Ultrafast dynamics of pyrrolidinium cation ionic liquids

Article abstract of DOI:10.1063/1.1893797

We have investigated the ultrafast molecular dynamics of five pyrrolidinium cation room temperature ionic liquids using femtosecond optical heterodyne-detected Raman-induced Kerr effect spectroscopy. The ionic liquids studied are N -butyl- N -methylpyrrolidinium bis(trifluoromethylsulfonyl)imide P 14 + NTf2-), N -methoxyethyl- N -methylpyrrolidinium bis(trifluoromethylsulfonyl)imide P 1EOE + NTf2-), N -ethoxyethyl- N -methylpyrrolidinium bis(trifluoromethylsulfonyl)imide P 1EOE + NTf2-), N -ethoxyethyl- N -methylpyrrolidinium bromide P 1EOE +, and N -ethoxyethyl- N -methylpyrrolidinium dicyanoamide P 1EOE + DCA-). For comparing dynamics among the five ionic liquids, we categorize the ionic liquids into two groups. One group of liquids comprises the three pyrrolidinium cations P 14 +, P 1EOM +, and P 1EOE + paired with the NTf2- anion. The other group of liquids consists of the P 1EOE + cation paired with each of the three anions NTf2-, Br-, and DCA-. The overdamped relaxation for time scales longer than 2 ps has been fit by a triexponential function for each of the five pyrrolidinium ionic liquids. The fast (~2 ps) and intermediate (~20 ps) relaxation time constants vary little among these five ionic liquids. However, the slow relaxation time constant correlates with the viscosity. Thus, the Kerr spectra in the range from 0 to 750 cm-1 are quite similar for the group of three pyrrolidinium ionic liquids paired with the NTf2- anion. The intermolecular vibrational line shapes between 0 and 150 cm-1 are fit to a multimode Brownian oscillator model; adequate fits required at least three modes to be included in the line shape.

Full text of DOI:10.1063/1.1893797

Ultrafast dynamics of pyrrolidinium cation ionic liquids
Hideaki Shirota, Alison M. Funston, James F. Wishart, and Edward W. Castner
Citation: J. Chem. Phys. 122, 184512 (2005); doi: 10.1063/1.1893797
View online: http://dx.doi.org/10.1063/1.1893797
View Table of Contents: http://jcp.aip.org/resource/1/JCPSA6/v122/i18
Published by the American Institute of Physics.
Additional information on J. Chem. Phys.
Journal Homepage: http://jcp.aip.org/
Journal Information: http://jcp.aip.org/about/about_the_journal
Top downloads: http://jcp.aip.org/features/most_downloaded
Information for Authors: http://jcp.aip.org/authors
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

THE JOURNAL OF CHEMICAL PHYSICS 122, 184512 ͑2005͒
Ultrafast dynamics of pyrrolidinium cation ionic liquids
Hideaki Shirota
Department of Chemistry & Chemical Biology, Rutgers, The State University of New Jersey, Piscataway,
New Jersey 08854-8087
Alison M. Funstona͒ and James F. Wishart
Brookhaven National Laboratory, Chemistry Department, Upton, New York 11973-5000
Edward W. Castner, Jr.
Department of Chemistry & Chemical Biology, Rutgers, The State University of New Jersey, Piscataway,
New Jersey 08854-8087
͑Received 3 January 2005; accepted 23 February 2005; published online 10 May 2005͒
We have investigated the ultrafast molecular dynamics of five pyrrolidinium cation room
temperature ionic liquids using femtosecond optical heterodyne-detected Raman-induced
Kerr effect spectroscopy. The ionic liquids studied are N-butyl-N-methylpyrro-
−
lidinium
bis͑trifluoromethylsulfonyl͒imide
P14+/NTf2 ͒,
N-methoxyethyl-N-methylpyrro-
lidinium bis͑trifluoromethylsulfonyl͒imide P1E−OE+/NTf2 ͒, N-ethoxyethyl-N-methylpyrrolidinium
−
bis͑trifluoromethylsulfonyl͒imide P1EOE+/NTf2 ͒, N-ethoxyethyl-N-methylpyrrolidinium bromide
P1EOE+, and N-ethoxyethyl-N-methylpyrrolidinium dicyanoamide P1EOE /DCA−͒. For comparing
+
dynamics among the five ionic liquids, we categorize the ionic liquids into two groups. One group
of liquids comprises the three pyrrolidinium cations P14++, P1EOM+, and P1EOE+ paired with the NTf2
−
anion. The other group of liquids consists of the P1EOE cation paired with each of the three anions
−
NTf2 , Br−, and DCA−. The overdamped relaxation for time scales longer than 2 ps has been fit by
a triexponential function for each of the five pyrrolidinium ionic liquids. The fast ͑ϳ2 ps͒ and
intermediate ͑ϳ20 ps͒ relaxation time constants vary little among these five ionic liquids. However,
the slow relaxation time constant correlates with the viscosity. Thus, the Kerr spectra in the range
from 0 to 750 cm−1 are quite similar for the group of three pyrrolidinium ionic liquids paired with
−
the NTf2 anion. The intermolecular vibrational line shapes between 0 and 150 cm−1 are fit to a
multimode Brownian oscillator model; adequate fits required at least three modes to be included in
the line shape. © 2005 American Institute of Physics. ͓DOI: 10.1063/1.1893797͔
enzymatic catalysis,17 and enantiomeric synthesis.18,19 Ana-
I. INTRODUCTION
lytical applications of ionic liquids include uses such as sta-
tionary phases for chiral separations,20 and as nonvolatile
solvents for matrix assisted laser desorption ionization time-
of-flight mass spectroscopy.21–23 Each of these applications
can benefit from a detailed physical understanding of the
dynamics and interactions in ionic liquids. Thus, we have
undertaken a spectroscopic study of five pyrrolidinium-
cation ionic liquids.
