Vibrational stark effects of nitriles I. Methods and experimental results

Article abstract of DOI:10.1021/jp002242r

The C-N stretch mode of several small nitriles was analyzed by measuring vibrational Stark effects. Conjugated and unconjugated nitriles as well as mono- and dinitriles were immobilized in frozen 2-methyl-tetrahydrofuran glass and examined in externally applied electric fields using FTIR. Room temperature absorption spectra of several nitriles (butyronitrile, hydrogen cyanide, valeronitrile, and deuterated aceonitrile) showed no concentration dependent effects between 2.4 mM and 0.8 M. Difference dipole moments, Δμ, equivalent to the linear Stark tuning rate, were 0.01/f-0.04/f Debye (0.2/f-0.7/f/cm(/MV/cm)) for most samples, with aromatic compounds toward the high end and symmetric dinitriles toward the low end. Most quadratic Stark effects were small and negative, while transition polarizabilities were positive and had a significant effect on Stark line shapes. For aromatic nitriles, transition dipoles and Δμ values agreed with Hammett numbers. Symmetric dinitrile Δμ values decreased monotonically with increasing conjugation of the connecting bridge, likely due to improved mechanical coupling and, to a lesser extent, an increased population of inversion symmetric conformations. Δμ values were close to those expected from bond anharmonicity and ab initio predictions.

Full text of DOI:10.1021/jp002242r

J. Phys. Chem. A 2000, 104, 11853-11863
11853
Vibrational Stark Effects of Nitriles I. Methods and Experimental Results
Steven S. Andrews and Steven G. Boxer*
Department of Chemistry, Stanford UniVersity, Stanford, CA 94305-5080
ReceiVed: June 21, 2000; In Final Form: October 19, 2000
Vibrational Stark effects, which are the effects of electric fields on vibrational spectra, were measured for the
C-N stretch mode of several small nitriles. Samples included unconjugated and conjugated nitriles, and
mono- and dinitriles. They were immobilized in frozen 2-methyl-tetrahydrofuran glass and analyzed in
externally applied electric fields using an FTIR; details of the methodology are presented. Difference dipole
moments, ∆µ, equivalent to the linear Stark tuning rate, range from 0.01/f to 0.04/f Debye (0.2/f to 0.7/f
cm-1/(MV/cm)) for most samples, with aromatic compounds toward the high end and symmetric dinitriles
toward the low end (the local field correction factor, f, is expected to be similar for all these samples). Most
quadratic Stark effects are small and negative, while transition polarizabilities are positive and have a significant
effect on Stark line shapes. For aromatic nitriles, transition dipoles and ∆µ values correlate with Hammett
numbers. Symmetric dinitrile ∆µ values decrease monotonically with increasing conjugation of the connecting
bridge, likely due to improved mechanical coupling and, to a lesser extent, an increased population of inversion
symmetric conformations. ∆µ values are close to those expected from bond anharmonicity and ab initio
predictions.
Introduction
small nitriles. Due to the paucity of work in vibrational Stark
spectroscopy, we decided to survey a wide variety of nitriles
When a weak electric field is applied across a molecule, both
the molecular vibrational energy levels and the transition dipoles
between the levels change slightly. These changes affect the
infrared absorption spectrum of the sample, and are collectively
called the vibrational Stark effect (VSE). VSE measurements
yield information on the physics of molecular vibrations,
including the anharmonicity of vibrational modes and the
perturbation of chemical bonds by electric fields. VSE measure-
ments can also be used to calibrate the sensitivity of a vibrational
frequency to an electric field, called the Stark tuning rate, which
allows an environmentally induced vibrational band shift to be
used as a local probe of electrostatic fields.1,2 Different
vibrational modes have widely varying Stark effects, so a VSE
spectrum can sometimes be used to identify a weak vibrational
band of interest from a collection of overlapping background
bands.2
Strong local electric fields are common in condensed matter
systems, including the solvation of charged or dipolar mol-
ecules,3 in ionic matrices, at electrode surfaces, across biological
membranes,4 and in the organized environment of proteins.5,6
These local fields can be very large, typically lying in the range
of 0.1 to 10 MV/cm. Vibrational band shifts have been studied
in all these systems,7-10 but in most cases neither the electric
field nor the Stark tuning rate of the probe molecule was known
accurately. We have recently shown that vibrational Stark
spectroscopy is a useful tool for understanding vibrational band
shifts for diatomic ligands bound to the heme iron in myoglo-
bin.1,2 We are also preparing to use vibrational Stark effects to
calibrate the sensitivity of vibrational frequencies of C-N
containing unnatural amino acids, such as cyano-phenylalanine
and cyano-alanine, to determine fields in proteins. VSE will
likely prove useful for understanding and predicting band shifts
in a wide variety of other systems as well.
including aromatic and aliphatic compounds, several homolo-
gous series of nitriles, and nitriles with varying degrees of
symmetry. In the course of the work, it was found that many
of the molecules studied, including several very simple ones,
have multiple absorption bands in the nitrile frequency region.
Where possible, the assignment of these bands is presented
below, both to characterize the system being studied and because
it influences the interpretation of the Stark data. Along with
methods development, a primary objective of this work is to
yield an understanding of the physical origins of vibrational
Stark effects, which is introduced below and described more
fully in a separate publication.11
The nitrile stretch mode was chosen for analysis for several
reasons. The frequency is high enough to be well separated from
most other vibrational modes and is in a spectral region that is
relatively easy to detect, as it occurs near the maximum
detectivity of an indium antimonide infrared detector. Bands
are generally intense and narrow, properties which contribute
to a strong Stark effect. A normal-mode analysis of organic
nitriles shows that the nitrile stretch mode is highly localized
to the nitrile bond,12 which both simplifies the interpretation of
VSE data and implies that changes to the rest of the molecule
minimally perturb the nitrile normal mode. Finally, there is a
great deal of experimental and theoretical literature on nitrile
vibrations.13-19
Theory
A vibrational Stark analysis requires an absorption spectrum
of the sample, A(νj), and a Stark difference spectrum, ∆Aø(νj),
which is the absorption in the presence of an external field,
minus the absorption without the field. ø is the angle between
the light polarization vector and the direction of the applied
electric field, and can be varied from 90 to about 45 degrees
using the experimental setup described in the next section. As
In this paper, a sensitive VSE measurement method is
presented and used to analyze the C-N stretch mode of several
10.1021/jp002242r CCC: $19.00 © 2000 American Chemical Society
Published on Web 11/30/2000

11854 J. Phys. Chem. A, Vol. 104, No. 51, 2000
Andrews and Boxer
2
2
Cø ) 5|∆µ| + (3cos2 ø - 1)(3cos2 ú - 1)|∆µ|
(6)
explained below, the absorption and Stark difference spectra
are fit to give the Stark fit coefficients, from which are computed
the Stark parameters of the sample.
Stark Parameters. For an absorption band, Stark effects are
expected to yield a slight peak shift for each molecule in a
sample, without any line shape change, where the shift is given
by
In eq 6, ú is the angle between M and ∆µ. There are also third
derivative and higher order terms in the derivative expansion,
but they are all proportional to the fourth or higher powers of
the field, a dependence which is expected to be small and which
was not seen in the data. The derivative coefficients were found
from experimental data by fitting a measured Stark difference
spectrum with a sum of the frequency-weighted derivatives of
the absorption spectrum using software developed for this
purpose. The fitting procedure is described below.
