Research: Science and Education

Teaching the Rovibronic Spectroscopy of Molecular Iodine

J. Charles Williamson
Department of Chemistry, Willamette University, Salem, OR 97301; jcwillia@willamette.edu

The spectroscopy of molecular iodine has been discussed

in the pedagogical literature for over fifty years (115). Iodine has a low-lying bound excited state, the 3+0u B state,
which is accessible from the ground 1+g X state by visible
photons (Figure 1). Both absorption and fluorescence spectra of molecular iodine can be collected with instruments of
modest resolution, common to teaching laboratories, and
sample preparation is simple. The ease with which students
may observe evidence of quantized electronic and molecular
behavior in iodine spectra is thus the primary attraction of
the experiment. Students can determine many electronic potential parameters for both the B state and the X state using
straightforward spectral analysis, and students can use their
results to generate simplified potential energy surfaces with
which they may connect back to the spectra. I2 rovibronic
spectroscopy has become a staple of undergraduate physical
chemistry laboratory textbooks (1618); the experiment has
also been the focus of an article on cooperative learning (19).
Despite this long history, though, there is a lack of clarity in the literature on the proper way to guide students in
the analysis of molecular iodine spectra. As one example, a
key parameter derived from the I2 visible absorption experiment is e, the fundamental vibrational energy of the B state.

One would hope that students could find affirmation of

quantum mechanical theory by determining a precise and accurate value of e from the spectrum. A survey of molecular
iodine articles published in this Journal and elsewhere, however, does not show such corroboration (Table 1). Reported
values for e are found to fall into two groups, clustered
around either 125 cm1 (2, 4, 5, 10) or 132 cm1 (6, 11, 15).
The first group of values is in good agreement with the accepted value for e of 125.6687 cm1 (20). Unfortunately,
though, refs 2, 4, 5, and 10 all contain mistakes in the interpretation of the molecular iodine absorption spectrum, as
does the article reporting a value of 130 cm1. Only refs 6
and 15 analyze the spectrum correctly! As will be explained
below, the value of e can only be determined to two significant figures of accuracy from analysis of the visible absorption spectrum in the undergraduate experiment. This is
due to an unavoidable extrapolation error.
Evaluation of refs 115 has found that at least half contain mistakes of various types. The intent of this article is to
clear up the confusion by reviewing which molecular iodine
potential parameters may be determined from the absorption and fluorescence spectra, specifying how the spectra are
properly analyzed, and summarizing how well the determined
values can be expected to agree with accepted literature values. The article is broken into three sections: (i) finding Bstate potential constants from the absorption spectrum, (ii)
finding X-state potential constants from fluorescence spectra, and (iii) combining data from both types of experiments
to gain additional insight into the system. Each section is
subdivided into basic analysis, clarifications, and possible
supplementary projects for students with more time to devote to the experiment. Brief explanations are provided for
subtle features of the molecular iodine spectra, such as the
aforementioned extrapolation error intrinsic to the determination of e. A more thorough discussion is available for the
interested reader in the accompanying Supplemental
Material.W

Table 1. Reported Molecular Iodine

e Values

/cm1
e

123

123.4
125

www.JCE.DivCHED.org

130

Accepted

5
10

126.5
Figure 1. X-state and B-state potential energy curves for molecular
iodine. The shape and placement of potential energy curves are
quantitatively accurate, using data compiled from primary literature sources (20, 2426, 31). Tick marks on the left edges of the
potential wells indicate the locations of every tenth vibrational level.
The arrow labeled ~
vabs shows the energy change when X-state iodine at v = 0 absorbs a photon and is excited to an arbitrary
high vibrational level (v 35) of the B state.

Ref

4
11

132.1

133.4

15

125.67

20

NOTE: Values reported in pedagogical discussions on the

visible absorption experiment. Comparison is made to the most
recent accepted literature value.