Femtosecond optical heterodyne-detected Raman-
induced Kerr effect spectroscopy ͑OHD-RIKES͒ is a nonlin-
ear optical polarization technique that measures the depolar-
ized Raman signal from a transparent condensed phase
sample in the time domain.24,25 Using OHD-RIKES, we can
measure the dynamics in condensed phases on time scales
ranging from 20 fs to 1 ns. The time domain data are con-
verted to the low-frequency depolarized Raman spectrum by
direct Fourier transform deconvolution.25–27 Spectral sensi-
tivity in the range from 0.1 to 750 cm−1 makes OHD-RIKES
a powerful tool to investigate the intermolecular dynamics
and interactions in condensed phases. Recently, femtosecond
OHD-RIKES has been used for simple molecular
liquids,24,25,28–30 and for more complex condensed phases
Room temperature ionic liquids ͑RTILs͒ are molten salts
for which the low temperature limit of the broad liquidus is
at or below ambient temperature. The strong electrical inter-
actions in these liquids cause them to have very low vapor
pressures, which result in solvent properties that are favor-
able for “green” chemistry.1 RTIL vapor pressures are orders
of magnitude lower than for neutral organic solvents. RTILs
show high solubility for both inorganic and organic com-
pounds, including peptides, proteins, and enzymes. As re-
placements for volatile organic solvents, the low vapor pres-
sures of RTILs provide opportunities for reducing emissions
of volatile organic compounds to the atmosphere.2–10 In par-
ticular the low vapor pressures of RTILs also provide oppor-
tunities for reducing the risk of uncontrolled gas phase chain
reactions, fires, and explosions. Many important chemical
reactions, such as hydrogenations,11,12 oxidations,13 and cou-
pling reactions14,15 have been investigated in RTILs to
test their use as organic solvent replacements. RTILs have
also proven useful as solvents for materials synthesis,16
a͒
Current address: School of Chemistry, The University of Melbourne, Vic-
toria, 3010 Australia.
0021-9606/2005/122͑18͒/184512/12/$22.50
122, 184512-1
© 2005 American Institute of Physics
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-2
Shirota et al.
J. Chem. Phys. 122, 184512 ͑2005͒
In a number of cases, cyclic aromatic molecules are more
toxic than molecules with the corresponding aliphatic ring
systems. If this proves true for RTILs, then the aliphatic cat-
ion RTILs might be better choices from an environmental
perspective. We have studied the following five different
pyrrolidinium ionic liquids shown in Fig. 1: N-butyl-
N-methylpyrrolidinium
bis͑trifluoromethylsulfonyl͒imide
−
͑P14+/NTf2 ͒,
bis͑trifluoromethylsulfonyl͒imide
N-ethoxyethyl-N-methylpyrrolidinium bis͑trifluoromethyl-
N-methoxyethyl-N-methylpyrrolidinium
−
͑P1EOM+/NTf2 ͒,
−
sulfonyl͒imide
N-methylpyrrolidinium
͑P1EOE+/NTf2 ͒,
N-ethoxyethyl-
+
bromide
͑P1EOE /Br−͒,
and
N-ethoxy-ethyl-N-methyl-pyrrolidinium
dicyanoamide
+
͑P1EOE /DCA−͒. We will compare the dynamical aspects
of the cations using measurements from the series of
−
−
−
NTf2 anion RTILs of P14+/NTf2 , P1EOM+/NTf2 , and
P1EOE+/NTf2 . The effects on liquid dynamics from anion
−
motions are investigated by comparing dynamics from
−
+
+
P1EOE+/NTf2 , P1EOE /Br−, and P1EOE /DCA−. For the case
+
of P1EOE /Br−, we can measure cation dynamics directly be-
cause the spherical bromide anion cannot directly contribute
to the OHD-RIKES signal, since this signal depends on fluc-
tuations of the polarizability anisotropy.
FIG. 1. Chemical structures of the pyrrolidinium cation ionic liquids used in
+
+
this study. P14 is N-butyl-N-methylpyrrolidinium cation, P1EOM is
+
N-methoxyethyl-N-methylpyrrolidinium cation, P1EOE is N-ethoxyethyl-
−
N-methyl-pyrrolidinium cation, NTf2 is bis͑trifluoromethylsulfonyl͒imide
anion, DCA− is dicyanoamide anion, and Br− is bromide anion.
such as nanoporous glasses,31,32 microemulsions,33,34 hy-
drated peptide and protein solutions,35–37 and polymer
solutions.34,38
II. EXPERIMENT
A. Sample preparation
The synthesis of the ionic liquids used in this study will
be reported in detail elsewhere,43 so only a brief description
of the procedures used are presented here. The pyrrolidinium
cations were synthesized via standard methods, viz. quater-
nization of N-methylpyrrolidine with the appropriate alkyl
halide ͑in this case bromide͒ to produce the pyrrolidinium
bromide salts.44–46 These were metathesized to the bis͑trif-
luoromethylsulfonyl͒imide salt using equimolar quantities of
the pyrrolidinium bromide and lithium bis͑trifluoromethyl-
sulfonyl͒imide in aqueous solution.45,47 Metathesis of the
pyrrolidinium bromide to the dicyanoamide salt was done
according to Refs. 47 and 48 by reacting equimolar quanti-
ties of silver dicyanoamide48 with the appropriate pyrroli-
dinium bromide in water, filtering off the silver bromide pre-
cipitate, and evaporating off the aqueous phase. The
pyrrolidinium dicyanoamide was extracted from the residue
into dichloromethane, which was filtered and evaporated to
yield the pure ionic liquid.48,49 The resultant liquids were
decolorized by stirring with activated charcoal overnight and
then passing a solution of the liquid in acetonitrile through
an activated alumina column. All ionic liquids were dried at
reduced pressure at 318 K for more than 24 h prior to char-
acterization and use. All starting materials were purchased
from Aldrich in the highest available purity grade and used
as received. Viscosity measurements were made using a
temperature-controlled Cambridge Applied Systems recipro-
cating electromagnetic piston viscometer ͑ViscoLab 4100͒.
Several groups have studied the ultrafast dynamics of
aromatic imidazolium ionic liquids using OHD-RIKES.
Quitevis and co-workers characterized how changes of the
alkyl substituent affected the intermolecular vibrational dy-
namics for a series of 1-alkyl-3-methylimidazolium cations
paired with bis͑trifluoromethylsulfonyl͒imide anion.39
Recently they also reported the temperature dependence of
the low-frequency Kerr spectrum for the RTIL 1-pentyl-3-
methylimidazolium
bis͑trifluoromethylsulfonyl͒imide.40
Giraud et al. studied five typical imidazolium ionic liquids.41
They analyzed the low-frequency Kerr spectra of the ionic
liquids using a multimode Brownian oscillator model and
concluded that the signals from each of the five ionic liquids
are dominated by the imidazolium cation dynamics. Cang,
Li, and Fayer studied the dynamics of the RTIL 1-ethyl-3-
methylimidazolium nitrate and concentrated on diffusive re-
laxation processes with the time range from 1 ps to 2 ns.42
They described the temperature dependence of the observed
diffusive relaxation decays using a mode-coupling theory for
the dynamics, in which the data are fit to a function where a
sum of two power-law decays describes the faster relaxation
processes, and this sum is multiplied by a slower exponential
decay assigned to the ␣ relaxation.