1
hc
1
2
∆νj ) -
∆µ‚F + F‚∆r‚F + ‚‚‚
(1)
(
)
∆µ and ∆r are called the difference dipole moment and the
difference polarizability, respectively, referring to the differences
between ground and excited states. These names are not entirely
accurate for VSE due to field-induced bond changes,20,11 but
the variables are still useful and completely general. An electric
field also affects the transition dipole moment, M(F), through
the transition polarizability, A, and the transition hyperpolar-
izability, B
As the ø dependence in eqs 4-6 is restricted to a linear
dependence on the term 3cos2ø - 1, only 6 scalar values of the
48 tensor components of the Stark parameters can be determined
from Stark measurements, so a number of assumptions are
needed to simplify the problem. The transition hyperpolariz-
ability, B, is assumed to be negligible, as it has been calculated
to contribute at most a few percent to Aø.22 It is simplest to
consider the molecular z axis to be parallel to the transition
dipole moment, M. In this system, ∆Rzz and Azz are parallel
M(F) ) M + A‚F + F‚B‚F + ‚‚‚
(2)
polarizability components, represented by ∆R and A , and ∆Rxx,
||
||
∆Ryy, Axx, and Ayy are the perpendicular polarizability compo-
nents, represented by ∆R and A , using the assumption that
the perpendicular components are equal on the two axes. The
off-diagonal components, representing polarizabilities perpen-
dicular to an applied field, are assumed to be zero. With these
simplifications, the six remaining Stark parameters are |∆µ|,
Again, the variables A and B are useful and general but the
names are not entirely accurate for VSE.
As the samples studied in this work were immobilized in a
frozen glass, the molecular orientations were isotropic (however,
field-induced poling can be a problem and is discussed below).
Integrating the peak shifts and extinction coefficient changes
given in eqs 1 and 2 over all orientations, combined with the
sample absorption spectrum, yields a computed Stark spectrum.
This computed spectrum is a function of the two field-
independent parameters, νj and M, and the four Stark parameters,
∆µ, ∆r, A, and B.
Stark Fit Coefficients. With the assumption that the absorp-
tion line shape is unaffected by the applied electric field, the
computed Stark spectrum can be expressed as a sum of
derivatives of the absorption spectrum21
∆R , A , ú, ∆R , and A , which can be found from experimental
||
||
data.
Further assumptions about the Stark parameters were made
for data where the ø dependence of the Stark difference spectrum
was not measured. Here, we assumed that a nitrile or dinitrile
vibration behaves as a one-dimensional oscillator, with the result
that ú, ∆R , and A are assumed to be zero. This assumption
of a one-dimensional system is expected to be exact for nitriles
with rotational symmetry about the nitrile bond (ignoring solvent
induced asymmetry); it was also shown to be reasonably good
for samples where the ø dependence was measured, described
below.
Due to the bulk dielectric properties of the solvent as well as
specific field-induced perturbations of the solvent near the
sample molecules, the electric field at a sample molecule is
greater than the average electric field (total applied voltage
divided by electrode spacing). This difference is corrected for
using the local field correction factor, f, which is estimated to
be between 1.1 and 1.3,23,24
2
A(νj)
|F|
∂
2
∆Ax(νj) ) |F| AxA(νj) +
Bøνj
+
15hc
∂νj νj
|F|
2
2
A(νj)
νj
∂
Cøνj
(3)
30h2c2
∂νj2
F is the electric field vector and Aø, Bø, and Cø, which are called
the zeroth, first, and second derivative coefficients, respectively,
are found from eqs 1 and 2 to be
Flocal ) f Fexternal
(7)
1
3cos2ø - 1
Ax )
Tr(ATA) +
(TrA)2 + Tr(A2) -
[
2
2
Whereas the exact value of f is not known, it is expected to be
the same constant for all experiments in this work, as well as
for a great deal of other work, allowing for meaningful
comparisons. Due to the lack of knowledge of this factor, results
are reported in terms of factors of f.
3|M|
10|M|
2
3
2
Tr(ATA) +
M B +
∑
i
ijj
]
2
i,j
3|M|
3cos2ø - 1
2
M B + M B - MiBijj (4)
∑
i
jij
i jji
(
)
2
Experimental
3
i,j
5|M|
Samples were solutions of small aliphatic and aromatic nitriles
in the glass forming solvent, 2-methyl-tetrahydrofuran (2-
MeTHF). VSE sample concentrations ranged from 0.1 to 0.3
M, and no concentration dependent effects were observed
throughout this range in either absorption or Stark spectra,
indicating that nitrile dimers18,25 did not form to a measurable
extent. In support of this assertion, room temperature absorption
spectra of several nitriles (butyronitrile, hydrogen cyanide,
3(3cos2 ø - 1)
10
Bx )
2 M‚A‚∆µ +
∆µ‚A‚M +
(
2
|M|
|M|
2
5
M‚∆µ TrA - M‚A‚∆µ + Tr∆r +
)
3
2
(3cos2 ø - 1)
3
2
1
M‚∆r‚M - M 2Tr∆r (5)
(
|
|
)
2
2
|M|

Vibrational Stark Effects of Nitriles
J. Phys. Chem. A, Vol. 104, No. 51, 2000 11855
valeronitrile, and deuterated acetonitrile) also showed no
concentration dependent effects for concentrations between 2.4
mM and 0.8 M. Low temperature extinction coefficients were
measured by taking spectra of known sample concentrations
with a setup that was essentially identical to that used for Stark
measurements. The 2-MeTHF solvent and all nitriles, except
for hydrogen cyanide, were purchased from Aldrich and used
without further purification.
Using this method, a single scan is made with the field turned
off, then a scan with 2.5 kilovolts applied to the sample in one
direction (about 0.9 MV/cm electric field), then another field-
off scan, and finally a 2.5 kilovolt potential in the opposite
direction. The alternating field pattern was chosen to avoid
charge build-up in the sample and to prevent sample poling;
both field-on scans gave identical spectra and were averaged
together. This cycle was repeated at least 512 times, which is
about 1 h, for each VSE measurement. At the end, the field-on
and field-off interferograms were averaged and Fourier trans-
formed; the ratio of their spectra were used to compute a Stark
spectrum. The field-off spectrum, relative to background scans
with the sample out of the light path, provided a good quality
absorption spectrum with exactly the same conditions as the
Stark spectrum. Sample molecules did not change their orienta-
tions due to the applied electric field in the immersion cryostat
to any measurable extent, as determined by (i) the lack of
differences between positive and negative field results after a
constant positive field had been applied for 15 min or longer
and (ii) the lack of a difference between sequential field-on
scans, using a different electric field pattern.30
For the synthesis of hydrogen cyanide (HCN), equal volumes
of 2-MeTHF and a 0.3 M aqueous solution of potassium cyanide
were put in a vial, forming an organic layer on top of an aqueous
layer. Dropwise addition of concentrated HCl formed the HCN
product which partitioned into the organic layer, from which
the sample was taken for spectroscopy. The partition coefficient
(organic/aqueous) is 4.62, determined from room-temperature
infrared peak intensities of HCN in the aqueous layer in contact
with various volumes of 2-MeTHF. Using the partition coef-
ficient, a known concentration of HCN was dissolved in the
organic layer and used to determine the low-temperature
extinction coefficient of HCN. It was shown with control
experiments that no CN- partitioned into the organic layer and
that all measurable aqueous CN- reacts to form HCN in the
presence of excess acid. The room-temperature peaks analyzed
here are aqueous CN- at 2080 cm-1, aqueous HCN at 2093
Data Analysis
Fits were made with at least one of three methods.31 Where
the absorption spectrum did not have overlapping bands, the
numerical derivatives of the measured spectrum were used so
as to eliminate errors due to naturally asymmetric line shapes
or any consistent measurement error, such as light scatter or
imperfect phase correction of the interferogram. This fitting
method introduces the least error but cannot be used to determine
separate Stark parameters for overlapping bands. For samples
with overlapping bands, and for several samples with single
bands in order the check the reliability of the method, the
absorption spectra were fit first with a modified Voigt line shape
for each peak. The form of a modified Voigt line shape is a
linear combination of a Lorentzian and a Gaussian, where the
Lorentzian and Gaussian are constrained to have the same peak
positions and the same full width at half-maximum.32 It is similar
to a Voigt line shape, a convolution of a Lorentzian and
Gaussian, and has similar versatility, but is easier to fit. The
analytical derivatives of the separate peaks were then used to
fit the VSE data, providing separate fit coefficients for each
peak. The third method used a simultaneous fit of the absorption
spectrum with modified Voigt line shapes, and of the Stark
spectrum with the analytical derivatives of these line shapes.