Vol. 84 No. 8 August 2007

Journal of Chemical Education

1355

Research: Science and Education

Determination of B-State Potential Constants

Basic Analysis
The visible absorption spectrum of molecular iodine
shows a rich forest of vibronic transitions made up of a
cold band and several hot bands (Figure 2). Each peak in the
cold band is a transition from the lowest vibrational level of
the ground electronic state, v = 0, to a particular vibrational
level v with energy G (v) in the B state (Figure 1). The corresponding
absorbed photon energy abs will be the sum of

G (v) and an energy offset. As seen in Figure 1, the energy

offset is the spacing between the bottoms of the two potential wells, T e, minus the vibrational energy of v = 0 in the
ground state, G (0). The absorbed photon energy abs at a
particular peak in the spectrum is therefore equal to

(1)
abs ( v , v = 0 = Te G 0 + G v

When G(v) is written as a series expansion in v + 1/2, then

abs becomes

a bs v , 0 = Te G 0
1
+
e v +
2
1
+
e ye v +
2

1

e x e v +
2
3

(2)

+ ...

where eis the fundamental vibrational wave number of the

B state, and the higher-order terms correct for anharmonicity in the potential well. The quantities xe and ye are anharmonicity constants; typically the products exe and eye
are tabulated rather than just the constants.
Students find the series expansion coefficients in eq 2
by plotting absorbed photon energies (obtained from the
spectrum) as a function of v + 1/2 and then fitting the data
to a polynomial of appropriate order, usually two or three.
The traditional BirgeSponer analysis is not necessary (15).
Extrapolation of the data points to their maximum yields E *,
the energy required to take molecular iodine from the lowest vibrational level of the X state to the dissociation

limit of
the B state (Figure 1). Students use E * and T e G (0), the
constant term from their
polynomial fit (eq 2), to calculate
the B-state well depth De:

De = E * Te G 0

(3)

Clarifications
Four clarifications are in order with respect to analysis
of the absorption spectrum. First, the absorption peak locations used in the plot of abs versus v + 1/2 are best assigned
to the valleys on the high-energy side of each peak (18, 21,
22), and not the peak maxima. Each peak is wide owing to
many unresolved rotational transitions. Rotational energy
changes are neglected in eq 2, so the proper center of the
peak for this analysis would be the Q branch. Q branches
do not appear in the rovibronic peaks because J = 0 is forbidden for the B X transition of molecular iodine. If Q
branches were present, though, they would be located near
1356

Journal of Chemical Education

Figure 2. The absorption spectrum of gas-phase molecular iodine

from 500 to 650 nm at 40 C. Approximate ranges of significant
cold band and hot band contributions are indicated by the horizontal lines. The 514.5 nm Ar+ lasing line accesses the 43 0 vibronic
transition at the indicated peak.

the edges on the high energy sides of each peak. This is because the average internuclear distances of the X state and
the B state are significantly different, leading to different rotational constants and to superposition of the R branch over
the P branch (21).
The second clarification has to do with the accuracy to
which the coefficients in eq 2 may be determined. The typical range of v for which absorption peaks can be identified
in Figure 2 is from 15 to 52. The upper limit on v has
some play: peaks get closer together and FranckCondon
overlap gets progressively poorer on the high-energy side of
the cold band, so instrumental resolution and signal averaging determine the actual upper limit for v. The lower limit
on v, however, is constrained by the onset of hot band peaks
that overwhelm the cold band peaks around 575 nm. Without data below v = 15, two problems arise in the fit. First,
the coefficients will be highly dependent upon the order of
the polynomial fit even though the R 2 for the fit may be almost exactly one. For example, a third-order fit and a fourthorder fit on the same set of data can give values of e that
differ by 10 cm1. Second, the true energy spacings of B-state
vibrational levels change more slowly below v = 12 than
above v = 12. This behavior cannot be predicted from the
collected data, so the value of e found by undergraduates
is systematically too high, usually by 5 to 10 cm1.
For similar reasons, the value of E* found from extrapolation of the polynomial fit to a maximum will be systematically too low by 50 to 200 cm1. In this case the actual
spacings of the vibrational energy levels decrease more slowly
beyond v = 55 than would be expected based on data prior
to v = 55. The experimental value of De will also be systematically lower than the literature value.
The third clarification is with respect to the hot band
region of the absorption spectrum. Although the hot band
region looks as simple as the cold band (Figure 1), the hot
band region is actually an intricate mixture of unresolved and
overlapping contributions originating from v = 0 through
at least v = 5 of the ground electronic state. Undergraduates should be instructed to ignore this portion of the spectrum when characterizing molecular iodine because much
higher instrumental resolution is necessary to track the hot