In this paper, we present high quality Kerr transient data
for each of the five pyrrolidinium RTILs shown in Fig. 1.
There are several reasons for investigating nonaromatic ionic
liquids. The electronic structure of the aromatic molecular
cations is more delocalized than for the aliphatic molecular
cations leading to ion-ion interactions of a different nature.
1
No impurity signals were detected in the H-NMR spectra of
the neat ionic liquids.
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-3
Ultrafast dynamics of ionic liquids
J. Chem. Phys. 122, 184512 ͑2005͒
−
dinium cations P1EOE+, P1EOM+, and P14+, and for the NTf2
B. OHD-RIKES measurement
and DCA− anions of the ionic liquids were performed by
using the GAUSSIAN 03 program package.51 Becke’s three-
parameter gradient-corrected exchange52 and Lee–Yang–Parr
gradient-corrected correlation functions53 ͑B3LYP͒ were
used with the 6-31+G͑d,p͒ basis set.51 Calculations were
done for the isolated molecular ion and again for the ion in a
dielectric continuum model using a dielectric constant of
36.64 for acetonitrile. The value for the acetonitrile dielectric
constant was chosen because of the similar solvatochromic
effects observed in ionic liquids.54 The integral equation for-
malism, polarized continuum model ͑IEF-PCM͒ of Men-
nucci and Tomasi55,56 was used to mimic a solvent environ-
ment similar to that observed for solvatochromic shifts in
several ionic liquids. Molecular ion geometries were opti-
mized with no constraints. The XYZ Cartesian coordinates
for cation and anion optimized by B3LYP/6-31+G͑d,p͒
calculations are summarized in the EPAPS supporting infor-
mation document.57
Details of the OHD-RIKES apparatus were reported
elsewhere.38,50 Briefly, a lab-built femtosecond titanium sap-
phire laser operating at 800 nm center wavelength was used
as a light source. The output power was about 320 mW when
the laser was pumped by 3.2 W of 532 nm light from a
frequency-doubled diode-pumped continuous Nd:VO4 laser
͑Spectra Physics, Millennia V͒. To compensate for the group
velocity dispersion introduced by the optics in the spectrom-
eter, the laser output was first routed through a retroreflecting
dispersive delay line consisting of a pair of fused silica
Brewster prisms separated by Ϸ120 cm. After the delay line
the laser pulses passed thorough a half-wave plate to rotate
their polarization to vertical, and were incident on an un-
coated fused silica interferometric wedge to produce pump
and probe beams with an intensity ratio of about 90:5. The
pump beam is routed though a half-wave plate and a Glan–
Thompson polarizer transmitting vertically polarized light
prior to the focusing lens. The pump beam intensity was
modulated at 1 kHz by a mechanical chopper ͑Stanford Re-
search Systems, SR540͒. The probe beam is routed through
an antireflection coated interferometer wedge, a variable op-
tical delay line ͑Pacific Precision Labs, 0.1 micron accuracy͒,
a half-wave plate and a Glan–Thompson polarizer to set the
polarization 45° from vertical, and then through a quarter-
wave plate before the fused silica focusing lens. The pump
and probe beams were focused on the sample and recolli-
mated with a matched pair of 100 mm focal length fused
silica lenses. The pump beam was blocked while the probe
was sent though a second Glan–Thompson polarizer and a
spatial filter consisting of two 150 mm lenses and a 0.15 mm
pinhole. To achieve heterodyne detection, an out of phase
local oscillator was introduced by rotating the input polarizer
by ±1° away from the homodyne orientation ͑defined at the
polarizer extinction angle͒. The OHD-RIKES probe beam
was detected by a large area amplified p-i-n photodiode
͑New Focus, 2032͒ and recorded as a function of pump-
probe delay by a digital lock-in amplifier ͑Stanford Research
Systems, SR830͒ referenced to the chopper frequency. The
laser pulse cross correlation measured with a 200 ␮m potas-
sium dihydrogen phosphate crystal was 36±3 fs FWHM
͑equivalent to a 25±2 fs pulse, assuming a Gaussian line
shape͒. The measured spectral bandwidth of the laser pulses
was about 55 nm. All the measurements were made at
295±2 K. Scans with high time resolution of 4096 points at
0.5 ␮m/step were recorded for a time window of 13.66 ps.
The slower orientational dynamics were captured in 115 ps
scans with 20 ␮m per data point and in 730 ps scans with
50 ␮m per data point. Ten scans were averaged for the short
time window transients ͑5 scans each for +1° and −1° rota-
tions of the probe input polarizer͒ and 20 scans were aver-
aged for the intermediate and long time windows ͑10 scans
each for both the +1° and −1° rotations of the probe input
polarizer͒.
III. RESULTS
A. Picosecond reorientational dynamics
Figure 2 shows the log-log plots of the Kerr transients of
the five pyrrolidinium ionic liquids. As shown in Fig. 2, the
−
−
Kerr transients of P1EOE+/NTf2 , P1EOM+/NTf2 , and
P14+/NTf2 are quite similar. On the other hand, the Kerr
−
+
+
transients for P1EOE /Br− and P1EOE /DCA− are clearly dis-
+
tinguishable. The Kerr transient of P1EOE /Br− shows that the
amplitude of the picosecond reorientational relaxation is
much smaller than that of the other ionic liquids.
In Fig. 2, one finds that for time delays from 10 fs to 1
ps, a strong beating pattern is observed in the Kerr transients.
The beats arise from multiple intramolecular normal modes,
and are most complex for the RTILs with the NTf2− anion. A
different beating pattern with a smaller modulation depth is
+
observed over the same time range for the P1EOE /Br− liquid.
+
For the P1EOE /DCA− liquid, the beating pattern is a more
simple sinusoidal decay over this time range, indicating the
presence of a single dominant low-frequency mode. In the 1
to 10 ps time window, the intramolecular modes are damped.
For longer time scales beyond 10 ps, the intramolecular vi-
brational damping is complete and a nonexponential decay is
observed for each of the five liquids.