The sum of the two fit errors was minimized, where each least-
squares error was weighted by the respective spectral rms noise,
to use all the spectral information available for all the fit
parameters. The simultaneous fitting method is less sensitive
to noise in the absorption spectrum, but more sensitive to noise
in the Stark data. Where multiple methods were carried out on
the same set of spectra, fit results were identical, within statistical
error.
cm-1, and organic HCN at 2079 cm-1
.
The infrared spectroscopy cell consisted of a pair of 13 mm
diameter by 1 mm thick calcium fluoride windows (Janos
Technology Inc.) with approximately 50 Å of nickel vacuum
deposited on the surfaces facing the sample, a design which is
similar to that used in previous work26 and in electronic Stark
spectroscopy.24 A layer of nickel typically transmits 50 to 80%
of the incident light throughout the infrared region, with the
rest being either absorbed or reflected, and has a resistance of
about 5 kΩ across the face of a window. The nickel electrodes
were connected to a high voltage DC power supply (Trek
Instruments Inc.), whose output voltage was synchronized to
the FTIR scan timing, described below, with a home-built
control unit. The windows were separated from each other by
a pair of 28 micron thick Teflon spacers and held in place with
a metal clamp. Using interference contour fringes, the windows
were made parallel to within 1° to ensure both a uniform optical
path length and a uniform field in the cell. The window
separation in the center of the cell was measured to ( 0.05
micron after each experiment by taking a visible spectrum of
the empty cell and measuring the frequency spacing between
spectral interference fringes. After a sample was loaded into
the cell by capillary uptake, it was rapidly frozen by immersing
the cell in liquid nitrogen. This formed a transparent glass
without observable cracks, crystallization, or other defects. The
nitrogen immersion dewar is a custom built optical cryostat with
several novel features designed to avoid bubbles and schlieren.27
Spectroscopy was carried out on a Bruker IFS 66V/S FTIR,
with a nitrogen cooled indium antimonide detector and 0.5 cm-1
resolution. We first tried to record Stark spectra with the FTIR
in step-scan mode, by locking into the second harmonic of a
200 hertz electric field applied to the sample. This is analogous
to the standard method for visible and near-infrared Stark
spectroscopy24 and to a previous study of VSE26 in which
dispersive optics were used. Step-scan with lock-in detection
has been used advantageously for measuring vibrational circular
dichroism.28 However, the noise reduction achieved by lock-in
detection was more than offset by the noise gained by using
step-scan mode. Better signal-to-noise was obtained using a DC
electric field with the FTIR in its normal rapid-scan mode.29
To achieve a good signal-to-noise ratio, most data were
obtained only at the experimental angle ø at 90°, where
unpolarized light can be used and less of the infrared beam is
clipped by the sides of the sample. These data were interpreted
using the one-dimensional assumptions, described above. Fits
to these data constrained Aø values to be nonnegative, as
negative values are impossible for a one-dimensional system
and for a negligible transition hyperpolarizability (eq 4). This
constraint was only required for data sets of poorer quality,
suggesting that those Aø values were dominated by noise and
that the constraint is appropriate.

11856 J. Phys. Chem. A, Vol. 104, No. 51, 2000
Andrews and Boxer
In some situations, strong interference fringes were observed
at normal incidence due to reflections between the sample cell
windows, which were reduced by rotating the sample away from
the normal by up to 20° (still using unpolarized light). Those
data were corrected to normal incidence, again with the
assumption that the nitrile Stark effects follow one-dimensional
behavior. The corrections are relatively insensitive to the
assumptions for small rotation angles and unpolarized light.33
Despite rotating some samples, Stark spectra often showed a
nonzero baseline, typically dominated by spectral interference
fringes (about 130 cm-1 period), but also sometimes including
a uniform offset or slope. To prevent the baseline from affecting
the fit parameters, it was fit along with the rest of the Stark
spectrum using a linear combination of a sine wave and a
straight line. This did not interfere with the rest of the fitting
because the baseline extended well beyond the nitrile peaks,
and the functions used to fit the baseline yield very different
curve shapes than the functions used to fit the VSE peaks.
The magnitudes of the transition dipoles were found from
the area of the absorption band, expressed as an extinction
coefficient spectrum, ꢀ(νj)34
3ꢀ0hcln10
|M| )
ꢀ(νj)dνj
∫
(8)
2π2N νj
x
A
Several samples showed multiple peaks in the nitrile stretch
region. For situations where the peaks were determined to be
due to multiple vibrational modes, the peak areas were integrated
separately using fits with modified Voigt line shapes (these
samples include all dinitriles with multiple peaks and deuterated
acetonitrile). On the other hand, where multiple peaks were
determined to be due to either heterogeneity or Fermi resonance,
the integration was carried out over all the peaks in order to
include the entire sample and all the intensity of the nitrile mode
(these samples include HCN and other aliphatic nitriles). Implicit
in the latter method is the assumption that the transition dipole
moments have the same magnitude for all nitrile stretches in a
heterogeneous sample.
Figure 1. Spectra of acetonitrile. These and subsequent spectra are of
samples dissolved in 2-MeTHF, at 74 K, and at normal incidence to
the infrared beam. (A) Extinction coefficient spectrum of the nitrile
stretch mode; the line is not a fit, but is simply shown to guide the
eye. (B) Dots represent the Stark data, the solid line is the best fit to
the Stark spectrum, using numerical derivatives of the data in panel A,
and the dashed line is the Stark baseline. In this and subsequent figures,
Stark spectra are scaled for a 1 MV/cm applied electric field. (C) The
solid line is the fit from panel B, with the baseline subtracted, and the
dashed lines are the zeroth, first, and second derivative contributions.
Results
From the second and zeroth derivative contributions, respec-
Aliphatic Nitriles. The nitrile stretch mode of acetonitrile
gives rise to an intense and narrow absorption band (Figure 1A).
There is also a weak Fermi resonant peak 50 cm-1 to higher
energy (not shown; peak maximum is at 2297.1 cm-1, intensity
is 6.90 M-1cm-1, fwhm is 20.1 cm-1), which arises from
resonance between the nitrile mode and the combination mode
of the C-C stretch and the symmetric C-H bend.8 To facilitate
comparison with other spectra, Figure 1 and similar figures in
this paper are scaled to show spectral extinction coefficients
and to show Stark spectra for an applied field of 1 MV/cm (eq
3). The unscaled acetonitrile data, which are fairly typical, had
a maximum absorption of 0.21 o.d., a minimum Stark absorp-
tion, relative to the baseline, of -4.4 × 10-4 o.d. using a 0.88
MV/cm electric field, and Stark rms noise levels of 5.2 × 10-6
o.d.
As seen in Figure 1C, the VSE spectrum is dominated by
the second derivative of the absorption spectrum, but also shows
significant zeroth and first derivative components (Table 1), all
of which are highly reproducible. The ø angle dependence of
the Stark spectrum was studied as well and is discussed below;
however, it is ignored here and in Table 1, using the one-
dimensional assumptions, to allow for a meaningful comparison
with other nitriles where the ø dependence was not measured.
tively, |∆µ| is 0.0258 ( 0.0004 D/f and A is 0.53 ( 0.03 Å3/f,
||
which imply a 0.43 cm-1 shift and a 6.1% increase of band
intensity for molecules aligned parallel to a 1 MV/cm electric
field (errors represent 1 standard deviation for 5 data sets). As
A is more than sufficient to account for the entire first derivative
||
contribution, ∆R is found to be slightly negative, with a value
||
of -0.81 ( 0.07 Å3/f , implying a band shift of 0.02 cm-1 to
2
higher energy for molecules aligned parallel to a 1 MV/cm field.