Vol. 84 No. 8 August 2007

www.JCE.DivCHED.org

Research: Science and Education

Figure 3. Argon ion 514.5 nm laser-induced fluorescence spectrum of

gas-phase molecular iodine. Every fifth peak is numbered with the final value of v reached during relaxation. Odd peaks at higher v are
not apparent on this graph owing to poor FranckCondon overlap.
The lasing transition is blocked with a holographic notch filter, but the
location of the lasing wavelength is indicated by the dashed line marked
with a zero. The inset shows low-resolution and high-resolution scans
of the 58 0 relaxation following 58 1 laser excitation. The resulting rotational doublet is distinct at 0.25 nm resolution.

bands correctly. The five lowest values of e listed in Table 1

all come from erroneous hot band assignments.
As a final point, students must be provided the assignment of v for one absorption peak in order to carry out their
polynomial fit. If students also measure the fluorescence spectrum generated from the 514.5 nm line of an argon ion laser, then there is a context for providing this extra
information. The 514.5 nm line excites molecular iodine from
v = 0 to v = 43, and students can make the assignment to
the appropriate peak in the absorption spectrum (Figure 2).

Supplementary Analysis
Students may wonder about the number of terms to include in a polynomial fit (eq 2). Analysis of the molecular
iodine absorption spectrum provides students with an excellent opportunity to learn how the order of a fit is limited by
the noise and span of the data. Programs such as PeakFit
(Systat Software) calculate P values automatically for each fitting parameter, or students may evaluate the statistical merits of higher-order terms using F tests. An excellent discussion
on this latter topic is found in Garland et al. (18).
A second project would be for students to explore why
the best assignment of an absorption peak location is on the
high-energy side of the peak. Students can use the rotational
constants for the two states to find the value of J at which
the R branch turns around and heads towards overlap with
the P branch (21). They can then compare this value of J
with the room temperature rotational population distribution predicted by the Boltzmann equation.
Determination of X-State Potential Constants

Basic Analysis
Characteristics of the X-state potential surface can be determined by exciting molecular iodine to a specific vibrational
www.JCE.DivCHED.org

level in the B state and then collecting the resulting fluorescence spectrum. Excitation sources used in undergraduate
laboratories include mercury lamps (3), heliumneon lasers
(7, 11, 14), and argon-ion lasers (13). The argon ion 514.5
nm laser-induced fluorescence (LIF) spectrum shown in Figure 3 consists of a series of spikes to the red of the laser line
that gradually diminish in intensity. The 514.5 nm line excites molecular iodine to v = 43 in the B state, and each successive spike away from the laser line corresponds to relaxation
from v = 43 down to progressively higher values of v in the
X state. The overall intensity envelope of the fluorescence depends on instrumental sensitivity and FranckCondon overlap between v = 43 and the X-state vibrational levels.
The energy difference between the excitation pho
ton laser and the fluorescence photon fl is equal to the energy difference between the initial and final X-state vibrational
levels:
=
laser
f l v = 43, v

(4)

= G v G 0

When G (v) is written as a series expansion in v + 1/2, then

becomes:
1

= G 0 +
e v +

2
1
e x e v +

2

1
e ye v +
+
2

+ ...

(5)

Students find values by subtracting the photon energies

of their fluorescence peaks from laser, which is 19429.7694
cm1 for Ar+ 514.5 nm laser output (23). The coefficients in
eq 5 are then determined by plotting for each peak as a
function of v + 1/2 and fitting the data to a polynomial of
appropriate order. With 514.5 nm excitation, the data usually warrant a polynomial of order three or four. Extrapola
tion of the data points to their maximum yields D0, the
energy required to take molecular iodine from the lowest vibrational level of the ground state to the dissociation limit
(Figure 1).