The overdamped nonexponential decay part of the Kerr
transients from 2 to 740 ps is fit to several model functions.
Exponential, stretched exponential, biexponential, and a sum
of exponential plus stretched exponential functions do not
adequately fit the data. Both triexponential and mode cou-
pling theory model functions42 provide improved quality fits,
but a triexponential model is generally superior. We note that
the long time noise floor for our Kerr transients ranges be-
tween 2ϫ10−5 to 5ϫ10−5 on an intensity scale normalized
to the peak of the electronic response. To obtain a reduced
noise floor, it would be necessary to either use a scheme with
multiple laser pulses of different durations and intensities as
done by the Fayer group,42 or to employ a multiple modula-
C. Density functional theory calculations
Optimized geometries, intramolecular vibrational modes,
and effective atom-centered charges for the three pyrroli-
tion method as demonstrated by Fourkas and co-workers.58
A
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-4
Shirota et al.
J. Chem. Phys. 122, 184512 ͑2005͒
notable points from Table I are the following: ͑1͒ the ampli-
+
tude of the diffusive relaxation for P1EOE /Br− relative to the
electronic contribution is only about 10% of that for the
other ionic liquids, ͑2͒ the fast and intermediate relaxation
time constants ␶1 and ␶2 do not vary much between these five
ionic liquids, and ͑3͒ the slow relaxation time constants ␶
3
for the five pyrrolidinium RTILs range from 160 ps to 3.2 ns,
and correlate with the values of the shear viscosity.
B. Femtosecond dynamics: Low-frequency Kerr
spectrum
The Kerr transients are analyzed by means of the Fourier
transform deconvolution procedure introduced by
McMorrow and Lotshaw.26,27 To obtain a sufficient number
of data points for a spectrum of reasonable accuracy, it is
necessary to extend the high time resolution data from Fig. 2
͑with 4096 data points over the time window from −1.5 to 12
ps͒ over a longer time range than it is practical to record. To
achieve this, the longer time scale Kerr transients ͑Fig. 2͒ in
the range from 2 to 740 ps were recorded with lower time
resolution, and are fit to
a triexponential function
͑a1 exp͑−t/␶ ͒+a2 exp͑−t/␶ ͒+a3 exp͑−t/␶ ͒͒. Prior to
1
2
3
Fourier-transform analysis, the 4096 point data set from −1.5
to 12 ps is extended to 131 072 points using the triexponen-
tial fit function with a time step of 3.3355 fs, yielding spectra
with about 0.0763 cm−1/point spacing. All the spectra ob-
tained in this study are well resolved up to 750 cm−1.
Figure 4 shows the Fourier-transform Kerr spectra of the
five ionic liquids. These spectra have had the contribution
from the diffusive components subtracted prior to Fourier
−
deconvolution. As shown in Fig. 4, P1EOE+/NTf2 ,
−
−
P1EOM+/NTf2 , and P14+/NTf2 show rich intramolecular vi-
+
brational bands spectra in comparison with P1EOE /Br− and
P1EOE /DCA−. The spectra of P1EOE+/NTf2 , P1EOM+/NTf2 ,
+
−
−
and P14+/NTf2 are quite similar. The frequencies of the in-
−
tramolecular vibrational modes observed in the Kerr spectra
+
of the ionic liquids are summarized in Table II. P1EOE in-
tramolecular vibrational bands are observed at band centers
+
of 283, 380, 527, and 642 cm−1 in P1EOE /Br− and
+
P1EOE /DCA−, though they are of much lower intensity com-
−
pared to the strong intramolecular bands of the NTf2 anion.
C. Density functional theory calculations
The electrostatic properties for each of the five molecu-
lar ions are obtained following geometry optimizations.
Atom-centered charges from the ChelpG model are obtained
for the gas-phase ion, for the PCM model using the gas-
phase geometry, and again for a geometry reoptimization for
the stable ground state with PCM polarization. The magni-
tudes of the atom-centered charges are enhanced relative to
the gas phase when the IEF-PCM model is applied, and are
further enhanced as the geometry is reoptimized from the gas
phase. The electric multipole moments obtained are for the
standard orientation, center-of-mass coordinate system. It is
important to note that dipole moments for ions depend on the
choice of origin for the atomic coordinates. Optimized ge-
−
FIG. 2. Log-log plots of Kerr transients for ͑a͒ P1EOE+ /NTf2
, ͑b͒
P1EOM+ /NTf2−, ͑c͒ P14+ /NTf2−, ͑d͒ P1EOE /Br−, and ͑e͒ P1EOE /DCA−.
+
+
Triexponential fit functions are also shown in the figure.
reduced noise floor will be required to differentiate between
the triexponential decay and mode-coupling theory42 models.
The triexponential fit curves are also shown in Fig. 2.
The fit parameters are summarized in Table I and plotted in
Fig. 3. The sum of the amplitudes is normalized to unity. The
amplitudes of the diffusive relaxation components relative to
the instantaneous electronic responses are also listed. The
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-5
Ultrafast dynamics of ionic liquids
J. Chem. Phys. 122, 184512 ͑2005͒
TABLE I. Triexponential fit parameters for diffusive Kerr transients of pyrrolidinium ionic liquids.
a
b
c
d
d
d
e
Ionic liquid
P1EOE+ /NTf2
aod
a1
␶ ͑ps͒
a2
␶
͑ps͒
a3
␶ ͑ps͒
1
2
3
−
3.52ϫ10−3
3.63ϫ10−3
3.41ϫ10−3
0.47ϫ10−3
5.66ϫ10−3
0.71
0.68
0.72
0.78
0.78
2.31
2.27
2.43
1.89
2.02
0.23
0.26
0.22
0.17
0.20
15.4
14.6
18.3
21.2
16.2
0.06
0.06
0.06
0.05f
0.02
165
161
253
P1EOM+ /NTf2
−
−
P14+ /NTf2
+
P1EOE /Br−
3200g
686
+
P1EOE /DCA−
aSum of amplitudes of triexponential function normalized by the instantaneous electronic Kerr response.
bError is ±10%.
cError is ±20%.
dError is ±30%.
eError is ±35%.
fError is ±50%.
gError is ±100%.