Due to the large width and low intensity of the resonant band,
as well as its overlap with infrared absorption from liquid
nitrogen in the immersion cryostat,35,27 its Stark effect could
not be measured reliably. However, its |∆µ| value appears to
be between 0.01 and 0.05 D/f, putting it in the same range as
the nitrile band Stark effect.
Deuterating acetonitrile lowers the C-H frequencies, which
moves the Fermi resonant band well below the nitrile mode
and moves the C-H stretch frequency to just below the nitrile
mode (Figure 2 A, B). The two peaks were determined to have
different Stark characteristics by the fact that derivatives of the
entire absorption spectrum cannot be satisfactorily fit to the Stark
spectrum. This required a separation of the peaks to allow
independent fitting. However, the absorption spectrum cannot
be fit with two modified Voigt line shapes due to the relatively

Vibrational Stark Effects of Nitriles
J. Phys. Chem. A, Vol. 104, No. 51, 2000 11857
TABLE 1: Vibrational Stark Effect Data for Unconjugated Aliphatic Nitriles
absorption data
Stark data
b
peak area
M-1 cm-2
|M|
q
e
Aø
Bø
Cø
|∆µ|
D/f
∆R|b,c
Å3/f 2
A|b,c
compound
νja cm-1
D
1020 m2/V2
1040 Jm2/V2
1063 J2m2/V2
Å3/f
HCN Ld
2054.9
2064.7
2072.4
2247.5
2243.5
2247.0
2239.2
2243.2
2255.5
2234.5
3989
3989
3989
830
778
813
1024
793
974
0.134
0.134
0.134
0.058
0.056
0.058
0.065
0.057
0.063
0.061
0.79
0.79
0.79
0.36
0.35
0.35
0.40
0.35
0.37
0.39
15.8
0
0
1.8
1.9
1.9
0.5
0.5
2.1
7.2
2.23
-2.92
0.43
0.22
0.52
0.92
0.87
-0.03
-0.17
1.77
28.89
20.12
14.89
5.56
6.42
6.69
5.95
6.55
6.30
14.16
0.0588
0.0491
0.0422
0.0258
0.0277
0.0283
0.0267
0.0280
0.0275
0.0412
-4.93
-1.76
0.26
-0.81
-0.71
-0.48
0.14
-0.52
-1.14
-1.89
3.55
0
0
0.53
0.52
0.53
0.23
0.25
1.09
0.60
HCN Md
HCN Hd
acetonitrilee
propionitrile
butyronitrilee
valeronitrilee
hexanenitrilee
acetonitrile-d3
acetonitrile-d3 (C-D)f
904
a Peak maximum. b Assumes one-dimensional behavior, i.e. ú)0°, and off-diagonal and perpendicular components of ∆r and A are 0. c B is
assumed to be 0. d Peak areas for HCN low (L), medium (M), and high (H) energy nitrile absorptions are combined. e Fermi resonant peaks are
observed, and their areas are combined with the nitrile peak area. f C-D stretch mode of deuterated acetonitrile.
Figure 2. Spectra of CD3CN, butyronitrile, and HCN. (A) Dots show the extinction coefficient spectrum for CD3CN, showing a C-D stretch
mode at 2235 cm-1 and the C-N stretch mode at 2256 cm-1. The solid line is a sum of two modified Voigt line shapes, demonstrating the resonant
absorption between the bands; these line shapes were used to separate the measured absorbance into two bands (see text). (B) Dots represent the
CD3CN Stark spectrum, the solid line shows the best fit to the Stark data, using different parameters for the two bands, and the dashed line shows
the Stark baseline. (C, D) Extinction coefficient and Stark spectra for butyronitrile, where the two peaks are interpreted as being due to Fermi
resonance. The line in panel C is shown to guide the eye, whereas the line in panel D is a fit to the Stark spectrum using numerical derivatives of
the absorption spectrum, fit with a single set of coefficients. (E, F) Extinction coefficient and Stark spectra for HCN, where the three peaks are
interpreted to result from heterogeneous hydrogen bonding to the solvent. Solid lines show fits that were made simultaneously to the two spectra,
using modified Voigts for the extinction coefficient spectrum and their analytical derivatives for the Stark spectrum. Dashed lines in panel E show
the three modified Voigt line shapes used to fit the peak, and the dashed line in panel F is the Stark baseline.
large absorption between the peaks, indicating the presence of
nonlinear coupling between the C-N and C-D stretching
modes.36 A theory has not been developed yet for Stark effects
of resonant modes, so an interpolation method was used to
separate the peaks. It yielded a good fit to the data despite its
lack of a physical basis. Two modified Voigt line shapes, called
FitL and FitH, were fit to the low and high energy peaks of the
absorption spectrum, respectively, while ignoring the fit error
of the region between the peaks (2237 to 2250 cm-1). The ratios
of the fits were used to separate the experimental spectrum, A,
yielding separate spectra for the two peaks, AL and AH (these
are all functions of νj)
good quality of the fit away from the region between the bands
suggests that the fit parameters are likely to be representative
of the isolated normal modes, and are thus comparable to
parameters for other nitriles. Not surprisingly, the nitrile VSE
parameters for deuterated acetonitrile (Table 1) are similar to
those for protonated acetonitrile. For the C-D stretch mode,
|∆µ| is 0.0412 D/f, which is about 50% larger than the nitrile
difference dipole.
A homologous series of aliphatic nitriles was studied, from
hydrogen cyanide to hexanenitrile. With the exception of HCN
(discussed below), the transition dipoles and difference dipoles
are nearly identical across the series (Table 1). Other Stark
parameters vary some, but with no clear trends and with less
magnitude than the variation between aliphatic and other nitriles.
The molecular polarizability changes by about a factor of 3 from
acetonitrile to hexanenitrile,37 suggesting that the molecular
polarizability has a minimal influence on vibrational Stark
effects for nitriles.
A‚FitL
FitL + FitH
A‚FitH
FitL + FitH
AL )
AH )
(9)
As AL + AH ) A, all the sample absorption is included without
requiring the use of additional fitted peaks; the interpolation
method also preserves the natural peak asymmetry. Derivatives
of AL and AH were then fit to the Stark data (Figure 2B). The
Lengthening the hydrocarbon chain to two units in propio-
nitrile results in an absorption spectrum with a single peak and

11858 J. Phys. Chem. A, Vol. 104, No. 51, 2000
Andrews and Boxer
no discernible Fermi resonance. However, on adding one, two,
or three more CH2 units to the hydrocarbon chain, giving
butyronitrile, valeronitrile and hexanenitrile, respectively, the
absorption spectra show a pair of peaks (Figure 2C shows
butyronitrile). In all cases, the peaks fit well with a pair of
modified Voigt line shapes, without any extra absorption
between the peaks. Using the fits, the peak separations are 9.7,
13.0, and 13.0 cm-1 for the respective nitriles, while the ratios
of the areas of the lower and higher frequency peaks are 0.22:
1, 1.38:1, and 0.006:1 (all at 74 K). Possible assignments for
the multiple peaks include Fermi resonant interactions as seen
for acetonitrile, heterogeneous sample conformations involving
anti and gauche conformations, partial dimer formation in the
sample, solvent heterogeneity, or a combination of these factors.
