez et al., Elec-
Electroanal. Chem. 1981, 124, 321. trochim. Acta 1992, 37, 2361–2367.
66. M. Zhang, C. P. Wilde, J. Electroanal. Chem. 93. R. J. Nichols, A. Bewick, Electrochim. Acta
1995, 390, 59–68. 1988, 33, 1691–1694.
67. X. H. Xia, T. Iwasita, J. Electrochem. Soc. 94. H. Kita, H.-W. Lei, J. Electroanal. Chem.
1993, 143, 2559–2565. 1995, 388, 167–177.
68. T. Iwasita, R. Dalbeck, E. Pastor et al., Elec- 95. T. Iwasita, X. Xia, E. Herrero et al., Lang-
trochim. Acta 1994, 39, 1817–1823. muir 1996, 12, 4260–4265.
598 4 Underpotential Deposition