ometry coordinates, rotational constants, CHelpG charges,
and electrostatic properties are reported in the supporting
information.57
g2
j2
␶
=
␶rot,
͑1͒
col
Harmonic frequencies for each of the optimized geom-
etries were calculated, both for the isolated gas-phase mo-
lecular ion, and with the polarized continuum model using
the value for the room temperature dielectric constant of ac-
etonitrile. The complex vibrational spectroscopy resulting
from the flexibility of each ion makes most comparisons of
observed versus calculated modes difficult. An exception is
the pentaatomic dicyanoamide anion, which has a strong
bending mode observed at 180 cm−1 for which calculated
values are 160 cm−1 ͑gas phase͒ and 174 cm−1 ͑PCM opti-
mized͒; and the symmetric in-plane cyano wagging mode
observed experimentally at 663 cm−1 and calculated to be at
650 cm−1 for the PCM model.
where g2 is the static pair orientational correlation parameter
and j2 is the dynamical orientational correlation pair
parameter.59 Although it is difficult to estimate g2 and j2 for
the RTIL molecular ions, the collective ͑structural͒ reorien-
tational time is proportional to the single molecule reorien-
tational time if the ratio of g2 to j2 does not vary between the
five pyrrolidinium RTILs. Comparisons between NMR and
neutron scattering experiments and molecular dynamics
simulations will be required to address the characteristics of
g2 and j2 for ionic liquids.
Results from the triexponential analysis of the longer
time Kerr transients are summarized in Table I. Only the
longest time constant ␶ varies with viscosity in a manner
3
consistent with diffusive relaxation. For diffusive orienta-
tional relaxation, the reorientation time constant ␶rot of a sol-
ute in many simple organic molecular liquids is well de-
scribed by the Stokes–Einstein–Debye hydrodynamic
model,60
IV. DISCUSSION
Reorientational relaxation signals from optical Kerr ef-
fect experiments provide a measure of the collective re-
sponse ␶col, unlike single particle experiments, such as NMR
6V␩f
L
rot
␶
=
,
͑2͒
L͑L + 1͒kBT
relaxation. The collective reorientational relaxation time ␶
col
is correlated with the single molecule reorientation time
␶ ,
rot
59
where V is the volume of the rotating species, ␩ is the shear
viscosity of the liquid, f is a shape factor, kB is the Boltz-
mann constant, T is the absolute temperature, and the rank L
equals 2 for depolarized Kerr reorientation signals. Figure 3
shows the plots of the longest reorientation time ␶ versus V
3
␩
for the pyrrolidinium ionic liquids. Because the fitted time
constant ␶3 for P1EOE+/Br− is 3.2 ns, which is more than four
times longer than the time window for the present OHD-
−
RIKES measurements, only the ␶ values for P +/NTf2 ,
3
14
P1EOM+/NTf2 , P1EOE+/NTf2 , and P1EOE /DCA− are plotted
in Fig. 3. The values for several physical properties including
formula weight MW, density d, formula volume V, viscosity
␩ of the pyrrolidinium RTILs are summarized in Table III.
The values of V for the ionic liquid cations and anions are
estimated by the method of van der Waals increments,61,62
with the value of V for the Br− ion taken from Israelachvili’s
book.63
−
−
+
FIG. 3. Plots of the slow relaxation time constant ␶ vs ␩V for the five
3
pyrrolidinium ionic liquids. Squares, triangles, and circles represent points
calculated using the molecular volumes of the anion, cation, and the sum of
cation and anion, respectively.
Figure 3 plots the longest Kerr relaxation time constant
␶3 versus the product of the shear viscosity times the volume
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-6
Shirota et al.
J. Chem. Phys. 122, 184512 ͑2005͒
of the viscosity-volume products Vanion␩ or Vcation␩. In a re-
cent molecular dynamics study of solvation in the RTIL
1-butyl-3-methylimidazolium hexafluorophosphate, Kobrak
and Znamenskiy found that the dynamics of anions and cat-
ions are strongly coupled.64 Margulis also showed the strong
correlation between cation and anion diffusions in 1-alkyl-3-
methylimidazolium hexafluorophosphate ionic liquids by
molecular dynamics simulation.65 Thus, we believe it is plau-
sible that the Kerr orientational signal is comprised of a sum
of orientational correlations of individual cations and anions,
as well as cation-anion pairs. We cannot exclude the possi-
bility that there is a contribution to the overall diffusive re-
sponse from more complex transient aggregates, such as
cation-anion-cation or anion-cation-anion trimers. To con-
tribute to the Kerr diffusive reorientation signal, such aggre-
gates would have to have a cluster lifetime greater than the
jump diffusion time.
Cang, Li, and Fayer measured the Kerr relaxation dy-
namics of 1-ethyl-3-methylimidazolium nitrate at the tem-
perature range of 295–410 K.42 They found that the relax-
ation was best fit to a model from mode coupling theory.
This model is the product of the sum of two power-law de-
cays multiplied by a slower exponential decay, which is as-
signed to the overall ␣ relaxation of the glassy liquid.42 Their
results showed that the longest ␣-relaxation time constant
follows the trend predicted by the Stokes–Einstein–Debye
hydrodynamic model if the rotating solute volume is taken as
the volume of only the cation. In the pyrrolidinium RTILs, it
is possible that the appropriate volume of a rotor may be that
of the sum of the cation and anion volumes. Temperature
dependent Kerr dynamics experiments of both nonspherical
anion and cation ionic liquids will be needed to obtain fur-
ther understanding of reorientation dynamics in room tem-
perature ionic liquids. To study whether clustering affects the
measured reorientation, it will also be helpful to have inde-
pendent experimental data from NMR and quasielastic neu-
tron scattering. In particular, temperature-dependent NMR
experiments will elucidate to what degree internal torsions
occur on competing time scales with overall diffusive tum-
bling motions of the molecular ions.
The values of the fast and intermediate relaxation time
constants ␶ and ␶ do not vary much between the five pyr-
1
2
rolidinium ionic liquids and have values of about 2 ps and 17
ps, respectively. The corresponding normalized amplitudes
for a1 and a2 are about 0.73 and 0.22, respectively. Several
groups have characterized the solvation dynamics in ionic
liquids by time dependent fluorescence Stokes shift measure-
ments of solvatochromic probes such as coumarin 153 and
4-aminopthalimide.54,66–71 The solvation dynamics show
strongly nonexponential relaxation, fitting well to a stretched
exponential function with a ␤ value of 0.4–0.5, for both aro-
matic and nonaromatic RTILs. The average solvation time
constants observed in these experiments are substantially
longer than the observed Kerr reorientation dynamics, typi-
cally occurring on subnanosecond to nanosecond time scales.