These assignments do not affect the analysis of VSE data,
although the interpretation varies considerably. From a variety
of experiments, all of the possible assignments involving simple
heterogeneity can be eliminated: (i) spectra are independent of
sample concentration over a wide range of concentrations at
both room temperature and at 74 K, indicating that dimer
formation does not occur at the concentrations used; (ii) the
relative areas of the peaks are different for different compounds,
and none agree with the Boltzmann population calculations that
the ratio of gauche:anti populations are expected to be about
0.07:1 in the frozen samples or 0.6:1 at room temperature;38
(iii) the peak separations are much larger than those between
homologous nitriles or than those calculated with AM1 methods
for gauche and anti conformations (<1 cm-1), also arguing
against assignment based on conformational heterogeneity; (iV)
room temperature spectra show multiple peaks and the solvent
phase transition is not apparent in temperature dependent spectra,
suggesting that solvation is not primarily responsible for the
multiple peaks; and (V) room temperature spectra in toluene are
very similar to those in 2-MeTHF, indicating again that specific
solvation effects are not important. Thus, simple heterogeneity
is unlikely, and the multiple peaks are almost certainly due, at
least in part, to Fermi resonance. However, it remains surprising
that Fermi resonance would yield significantly different spectra
for each of the aliphatic nitriles studied. It is also surprising
that the valeronitrile absorption spectrum shows a strong
temperature dependence, whereas the other nitriles show a weak
temperature dependence (data not shown).
Figure 3. Spectra of 4-chloro-benzonitrile. (A) Extinction coefficient
spectrum, where the line is shown to guide the eye. (B) Dots represent
the Stark spectrum, the solid line is the best fit to the Stark spectrum
using numerical derivatives from panel A, and the dashed line is the
Stark baseline. (C) The solid line is the fit from panel B with the
baseline subtracted, and the dashed lines are the zeroth, first, and second
derivative contributions.
overlapping modified Voigt line shapes, characterized by
different Stark parameters. Due to the simplicity of HCN, there
are no resonant modes or alternate conformations; consequently,
the three peaks most likely represent heterogeneous hydrogen
bonding from the sample to the 2-MeTHF solvent. This
interaction, which cannot occur for the other samples, would
be expected to significantly affect the nitrile bond. In agreement
with this assignment, the spectrum of HCN exhibits a single
broad peak when it is dissolved in 2-MeTHF at room temper-
ature, and a single sharp peak when in a toluene glass at low
temperature. The transition dipoles were considered to be the
same for the three heterogeneous peaks, despite the fact that
this is likely to be a poor assumption, because they cannot be
determined independently. It was found that both the transition
dipole moment and the difference dipole moments of HCN are
much larger than they are for the other aliphatic nitriles.
Aromatic Nitriles. With a few exceptions, aromatic nitrile
absorption spectra show a single strong band in the nitrile stretch
region. A notable exception is the symmetric compound 1,4-
dicyanobenzene, which is discussed below with the other
symmetric dinitriles. Also, benzonitrile and 2-chloro-benzonitrile
show weak absorptions at 16 and 35 cm-1 higher frequency,
respectively, than the strong nitrile band. The Stark spectra of
the small peaks fit well with the same derivative components
as the major peaks, but were too small for a thorough analysis.
The spectrum of 4-chloro-benzonitrile is shown in Figure 3
As the intensity of the multiple peaks for butyronitrile,
valeronitrile, and hexanenitrile is expected to arise from the
nitrile mode (regardless of assignment), the entire band shapes
were integrated to find the transition dipole moments. From VSE
spectra (Figure 1D shows butyronitrile), it was found that the
Stark parameters of the pairs of peaks are identical within
measurement error, so they are reported in Table 1 with one
value for each nitrile. Assuming the Fermi resonance assignment
is correct, it is found that the Stark parameters of these Fermi
resonant bands are identical to those of the parent bands,
suggesting that VSE parameters represent the Stark effects of
the source of the absorption intensity, rather than those of the
forbidden mode. In contrast, recent work on the VSE of nitric
oxide bound to the heme iron in myoglobin showed a case where
Fermi resonance affects the Stark effect of the parent band.2 In
general, Stark effects of Fermi resonances are not well
understood.
The simplest member of the aliphatic series, hydrogen
cyanide, does not behave like the other aliphatic nitriles. Its
absorption band is at lower energy, is broader, and is about 4
times more intense (Figure 2E, F). The simplest fit that can be
made to the absorption and Stark spectra requires three

Vibrational Stark Effects of Nitriles
J. Phys. Chem. A, Vol. 104, No. 51, 2000 11859
TABLE 2: Vibrational Stark Effect Data for Aromatic Nitriles
absorption data
Stark data
b
peak area
M-1 cm-2
|M|
q
e
Aø
Bø
Cø
|∆µ|
D/f
∆R|b,c
Å3/f 2
A|b,c
compound
benzonitrile
νja cm-1
D
1020 m2/V2
1040 Jm2/V2
1063 J2m2/V2
Å3/f
2227.8
2235.3
2236.3
2231.7
2243.7
2232.1
2232.7
2230.6
2223.5
2083
819
868
862
138
1205
1260
1921
3724
0.093
0.058
0.060
0.060
0.024
0.070
0.072
0.089
0.124
0.57
0.36
0.37
0.37
0.15
0.43
0.44
0.55
0.76
2.6
1.4
5.5
2.1
4.9
3.4
2.9
2.7
1.8
1.52
2.01
2.23
2.76
1.95
1.94
2.04
2.10
1.33
10.86
5.80
9.08
3.20
1.30
7.81
8.45
10.10
20.60
0.0361
0.0263
0.0330
0.0196
0.0125
0.0306
0.0318
0.0348
0.0497
-0.65
0.60
-0.71
0.89
1.00
0.32
0.92
0.58
0.35
0.87
0.82
0.97
0.49
1,2-dicyanobenzened
1,3-dicyanobenzened
1,4-dicyanobenzene (A)d,e
1,4-dicyanobenzene (S)d,f
2-Cl-benzonitrile
0.43
-0.35
-0.23
-0.26
-0.02
3-Cl-benzonitrile
4-Cl-benzonitrile
4-methoxybenzonitrile
a-c See the corresponding notes in Table 1. d Peak area and transition dipole are divided by two, to give results on a per nitrile basis. e Antisymmetric
(A) nitrile mode. f Symmetric (S) nitrile mode.
and its Stark parameters are listed in Table 2 along with those
of other aromatic nitriles. As with the aliphatic nitriles, fits to
the aromatic Stark spectra are dominated by the second
derivative contribution, and the source of the Stark effect is
dominated by the difference dipole moment. However, aromatic
nitrile VSE spectra exhibit slightly larger zeroth derivative
contributions and much larger first derivative contributions,
leading to significantly different line shapes. Upon analysis, it
is seen that the large first derivative contribution is due to larger
difference dipoles and larger transition polarizabilities, rather
than to a significant change in difference polarizabilities (Table
2), which are again small and slightly negative. This result
contrasts with our earlier work on 4-methoxy-benzonitrile in
which it was concluded that the zeroth derivative contribution
was sufficiently small to be ignored, and that the entire first
derivative contribution represented a positive difference polar-
izability.26 The change in interpretation is appropriate, because
of the substantial improvement of data quality that has been
achieved.
Aromatic nitrile VSE spectra show a relatively large variation
in |∆µ| values with molecular structure, in contrast to the
aliphatic data. As seen in Figure 4, both the transition dipole
Figure 4. Correlation of the transition dipole and difference dipole
moments for the aromatic mononitriles and nonsymmetric aromatic
and difference dipole moments correlate well with Hammett
parameters, which are unitless values that characterize the
electron donating or electron withdrawing nature of substituents
on an aromatic ring.39 As the transition dipole of a vibrational
excitation is directly related to the partial charges on the excited
group, the correlation with Hammett parameters is reasonable.