Clarifications
In contrast to the B-state analysis, excellent quantitative
agreement is obtained between
experimental and literature

values of e, exe, and G (0) because the data begin at the

lowest vibrational level and there is no extrapolation to small
v. Calculated values of eye and eze show only qualitative agreement with literature values, but these terms are more
heavily influenced by the span of the data and the overall
order of the polynomial fit.

Determination of the X-state spectroscopic well depth D0

depends on a lengthy extrapolation. The last detected Ar+
514.5 nm LIF line typically occurs at 858 nm (v = 42), but
the highest bound X-state vibrational level is reported to be
v = 113 (24). Still, an extrapolation based on a fourth-order
polynomial fit to these data predicts the X-state vibrational
level spacing at high v well enough to match the literature
value of 12440.2 cm1 (25, 26) to within one to two percent.
With fewer
data points and a lower order polynomial fit,
though, D0 will be overestimated by 1000 cm1 or more.

Vol. 84 No. 8 August 2007

Journal of Chemical Education

1357

Research: Science and Education

Table 2. Summar y of X-State and B-State Potential Energy Parameters Determined by the Visible Absorption
and 514.5 nm Ar+ LIF Experiment
This Worka

Precisionb

107.11

107.28 0.27

Ar+ and HeNe LIF data yield a value precise to four significant
figures. Absorption spectrum hot band analysis is precise to three.

214.53

214.36 0.09

Same as G(0).

0.6130
1.03 x 104

0.603 0.008

Systematically low by 0.01 to 0.05 cm1.

(1.0 0.3) x 103

Qual

Highly dependent on order of polynomial fit.

1.43 x 104

(1.8 0.3) x 105

Qual

Most likely not statistically valid with HeNe LIF experiment.

D0/cm

12440.2

12440 110

D /cm1
e
1

Value for a fourth-order fit to eq 5. Lower-order fits will

overestimate the true value.

12547.3

12547 110

Accuracy and precision depend on D0.

0.037368

0.0372 0.0007

/cm1

Dependent on monochromator. Quoted precision is for 0.025 nm

resolution.

1.138 x 104

(1.17 0.04) x 104

See Supplemental Material.

266.64

267.2 2.5

Accuracy and precision depend on Be.

125.67

128 3

Systematic error due to extrapolation. Magnitude of error depends

on order of fit. Most likely to be 510 cm1 too high.

0.7504
4.14 x 103

0.79 0.15

Qual

(6 3) x 103

Qual

E*/cm1

20043.2

19981 25

Te/cm1

Systematic error due to extrapolation. Likely to be underestimated

by 50200 cm1.

15769.1

15757 26

Systematic error due to extrapolation dependent on order of fit.

4381.2

4330 40

Affected by systematic errors in E* and Te. Likely to be

underestimated by 50200 cm1.

4318.6

4270 40

Same as De.

302.67

298.7 2.7

See Supplemental Material.

0.029001

0.0298 0.0004

Calculated from value of Re.

7540 110

Affected by systematic errors in D0 and E*.

Quantity

Literature

Comments

X State (20, 25, 26)

G(0)/cm1

/cm1
e
x /cm1
e e
y /cm1
eze/cm1
e e 1

Be/cm
e

Re/pm

B State (20, 2426)

/cm1
e

x /cm1
e e
y /cm1
e

De/cm

D0/cm1
Re/pm

Be/cm1

Highly dependent on order of polynomial fit.

Highly dependent on order of polynomial fit.

Atomic Iodine (32)

E(I*)/cm1

7602.98

aLiterature values are compared with results from analysis of the experimental spectra shown in Figures 2 and 3. Both the absorption and the
fluorescence data were fit to fourth-order polynomials. Uncertainties are expressed at the 95% confidence level. Several B-state parameters are highly
dependent on the order of the fit. As an example, the value of e obtained from a third-order polynomial fit is 136 1 cm1 while the fourth-order fit
gives 128 3 cm1. bThis column accounts for both the propagated uncertainty and unavoidable systematic error. Qual indicates only order of
magnitude agreement can be expected.