We have measured solvation dynamics for the pyrrolidinium-
FIG. 4. Fourier-transform frequency power spectra of ͑a͒ P1EOE+ /NTf2−, ͑b͒
+
+
P1EOM+ /NTf2−, ͑c͒ P14+ /NTf2−, ͑d͒ P1EOE /Br−, and ͑e͒ P1EOE /DCA−. The
contribution of the picosecond relaxation was subtracted.
+
for each of the pyrrolidinium RTILs, excepting P1EOE /Br−.
Using either the cation or anion volume to calculate the
graph abscissa, a rough correlation is observed. However,
noticeably better agreement and a better statistical linear cor-
relation are found when the volume is taken to be the sum of
the cation plus anion molecular volumes, Vtotal=Vcation
+Vanion. If the slowest orientational relaxation process ob-
served in the Kerr signal ͑characterized by the time constant
−
−
cation RTILs P14+/NTf2 and P1EOE+/NTf2 . Preliminary
␶ ͒ arises from predominantly cation or anion reorientation,
3
one expects to find a linear relationship between ␶3 and either
analysis shows that the average relaxation time constant
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-7
Ultrafast dynamics of ionic liquids
J. Chem. Phys. 122, 184512 ͑2005͒
TABLE II. Intramolecular vibrational band frequencies observed in Fourier-transform Kerr spectra of ionic liquids. In the Assignment column, A denotes an
anion vibration and C a cation vibration.
−
−
−
+
+
P14+ /NTf2 ͑cm−1
͒
P1EOM+ /NTf2 ͑cm−1
͒
P1EOE+ /NTf2 ͑cm−1
͒
P1EOE /Br−͑cm−1
͒
P1EOE /DCA−͑cm−1
͒
Assignment
121
168
121
168
121
168
A
A
A
C
A
A
C
A
A
A
A
A
A
A
C
C
A
A
C
C
C
C
A
A
A
A
C
C
C
A
A
180
282
203
284
208
278
208
278
208
278
288
297
313
326
331
340
350
371
288
297
313
326
331
341
350
288
297
313
326
332
341
350
381
380
399
407
398
408
399
408
410
440
411
439
440
475
435
433
526
528
534
551
557
571
588
630
535
551
557
571
589
535
551
557
571
587
587
641
587
643
663
740
741
740
TABLE III. Formula weight MW, density d, formula volume V, and viscosity ␩ of the five pyrrolidinium ionic
liquids.
c
Ionic liquid
P1EOE+ /NTf2
Constituent
MW͑g/mol͒
da͑g/cm3͒
Vb͑Å3͒
␩ ͑cP͒
−
438.4
158.3
280.1
1.41
378
175
203
60
+
P1EOE
−
NTf2
−
P1EOM+ /NTf2
424.4
144.3
280.1
1.46
1.40
1.31
1.10
361
158
203
59
85
+
P1EOM
−
NTf2
−
P14+ /NTf2
422.4
142.3
280.1
372
169
203
+
P14
−
NTf2
+
P1EOE /Br−
238.2
158.3
79.9
206
175
31
11 940
480
+
P1EOE
Br−
+
P1EOE /DCA−
219.8
158.3
66.0
231
175
56
+
P1EOE
DCA−
aValues at 295 K.
bEstimated by van der Waals increments ͑Refs. 61 and 62͒.
cValues at 296 K.
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-8
Shirota et al.
J. Chem. Phys. 122, 184512 ͑2005͒
from fluorescence Stokes shift measurements seems to be
substantially slower than the slowest observed Kerr relax-
ation time constant ␶3.
+
The bands observed in the P1EOE /Br− spectrum are as-
+
signed to the intramolecular vibrational bands of the P1EOE
cation because the monatomic Br− ion has no vibrational
+
modes. Comparing the spectra of P1EOE /Br− and
+
+
P1EOE /DCA−, we find that P1EOE /DCA− shows a strong
vibrational mode at 180 cm−1. Lower intensity intramolecu-
+
lar modes for P1EOE /DCA− are observed in the
200–750 cm−1 range and are quite similar to those of
+
P1EOE /Br−, except for the 663 cm−1 band which is present
+
only in P1EOE /DCA−. The normal modes calculated at the
B3LYP/6-31+G͑d,p͒ level, listed in Table II, show that the
180 cm−1 DCA− band is the cyano-N-cyano bending mode
and the 663 cm−1 band is the N-cyano bending mode.
−
The ionic liquids having the NTf2 anion display a rich
vibrational structure, as seen in Fig. 4. The intramolecular
−
vibrational bands of the NTf2− anion dominate the spectra of
−
−
P1EOE+/NTf2 , P1EOM+/NTf2 , and P14+/NTf2 because the
−
NTf2 anion contribution to the Raman spectral density is
more than an order of magnitude greater than from the pyr-
rolidinium cations. For each of the pyrrolidinium NTf2− ionic
liquid spectra, there are ten intense vibrational transitions in
the 250–420 cm−1 frequency range ͑Table II͒. An electronic
structure calculation at the B3LYP/6-31G+͑d,p͒ level for a
gas-phase C2 ͑trans͒ symmetry NTf2− anion shows only eight
intramolecular vibrational bands in the corresponding fre-
quency range. The inconsistency between the number of
bands observed in the experimental liquid spectra and the
density functional theory normal mode calculations likely
FIG. 5. Multimode Brownian oscillator fits to the intermolecular spectrum
+
of P1EOE /Br− from 0 to 200 cm−1. The bottom figure presents the best fit
using three Brownian oscillators, the middle figure the best fit with two
Brownian oscillators, and the residuals for both fits are shown at top.