The observation that the difference dipoles also correlate with
Hammett parameters suggests that |∆µ| is determined by the
same partial charges, a result which is quantified below and is
discussed in a separate publication.11
Dinitriles. Centrosymmetric compounds are particularly
interesting for Stark analysis because all parameters that are
linear with the electric field should vanish by symmetry,
including ∆µ and A (eqs 1 and 2). Also, an inversion symmetric
compound has both symmetric and antisymmetric normal
modes, where the former are expected to have no absorption
intensity and the latter are expected to have twice the intensity
observed for comparable nonsymmetric compounds. The sym-
metric modes would be expected to gain intensity in an electric
field due to the symmetry breaking of the field and the effect
of the transition hyperpolarizability. Upon a weakening of the
coupling between the two nitrile groups of a symmetric
compound, relative to coupling to other internal degrees of
freedom and to the solvent, the nitrile vibrations become more
independent. Thus, absorption and Stark spectra of weakly
coupled symmetric dinitriles are expected to approach those of
nonsymmetric molecules.
dinitriles with Hammett numbers. 4-OCH3 is 4-methoxybenzonitrile;
C6H5CN is benzonitrile; 2-Cl, 3-Cl, and 4-Cl are 2-, 3-, and 4-chloro-
benzonitrile; and 1,2-CN and 1,3-CN are 1,2- and 1,3-dicyanobenzene.
As expected, most dinitriles show two absorption peaks in
the nitrile stretch region. With the exception of cis,cis-
mucononitrile (1,4-dicyano-cis,cis-1,3-butadiene), one peak is
always about an order of magnitude stronger than the other,
giving the antisymmetric and symmetric mode assignments,
from the discussion above. 1,4-dicyanobenzene (not shown)
exhibits strong resonance effects between the two nitrile modes,
leading to an absorption spectrum analogous to that of deuterated
acetonitrile (Figure 2A, B), and requiring a similar analysis
method (eq 9).
The VSE predictions are borne out in the symmetric dinitrile
data, presented in Table 3 and Figure 5, although other factors
partially obscure this simple picture. In particular, the symmetric
and antisymmetric mode frequencies are close enough for
dinitriles that there is often significant intensity sharing between
the bands due to nonlinear coupling.36 It is also calculated that
approximately 7% of succinonitrile (dicyanoethane) and 12%
of adiponitrile (1,4-dicyanobutane) molecules are expected to
be in low symmetry conformations.38 Additionally, the dinitrile
Stark data often exhibited a poor signal-to-noise ratio due to
low intensity absorptions, small Stark effects, and low dinitrile
solubilities in 2-MeTHF. Cyanogen40 is likely to be an especially

11860 J. Phys. Chem. A, Vol. 104, No. 51, 2000
Andrews and Boxer
TABLE 3: Vibrational Stark Effect Data for Aliphatic Dinitriles
absorption data
Stark data
b
peak area
M-1 cm-2
|M|
q
e
Aø
Bø
Cø
|∆µ|
D/f
∆R|b,c
Å3/f 2
A|b,c
compound
νja cm-1
D
1020 m2/V2
1040 Jm2/V2
1063 J2m2/V2
Å3/f
succinonitrile
2249.9
2244.4
2261.4
2225.0
2236.5
2251.3
2214.7
2223.1
374
697
68
0.039
0.053
0.017
0.010
0.033
0.046
0.049
0.048
0.24
0.33
0.10
0.06
0.20
0.28
0.30
0.29
4.7
0
0
192.0
0
0.2
9.7
0.9
0.04
0.70
4.74
4.87
0.32
1.24
3.15
-0.16
6.42
5.75
6.28
4.71
0.71
2.84
2.45
2.38
0.0277
0.0262
0.0274
0.0237
0.0092
0.0183
0.0171
0.0168
-1.58
0.42
2.84
-5.91
0.19
0.62
0.46
-0.38
0.57
0
0
0.91
0
0.09
1.03
0.21
adiponitrile (A)e
adiponitrile (S)f
fumaronitrile (S)f
23
fumaronitrile (A)e
268
515
587
548
1,4-dicyano-2-butene
cis,cis-mucononitrile (L)g
cis,cis-mucononitrile (H)h
a-c See the corresponding notes in Table 1. d Peak areas and transition dipoles are divided by two to give results on a per nitrile basis. e Antisymmetric
(A) nitrile mode. f Symmetric (S) nitrile mode. g Low (L) energy peak. h) High (H) energy peak.
Figure 5. Spectra of a homologous series of dinitriles. (A, B) Extinction coefficient and Stark spectra of adiponitrile (1,4-dicyanobutane); peaks
at 2244 and 2261 cm-1 represent the antisymmetric and symmetric modes, respectively. The solid line in panel A is a fit of modified Voigts, which
was used to separate the measured absorption into two bands. (C, D) Extinction coefficient and Stark spectra of 1,4-dicyano-2-butene. The VSE
spectrum is fit with the numerical derivatives of the absorption. (E, F) Extinction coefficient and Stark spectra of cis,cis-mucononitrile (1,4-dicyano-
cis,cis-1,3-butadiene). The VSE spectrum is fit with analytical derivatives of two modified Voigt line shapes, fit to the absorption in panel E.
interesting compound due to its high symmetry and strong
internal coupling, but was found to be insoluble in 2-MeTHF
and thus could not be analyzed.
zeroth, first, and second derivative fit contributions were plotted
as a function of the angle parameter, 1 - 3cos2ø (Figure 6).
For a purely one-dimensional system (ú ) 0° and off-diagonal
and perpendicular components of ∆r and A are zero), the
theoretical slope-to-intercept ratio for all three derivative
contributions is 0.4, using eqs 4-6.
Whereas significant linear Stark effects were seen for all
dinitriles studied, the difference dipole moments for both bands
were invariably lower for dinitriles than for the comparable
mononitriles, as expected from the discussion above. Further-
more, the |∆µ| values for both bands decreased with increased
conjugation of the hydrocarbon bridge, in agreement with their
having greater mechanical coupling between the nitriles. This
trend (Figure 5) starts with succinonitrile and adiponitrile (Figure
5A, B), which have |∆µ| values and Stark spectra that are similar
to those of aliphatic mononitriles. Increasing the succinionitrile
conjugation to give fumaronitrile (dicyanoethene) and the
adiponitrile conjugation to give 1,4-dicyano-2-butene (Figure
5C, D) and cis,cis-mucononitrile (Figure 5E, F), yield signifi-
cantly different Stark spectra with progressively smaller second
derivative components, implying smaller difference dipoles. The
presence of low symmetry conformations may contribute to the
observed trend, but this contribution is expected to be small
based on the conformational calculations presented above.
Angle Dependence: After a survey of the Stark effects of a
diverse collection of nitriles, acetonitrile (Figure 1) and 4-chloro-
benzonitrile (Figure 3) were chosen for more detailed analysis
due to their single nitrile peaks, minimal resonant interactions,
high extinction coefficients, and high molecular symmetry. To
examine the one-dimensional assumptions made above, the
It was found that 4-chloro-benzonitrile exhibits nearly one-
dimensional behavior for all three derivative components, with
ú ) 5°, A /A| ) 0.11, and ∆R /∆R| ) -0.001. Using the fits
to the angle dependent data, |∆µ| is 0.0348 D/f, A| is 1.09 Å3/f,
2
and ∆R| is -0.20 Å3/f , which are in close agreement with
the values calculated with the one-dimensional approximations
(Table 2).