Supplementary Analysis
Students can obtain the equilibrium internuclear distance
Re of ground-state molecular iodine from the Ar+ 514.5 nm
LIF spectrum by using rotational splitting
in the spectrum

to calculate the rotational constant, B . The 514.5 nm line

accesses several specific rovibronic transitions (20, 24, 27).
The 43 0 vibronic transition has already been mentioned,
but the rotational splitting associated with this transition is
difficult to resolve. The 514.5 nm line will also access a weak
hot band transition, v = 1 to v = 58, in which the rotational quantum number changes from J = 98 to J = 99
(28). Upon relaxation from the B state back to the X state,
the molecule ends up as either J = 98 again or J = 100, so
each vibronic line will be split into two. With minor theoretical approximations,
the energy
difference of these two lines
is equal to (4J + 6)B , or 398B . Thus if students
are able

to resolve the doublet, they can calculate B and then Re

since
h

B =
(6)
8 2 c R e 2
1358

Journal of Chemical Education

In eq 6, h is Plancks constant, c is the speed of light, and

is the reduced mass of 127I2.
The 58 1 rotational doublets are buried in most of the
LIF spectrum because the transition is about two orders of magnitude weaker than the 43 0 transition. However, since the
transition starts from v = 1, relaxation back down to v = 0
will release more energy than was absorbed, and one emission
doublet can be found to the blue of the lasing wavelength in a
relatively uncluttered region of the spectrum. The energy difference for this pair of peaks is 14.81 cm1, and the rotational
doublet is distinct at 0.25 nm resolution (inset in Figure 3).
If students have LIF instrumentation capable of 0.025 nm
resolution, they might also investigate one of the 43 0
fluorescence lines more closely. Each line in Figure 3 has substantial structure. The strongest portion of the signal is due
to an overlapped pair of closely spaced rotational doublets
because the Ar+ 514.5 nm line actually accesses two 43 0
rovibronic transitions: J = 15 to J = 16 and J = 13 to J = 12.
There is also low-lying structure in the signal that has been
attributed to changes in J caused by collisions prior to relaxation (21, 29).

Vol. 84 No. 8 August 2007

www.JCE.DivCHED.org

Research: Science and Education

Combining Absorption and Fluorescence Data

Basic Analysis
After collecting and analyzing fluorescence data, students
may
return to their absorption spectrum data and calculate
T e from the constant term in the polynomial fit of eq 2.
They may also calculate E(I*), the excitation energy corresponding to the lowest 2P12 2P32 transition of atomic
iodine (Figure 1):

E I* = E * D0

(7)

Supplementary Analysis
Reinforcement of how the energy of molecular iodine
changes during both the visible absorption experiment and
the laser-induced fluorescence experiment comes about by
having students plot approximate potential energy surfaces
for the X state and the B state. The simplest approximation
is a Morse potential
(30), for which students must provide

De, e, Re, and T e:

E = Te + De 1 e x

Supplemental Material

Expanded theoretical development, further clarification

on how to interpret the spectra, and a flowchart for students
showing the main analysis steps necessary to calculate model
potential energy curves from the collected spectra are available for the interested reader in this issue of JCE Online.
Literature Cited
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.

2 c
e
where x = R R e ; =

h De

(8)

Results from the Ar+ 514.5 nm LIF experiment give students all
the information they need to plot the X state as a Morse potential. Combining the visible absorption and LIF data yields all of
the B-state parameters except Re. Students can either be given
this value or else they can follow a method described by Dalterio
for estimating Re from the visible absorption spectrum (5).
Conclusion

12.
13.
14.
15.
16.

17.

18.

Exploration of the visible spectroscopy of molecular iodine provides students with a wonderful opportunity to test
many key quantum mechanical concepts they have learned
in physical chemistry. This article has presented guidelines
for molecular iodine spectral analysis to assist instructors and
students in reaching their experimental objectives correctly.
One major objective of this article has been to clearly identify the accuracy and precision in potential parameters that
students may expect to obtain from their measurements. Table
2 summarizes the results, listing recent primary literature values, the corresponding experimental values determined from
the spectra presented in this article, and brief commentary
on each one. Expanded theoretical development and further
clarification on how to interpret the spectra are available online in the Supplemental Material.W
Acknowledgments
The author thanks Karen Holman, Todd Silverstein, Sarah Kirk, and the reviewers for their helpful suggestions regarding this manuscript. Some instrumentation for these
experiments was purchased through the generous assistance
of a Camille and Henry Dreyfus New Faculty Start-Up Grant
for Undergraduate Institutions, and through the support of
the National Science Foundation under Grant No. 0116713.
www.JCE.DivCHED.org

19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.