One of the primary challenges in the interpretation of
low-frequency Kerr spectra is to fully understand the inter-
mediate time scale dynamics bridging the intermolecular vi-
brational to diffusive time scales. In a pure liquid, the pro-
gression of time scales in the observed dynamics begins at
the shortest time scales with the intramolecular vibrations,
followed by underdamped intermolecular vibrations ͑includ-
ing librational motions͒, with the longer time scales com-
prised of a quasiexponential decay for the intermediate time
dynamics, with the longest time scale relaxation being rep-
resented in the simplest case by a single exponential decay
that is correlated with diffusive reorientational motions. De-
tailed tests of the models have been made for two different
cases. Loughnane et al. showed that the intermediate time
scale exponential decay can be described as resulting from
motional narrowing of intermolecular oscillators.75 They
showed that this explanation was consistent for six symmet-
ric top liquids for a wide variety of temperatures. On the
other hand, McMorrow et al. showed that a single “quan-
tum” harmonic oscillator reproduces the dynamics of CS2
over a range of dilution concentrations for which the chemi-
cal environment changes substantially ͑as evidenced by the
change in observed linewidth͒.76,77 The quantum harmonic
oscillator has the advantage that the dephasing rate is not
coupled to the frequency of the oscillator. A limitation of the
“classical” harmonic oscillator model is that it exhibits fre-
quency pulling, and hence distortions of the lineshapes be-
cause of the strong overlap between inhomogeneous spectral
widths and dynamical time scales. For the room temperature
ionic liquids discussed here, the low-frequency spectra mea-
−
arises because the flexible NTf2 anion has multiple confor-
mations that lead to shifts in some of the vibrational bands.
We note that the Kerr spectra of imidazolium ionic liquids
with the NTf2− anion also show ten strong vibrational modes
in the same frequency region,41 in common with the pyrro-
−
lidinium NTf2 ionic liquid spectra presented here. Holbrey,
Reichert, and Rogers have reported crystal structures of alky-
−
limidazolium salts, with the NTf2 anion having the cis
conformation.72 We find overall agreement with the anion
and cation vibrational spectra with the Raman spectra of
73
+
P13+/NTf2− and P13 /I− reported recently by Castriota et al.
and with the IR and Raman spectra of NTf2− reported by Rey
et al.74
In order to emphasize the intermolecular vibrational dy-
namics, Fig. 6 shows the Kerr spectra from Fig. 4 plotted
over a narrower spectral window of 0–200 cm−1. The low-
frequency dynamics arise from several types of intermolecu-
lar interactions including collision-induced and dipole-
induced motions with the dominant contributions likely
arising from intermolecular librational motions. As shown in
Figs. 5 and 6, the broad low-frequency spectra of the
−
−
−
P1EOE+/NTf2 , P1EOM+/NTf2 , and P14+/NTf2 liquids are
quite similar. However, the low-frequency spectra of
+
+
P1EOE /Br− and P1EOE /DCA− are quite distinct. The inter-
−
molecular dynamics and interactions between NTf2 and
the three pyrrolidinium cations lead to similar spectra be-
cause of the structural and dynamical similarities of the
+
P1EOE+, P1EOM+, and P14 cations.
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-9
Ultrafast dynamics of ionic liquids
J. Chem. Phys. 122, 184512 ͑2005͒
The low-frequency intermolecular dynamics in the pyr-
rolidinium RTILs are analyzed by line shape fitting of the
broad spectral band to a multimode Brownian oscillator
model,78,79
3
aBOi
␻
IBOi͑␻͒ =
,
2
͑3͒
͚
2
2
2 2
2␲͓͑␻
− ␻ ͒ + ␥BOi ␻ ͔
i=1
BOi
where aBOi, ␻BOi, and ␥BOi are the amplitude, the frequency,
and the damping parameters for the ith Brownian oscillator,
respectively. Though the linewidth coefficient ␥BOi was origi-
nally assumed to be frequency dependent,78,79 here it is as-
sumed to be constant for simplicity. Fits to the intermolecular
spectral bands using two-mode versus three-mode Brownian
+
oscillators are shown in Fig. 5 for P1EOE /Br−. For the two-
mode model, the fit curve deviates somewhat from the data
and the fit residuals are noticeably worse than for the three-
mode model for which we determined that adequate fits were
obtained. We note that even small molecular liquids normally
show at least two features in fits to the intermolecular Kerr
spectrum. For the more complex ionic liquid case, it is plau-
sible that even more than three modes could be present,
likely with substantial overlap between modes. At least three
Brownian oscillators are needed to obtain good fits to the
intermolecular dynamics of the pyrrolidinium RTILs. We
note that Giraud et al. used three Brownian oscillators to
describe the intermolecular Kerr spectra of five imidazolium
cation ionic liquids.41 A Lorentzian function is added to the
model to account for the intramolecular bands observed at
+
180 cm−1 for P1EOE /DCA− and at 122 cm−1 for
−
−
−
P1EOE+/NTf2 , P1EOM+/NTf2 , and P14+/NTf2 . Figures 5
and 6 show the low-frequency spectra and Brownian oscilla-
tor fits. The fit parameters are summarized in Table IV. The
sum of the spectral area ABOi=͐IBOi͑␻͒d␻/͚ ͐IBOi͑␻͒d␻ for
i
the Brownian oscillators is normalized to unity.
−
From the line shape analysis of the three NTf2 anion
RTILs shown in Fig. 6, one finds that the band center fre-
quencies ␻ of about 28, 45, and 72 cm−1 are very similar
BOi
−
−
−
for P14+/ NTf2 , P1EOM+/NTf2 and P1EOE+/NTf2 . The
+
P1EOE /Br− intermolecular spectrum arises predominantly
+
from P1EOE cation dynamics, because the spherical bromide
ion has no reorientational contribution to the depolarized
OHD-RIKES signal. Three Brownian oscillator components
+
centered at 43, 66, and 91 cm−1 are found in the P1EOE /Br−
spectrum, with the band center for the intermolecular spec-
trum shifted to higher frequency, as observed in Fig. 6. In-
−
FIG. 6. Low-frequency power spectra of ͑a͒ P1EOE+ /NTf2
, ͑b͒
P1EOM+ /NTf2−, ͑c͒ P14+ /NTf2−, ͑d͒ P1EOE /Br−, and ͑e͒ P1EOE /DCA− ͑bold
dotted lines͒. Entire fit curves ͑solid lines͒ and each Brownian oscillator fits
͑small dotted, short dashed, and long dashed lines͒ are also shown. Intramo-
lecular modes are also fitted by a Lorentzian function ͑dot-dash lines͒.
+
+
+
termolecular dynamics for P1EOE /DCA− liquids also show a
higher frequency first moment of the spectrum. Brownian
oscillator line shape components are fit to the center frequen-
cies of 34, 64, and 97 cm−1. We note that the reduced mass
−
between cation and anion for the NTf2 liquids is approxi-
+
mately two times greater than for either of the P1EOE /Br− or
+
P1EOE /DCA− liquids. Thus a spectral component arising
sured in the 0–150 cm−1 range arise from intermolecular dy-
namics from both the cations and anions. The complexity of
the observed ionic liquid spectra does not permit us to dis-
tinguish between the motional narrowing or quantum har-
monic oscillator models in the Kerr dynamics observed for
the pyrrolidinium RTILs.
from a vibrational coordinate between cation and anion
could account for the observed higher band center frequen-
cies for the latter two liquids.