Acetonitrile only exhibits one-dimensional behavior for the
second derivative component, showing much steeper slopes for
the other two components. Analysis shows that cosú ) 1.04 (
0.002 (clearly leading to an impossible ú value), A /A| ) 0.39,
∆R /∆R| ) 0.043, |∆µ| ) 0.0263 D/f, A| ) 0.79 Å3/f, and ∆R
||
||
) -0.60 Å3/f 2. The latter three parameters, |∆µ|, A , and ∆R ,
||
are in moderate agreement with the values calculated with the
one-dimensional assumptions (Table 1). From the polarizability
ratios, A /A| and ∆R /∆R|, the dominant cause of both the zeroth
and first derivative slopes would be interpreted to be a significant
transition polarizability perpendicular to the bond axis. However,
due to the 3-fold rotational symmetry of acetonitrile, a perpen-
dicular component (or an off-diagonal component) of the
transition polarizability cannot occur without significant sym-

Vibrational Stark Effects of Nitriles
J. Phys. Chem. A, Vol. 104, No. 51, 2000 11861
Figure 7. Correlation of difference dipoles with transition dipoles for
all samples. Solid circles represent aliphatic mononitriles, solid squares
represent aromatic mononitriles, open circles represent aliphatic di-
nitriles, and open squares represent aromatic dinitriles. The solid line
is the best fit to the aliphatic and aromatic mononitrile data, with a
slope of 0.33/f and intercept of 0.007 D/f.
detail.11,22,20 The transition dipole of an harmonic oscillator is
given by
q
h
|M| )
(10)
2π x2cmνj
where q is the effective charge of the oscillator and m is the
reduced mass. From AM1 calculations on the mononitriles
studied here, reduced masses are found to be 6.55 ( 0.05 atomic
mass units, where the small range reflects the fact that nitrile
vibrations are highly localized to the nitrile group. As the
vibrational frequency is similar for all the mononitriles as well,
the variation of the transition dipoles reflects, for the most part,
the variation of the effective charge.
Figure 6. Polarization angle dependence of the fit coefficients for
acetonitrile (circles) and 4-chlorobenzonitrile (crosses). Error bars
represent one standard deviation. Dashed lines are drawn with a slope-
to-intercept ratio of 0.4 as guides to show the expected behavior for fit
coefficients of one-dimensional systems (see text).
A simple interpretation of the difference dipole is that it
reflects a displacement of the effective charge by a distance
∆x, where the displacement is due to the anharmonicity of the
potential surface
metry breaking by asymmetric solvation. A more likely
explanation is that the band shape of the nitrile mode changes
in an electric field, most likely due to the weak Fermi reso-
nant band 50 cm-1 to higher energy. This possibility is ex-
plicitly ignored in the Stark analysis presented here, likely
leading to a mis-interpretation of the Aø, Bø, and Cø angle
dependence. This explanation is supported by the observation
that statistically significant curvature is seen in Figure 6B for
acetonitrile and that cosú > 1, neither of which can be explained
by the theory.
|∆µ| ) q∆x
(11)
From eqs 10 and 11, the slope of the correlation shown in Figure
7 (with slight corrections to account for the different νj values)
corresponds to a ∆x value of 0.012 Å/f. The physical interpreta-
tion is that the expectation value of the C-N bond length is
0.012 Å/f longer in the first vibrationally excited state than it is
in the ground state. This displacement can be compared against
that expected from published values of the anharmonicity. In a
perturbed harmonic oscillator model,41 the potential is defined
as
Discussion
It is found that difference dipoles correlate well with tran-
sition dipoles for all the mononitriles (Figure 7), where the
slope of a linear fit is 0.33/f and the intercept is 0.007 D/f.
Dinitrile |∆µ| values generally agree with the correlation as well,
but with significantly more scatter. In agreement with the
predictions above, symmetric dinitriles exhibit small difference
dipoles for both symmetric and antisymmetric modes and small
transition dipoles for symmetric modes. The decrease of
transition dipoles for antisymmetric modes (computed on a per
nitrile basis) is expected for electrostatic reasons since two
electron withdrawing groups compete for a finite amount of
polarizable charge, thus decreasing the amount of partial charge
on each nitrogen.
γ3
U(x) ) x2 + x3
k
2
(12)
6
The cubic anharmonicity of the C-N bond, γ3, has been
determined from HCN vapor phase spectroscopic data42 to be
126 aJ/Å3, leading to a ∆x value of 0.0078 Å. Thus, about 70%
of the difference dipole values may be explained as being due
to nitrile bond anharmonicity (assuming an f value of about 1.1).
The discrepancy between the ∆x values may result from (i)
factors other than anharmonicity that contribute to |∆µ| values,
(ii) differences between the nitrile bond of HCN and of normal
modes of larger nitriles, and (iii) solvation effects on bond
anharmonicity.
The correlation for mononitriles can be interpreted with a
very simple model, which is presented elsewhere in more
In Table 4, hydrogen cyanide and acetonitrile VSE results
are compared against theoretical calculations of the molecules

11862 J. Phys. Chem. A, Vol. 104, No. 51, 2000
Andrews and Boxer
TABLE 4: Calculated and Observed Vibrational Stark
Effects of HCN and CH3CN.
Stark and absorption results agreed with expected trends,
showing small difference dipoles and greater symmetry effects
upon improved mechanical coupling.
HCN
obsa
CH3CN
obsc
For the aromatic nitriles studied, Hammett numbers correlate
with both transition dipole moments and |∆µ| values, likely
making them a good predictive tool. More broadly, it was shown
that transition dipoles correlate well with |∆µ| values for all
mononitriles and, to a lesser extent, for dinitriles. To a large
extent, the correlation for mononitriles is consistent with that
expected from the anharmonicity of the nitrile bond. However,
Stark effects are larger than just those expected from anhar-
monicity, suggesting that other contributions are important as
well.
The ability to routinely measure vibrational Stark effects to
good precision makes it possible to investigate molecular
vibrations in condensed phases in a new and sensitive manner.
This technique has previously been shown to be useful for
measuring an electric field change in a series of myoglobin
mutants,1 and has been studied here to clarify the physical
origins of vibrational Stark effects.
property
unit
calcb
calcb
|M|
|∆µ|
∆R|
A|
D
0.134
0.0055 0.0582
0.0010 0.0263/f
0.0402 -0.60/f
0.0172
0.0066
0.0425
0.171
0.384
0.0082
0.033
D
0.05/f
Å3
-2/f
1.2/f
0.37/f
2
2
Å3
0.157
0.182
0.79/f
|∆µ|/|M| unitless
0.452/f
0.0244 -0.034/f
0.0952 0.045/f
∆R|/|M| (MV/cm)-1 -0.05/f
2
2
A|/|M|
(MV/cm)-1 0.03/f
a Unweighted averages of the values for the three subbands; assumes
one-dimensional behavior and that B is 0. b Highest level of calculation
from Reimers and Hush.22 c From data where ø angle dependence was
measured; values do not assume one-dimensional behavior, but do
assume B is 0.
in the absence of solvent.22 It is seen that all of the experimental
Stark parameters are several times greater than the calculated
values, and are often larger by more than an order of magnitude.
However, consideration of the Stark parameters in relation to
the transition dipole values results in significantly better
agreement. In particular, calculations for the acetonitrile transi-
tion polarizability and the difference dipole moment are within
about 25% of the measured values. The large discrepancies seen
for hydrogen cyanide are not surprising due to the strong solvent
effects that were observed.
Acknowledgment. We thank Dr. Arun Chattopadhyay who
made early contributions to the VSE of simple nitriles, and Prof.
Noel Hush for helpful discussions. This work was supported in
part by a grant from the NSF Chemistry Division. The FTIR
facilities are supported by the Medical Free Electron Laser
Program of the Office of Naval Research under Contract
N00014-94-1-1024.
Conclusions
A method for measuring vibrational Stark spectra has been
developed and applied to the infrared absorption of the nitrile
stretch mode of several aliphatic and aromatic nitriles. While
most compounds were not analyzed enough times for detailed
error estimates, the precision of the method can be reasonably
estimated by averaging the standard deviations for all repeated
experiments. Difference dipoles, which invariably dominated
the Stark effects, were reproducible to within (0.001 D/f, giving
a Stark tuning precision of (0.02/f cm-1/(MV/cm). Difference
polarizabilities and transition polarizabilities led to smaller
absorption changes, and could be measured reliably with
precisions of 0.3 Å3/f 2 and 0.1 Å3/f, respectively.
References and Notes
(1) Park, E. S.; Andrews, S. S.; Hu, R. B.; Boxer, S. G. J. Phys. Chem.
B 1999, 103, 9813.
(2) Park, E. S.; Thomas, M. R.; Boxer, S. G. J. Am. Chem. Soc., in
press.