Davies, M. J. Chem Educ. 1951, 28, 474477.

Stafford, F. E. J. Chem. Educ. 1962, 39, 626629.
Steinfeld, J. I. J. Chem. Educ. 1965, 42, 8587.
Hollenberg, J. L. J. Chem. Educ. 1970, 47, 214.
Dalterio, R.; Mattson, R.; Harris, R. J. Chem. Educ. 1974,,
51, 282284.
McNaught, I. J. J. Chem. Educ. 1980, 57, 101105.
Tellinghuisen, J. J. Chem. Educ. 1981, 58, 438441.
Cartwright, H. M. J. Chem. Educ. 1983, 60, 606.
Snadden, R. B. J. Chem. Educ. 1987, 64, 919921.
George, S.; Krishnamurthy, N. Am. J. Phys. 1989, 57, 850
853.
Lewis, E. L.; Palmer, C. W. P.; Cruickshank, J. L. Am. J. Phys.
1994, 62, 350356.
Lessinger, L. J. Chem. Educ. 1994, 71, 388391.
OBrien, L. C.; Kubicek, R. L. J. Chem. Educ. 1996, 73, 8687.
Muenter, J. S. J. Chem. Educ. 1996, 73, 576580.
Pursell, C. J.; Doezema, L. J. Chem. Educ. 1999, 76, 839
841.
Salzberg, H. W.; Morrow, J. I.; Cohen, S. R.; Green, M. E.
Physical Chemistry Laboratory: Principles and Experiments;
Macmillan: New York, 1978; pp 232237, 453455.
Sime, R. J. Physical Chemistry: Methods, Techniques, and Experiments; Saunders College Publishing: Philadelphia, 1990;
pp 660668.
Garland, C. W.; Nibler, J. W.; Shoemaker, D. P. Experiments
in Physical Chemistry, 7th ed.; McGraw-Hill: Boston, 2003;
pp 423433.
Long, G.; Sauder, D.; Shalhoub, G. M.; Stout, R.; Towns, M.
H.; Zielinski, T. J. J. Chem. Educ. 1999, 76, 841847.
Gerstenkorn, S.; Luc, P. J. Physique 1985, 46, 867881.
Steinfeld, J. I. Molecules and Radiation, 2nd ed.; MIT Press:
Cambridge, MA, 1985.
Gerstenkorn, S.; Luc, P. Atlas du Spectre dAbsorption de la Molecule dIode: 14,80020,000 cm1; CNRS: Paris, 1978.
NIST Atomic Spectra Database. http://physics.nist.gov/
PhysRefData/ASD/index.html (accessed May 2007).
Martin, F.; Bacis, R.; Churassy, S.; Vergs, J. J. Mol. Spectrosc.
1986, 116, 71100.
Gerstenkorn, S.; Luc, P. Laser Chem. 1983, 1, 83112.
Tromp, J. W.; Le Roy, R. J.; Gerstenkorn, S.; Luc, P. J. Mol.
Spectrosc. 1983, 100, 8294.
Gerstenkorn, S.; Luc, P. J. Mol. Spectrosc. 1979, 77, 310321.
Foth, H. J.; Spieweck, F. Chem. Phys. Lett. 1979, 65, 347352.
Kurzel, R. B.; Steinfeld, J. I.; Hatzenbuhler, D. A.; Leroi, G.
E. J. Chem. Phys. 1971, 55, 48224832.
Morse, P. M. Phys. Rev. 1929, 34, 5764.
Barrow, R. F.; Yee, K. K. J. Chem. Soc. Faraday II 1973, 69,
684700.
Luc-Koenig, E.; Morillon, C.; Vergs, J. Physica 1973, 70,
175189.

Vol. 84 No. 8 August 2007

Journal of Chemical Education

1359