(3) Onsager, L. J. Am. Chem. Soc. 1936, 58, 1486.
(4) Stryer, L. Biochemistry, Fourth ed.; W. H. Freeman and Com-
pany: New York, 1995.
(5) Laberge, M.; Sharp, K. A.; Vanderkooi, J. M. J. Phys. Chem. B
1997, 101, 7364.
(6) Phillips, G. N. J.; Teodoro, M.; Li, T.; Smith, B.; Gilson, M. M.;
Olson, J. S. J. Phys. Chem. B 1999, 103, 8817.
(7) Kolling, O. W. J. Phys. Chem. 1996, 100, 16087.
(8) Nyquist, R. A. App. Spectrosc. 1990, 44, 1405.
(9) Spitzer, R. C.; Sievers, A. J.; Silsbee, R. H. J. Opt. Soc. Am. B
1992, 9, 978.
The VSE data show that |∆µ| values are typically smallest
for symmetric dinitriles, larger for aliphatic nitriles and largest
for aromatic species. The values of |∆µ| are virtually all within
the range of 0.01 D/f to 0.04 D/f, which is several orders of
magnitude smaller than typical values for electronic Stark
effects24 and a factor of 2 to 8 smaller than those seen for the
VSE of carbon monoxide1 or nitric oxide2 bound to the heme
iron in myoglobin and heme model compounds. Where com-
parable experimental data exist, the |∆µ| values are in good
agreement.26,9 Compared to theoretical results, they are several
times larger than those predicted from ab initio calculations for
isolated molecules,22 but agreement improves significantly when
Stark parameters are considered relative to transition dipoles.
While the nitrile stretch mode is typically well separated from
other molecular vibrations, only about half of the samples
studied exhibited well defined single bands. The other samples
had multiple bands arising from Fermi resonant interactions,
varying degrees of hydrogen bonding, and perhaps multiple
sample conformations. Most dinitriles exhibited multiple bands
as well, reflecting symmetric and antisymmetric stretch modes.
It was found that Stark parameters were generally similar for
bands that arose from Fermi resonance, whereas they differed
between bands arising from different modes. While the dinitrile
data were not as simple as their symmetry might suggest, both
(10) Lambert, D. K. J. Chem. Phys. 1988, 89, 3847.
(11) Andrews, S. S.; Boxer, S. G., in preparation.
(12) AM1 level calculations using MacSpartan (Wave function Inc.).
(13) Charles, S. W.; Cullen, F. C.; Owen, N. L. J. Mol. Struct. 1976,
34, 219.
(14) DeLeon, R. L.; Muenter, J. S. J. Chem. Phys. 1984, 80, 3992.
(15) Kumar, A. P.; Rao, G. R. Sepctrochim. Acta Part A 1997, 53, 2033.
(16) Michel, H.; Lippert, E. Acetonitrile-The structure of the liquid and
solid phases and the nature of the liquid-solid-phase transition. In Organic
Liquids. Structure, dynamics, and chemical properties; Buckingham, A. D.,
Lippert, E., Bratos, S., Eds.; John Wiley & Sons: New York, 1978; p 293.
(17) Pace, E. L.; Noe, L. J. J. Chem. Phys. 1968, 49, 5317.
(18) Reimers, J. R.; Hall, L. E. J. Am. Chem. Soc. 1999, 121, 3730.
(19) Thomas, B. H.; Orville-Thomas, W. J. J. Mol. Struct. 1971, 7, 123.
(20) Hush, N. S.; Reimers, J. R. J. Phys. Chem. 1995, 99, 15798.
(21) Mathies, R. A. Experimental and theoretical studies on the excited
electronic states of some aromatic hydrocarbons through electric field
perturbation and through chemical substituents, Cornell University, 1974.
(22) Reimers, J. R.; Zeng, J.; Hush, N. S. J. Phys. Chem. 1996, 100,
1498.
(23) Wortmann, R.; Bishop, D. M. J. Chem. Phys. 1998, 108, 1001.
(24) Bublitz, G. U.; Boxer, S. G. Annu. ReV. Phys. Chem. 1997, 48,
213.
(25) Jucks, K. W.; Miller, R. E. J. Chem. Phys. 1988, 88, 6059.
(26) Chattopadhyay, A.; Boxer, S. G. J. Am. Chem. Soc. 1995, 117,
1449.
(27) Andrews, S. S.; Boxer, S. G. ReV. Sci. Inst. 2000, 71, 3567.
(28) Long, F.; Freedman, T. B.; Tague, T. J.; Nafie, L. A. Appl.
Spectrosc. 1997, 51, 508.

Vibrational Stark Effects of Nitriles
J. Phys. Chem. A, Vol. 104, No. 51, 2000 11863
(29) Three factors appear to improve the signal-to-noise for rapid-scan
methods relative to step-scan methods: (i) it is not necessary to wait for a
lock-in to settle after each mirror step, (ii) multiplexing of signal frequencies
is used, and (iii) interferogram oversampling requires no additional time.
This is discussed more fully in Andrews, S. S.; Boxer, S. G. In preparation.
(30) Early experiments were performed in a miniature Joule-Thompson
refrigerator (MMR Technologies) operating at 80 K, but we found significant
sample orientation in an electric field, leading to the development of the
immersion cryostat. Poling may not be a problem for AC electric fields.
(31) Data fitting was done with Macintosh software created for this
purpose. The software and C language source code are available upon
request.
gives absorption between the peaks, and is responsible for Fermi resonance.
The excess absorption is analogous to the coalescence seen in NMR spectra
of dynamic systems (see e.g. Gasparro, F. P.; Kolodny, N. H. J. Chem.
Educ. 1977, 54, 258-261).
(37) Miller, T. M. Atomic and molecular polarizabilities. In Handbook
of Chemistry and Physics, 71st ed.; Lide, D. R., Ed.; CRC Press: Boca
Raton, FL, 1990; p 10.
(38) The transition state enthalpy is expected to be about 16 kJ/mol,
from butane data (see e.g., Kemp, D. S.; Vellaccio, F. Organic Chemistry;
Worth Publishers: New York, 1980; p. 472), implying an interconversion
rate below 1 ms down to the 100 K glass transition temperature, which is
fast compared to the freezing rate. Thus, low-temperature population ratios
were computed for 100 K, with the assumption that the sample is at
equilibrium before freezing and immobilized thereafter. Boltzmann popula-
tion calculations use an enthalpy difference between anti and gauche
conformations of 2.8 kJ/mol, a value found by AM1 calculations for the
relevent nitriles.
(32) The parameters of a modified Voigt are: peak height, frequency,
full width at half-maximum, and the Gaussian:Lorentzian ratio. These
parameters are adjusted only to obtain a best fit to the absorption spectrum,
and are not considered to have any physical meaning.
(33) Assuming the maximum possible ú angle of 90° and the maximum
sample rotation angle of ø ) 20°, Cø would be corrected to be 13% too
small under the one-dimensional assumptions, contributing to a 7% error
in the calculation of |∆µ|.
(39) Lowry, T. H.; Richardson, K. S. Mechanism and Theory in Organic
Chemistry, 3rd ed.; Harper Collins: New York, 1987.
(34) Atkins, P. W. Physical Chemistry, 5th ed.; W. H. Freeman and
Company: New York, 1994.
(40) Cyanogen was synthesized by an adaptation of a method from:
Janz, George J. Inorg. Synth. 1957, 5, 43-48.
(41) McQuarrie, D. A. Quantum Chemistry; University Science Books:
Sausalito, CA, 1983.
(35) Smith, A. L.; Keller, W. E.; Johnston, H. L. Phys. ReV. 1950, 79,
728.
(36) Linear coupling between two modes results in redefined normal
modes and a separation of the absorption peaks, whereas nonlinear coupling
(42) Strey, G.; Mills, I. M. Mol. Phys. 1973, 26, 129.