For the five pyrrolidinium ionic liquids considered here,
the case of the intermolecular vibrations is simplest to con-
+
sider for P1EOE /Br− because of the spherical bromide anion.
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-10
Shirota et al.
J. Chem. Phys. 122, 184512 ͑2005͒
TABLE IV. Brownian oscillator fit parameters for the low-frequency power spectra of the pyrrolidinium ionic liquids.
a
b
c
b
d
d
b
d
d
Ionic liquid
ABO1
␻
͑cm−1
͒
␥
͑cm−1
͒
ABO2
␻
͑cm−1
͒
␥
͑cm−1
͒
ABO3
␻
͑cm−1
͒
␥
BO3
͑cm−1
͒
BO1
BO1
BO2
BO2
BO3
−
P1EOE+ /NTf2
0.23
0.17
0.21
0.27
0.17
28
29
28
43
34
61
71
68
72
76
0.44
0.45
0.55
0.40
0.48
47
41
45
66
64
76
81
86
62
79
0.33
0.38
0.24
0.32
0.35
74
73
69
91
97
65
73
61
57
72
−
P1EOM+ /NTf2
P14+ /NTf2
−
P1EOE+ /Br−
P1EOE+ /DCA−
aError is ±30%.
bError is ±15%.
cError is ±20%.
dError is ±10%.
The only way in which the Br− motions can couple to the
vibrational coordinates that should result in more clearly de-
fined modes than are observed for the intermolecular inter-
actions found between aprotic organic liquid molecules.
These proposed vibrational coordinates between cations and
anions would likely overlap with both the translational
͑interaction-induced͒ and orientational ͑librational͒ compo-
nents of the Kerr spectrum. It seems clear that careful mo-
lecular dynamics simulations followed by calculation of the
Kerr spectra will be required to obtain any more definitive
assignments of the low-frequency spectra of ionic liquids.
Our conclusion is that the clearest contributions from a line
shape analysis are that one obtains an accurate parametriza-
tion of the line shape and a reasonably good picture about the
minimum number of different modes that contribute to the
intermolecular line shape.
The intermolecular dynamics for the five pyrrolidinium
RTILs studied here are qualitatively similar to those ob-
served for the imidazolium cation based RTILs studied by
Giraud et al.41 and Rajian et al.40 Despite the qualitative
similarity in intermolecular dynamics for the two classes of
RTILs, the solvent reorganization dynamics probed using
solvatochromic fluorescent dyes very different. For each of
the imidazolium-cation RTILs investigated thus far, the sol-
vation dynamics present an ultrafast decay component that is
qualitatively assigned to librational motions of the aromatic
imidazolium cation. Solvation dynamics of RTILs having the
same anions but with the aromatic cation replaced by an
aliphatic tetraalkylammonium, dialkylpyrrolidinium, or tet-
raalkylphosphonium cation show slower, glassylike relax-
ation with long time scale solvation time constants in the
range from 50 ps to 50 ns.54,66–71,97
+
observed Kerr signal is if P1EOE /Br− interactions lead to
+
substantial changes to the P1EOE angular momentum. The
complex nature of ion-ion structure and interactions in ionic
liquids leads us to consider the possibility that low-frequency
cation-anion modes may contribute to the intermolecular
spectrum along with librations and interaction-induced mo-
tions. Visualization of the normal modes for AM1 gas-phase
+
semiempirical cluster calculations on the P1EOE /Br− dimer
+
and the Br−/P1EOE /Br− trimer gave preliminary evidence
that several of the numerous low-frequency torsional modes
of the cation are strongly coupled to bromide translations,
leading to effective cation-anion stretching coordinates.
Several line shape models have been used to analyze the
low-frequency intermolecular part of OHD-RIKES spectrum
for molecular liquids.33,34,38,50,80–91 The low-frequency inter-
molecular spectrum of liquids contains the interaction-
induced ͑translational͒ and librational ͑reorientational͒
motions.92–96 The recent study by Ryu and Stratt on the in-
termolecular dynamics and calculated Kerr spectra of ben-
zene and CS2 liquids provides insight for how we may con-
sider ionic liquids in a way that goes beyond line shape
analyses.96 They demonstrated that significant overlap occurs
between translational and orientational components of the
calculated Kerr low-frequency spectrum in the range from 0
to 200 cm−1. We believe that similar overlap and coupling
between translational and orientational modes will also occur
for ionic liquids in general. Our geometry optimized gas-
phase structures for the three pyrrolidinium cations and two
molecular anions show that each ion is an asymmetric rotor.
Structures of these ions are actually surprisingly close to
symmetric rotors, showing a crude similarity to the rotational
symmetry of benzene in the sense that they have two very
similar inertial axes for the prolate rotors. In order to unravel
the complex intermolecular spectrum in a binary system such
as an ionic liquid, we must recognize that there will be low-
frequency interaction-induced contributions from both cation
and anion, as well as cross terms. Because of the near sym-
metric top nature for both the cation and anion geometries,
there will exist librational motions on two different time
scales. Because of the asymmetric nature of the ions, we
expect depolarized contributions from all modes. We believe
that the strong electrostatic interactions between ions will
lead to a substantially higher degree of clustering for ionic
liquids compared to neutral organic liquids. The specific at-
tractive associations between cations and anions will lead to
It is well known that the dipole moment for an ion is not
a very meaningful quantity, since its magnitude depends on
the choice of origin for the atomic coordinates. Wynne and
co-workers avoid this difficulty by estimating cation effec-
tive dipole moments from chemically and structurally similar
neutrals.41 Kobrak has described the how the concept of a
“charge arm” can be used in place of the dipole moment for
describing polarity in ionic liquids.98 The charge arm is de-
fined as the charge times distance separating the center of
charge and the center of mass. This picture may qualitatively
explain the similarity of low-frequency dynamics between
−
+
NTf2 ionic liquids with P14 and the aromatic analog
1-butyl-3-methylimidazolium cations because the charge-
+
arms for P14 and 1-butyl-3-methylimidazolium cation have
similar values of 0.9 and 1.0 e Å, respectively.98
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-11
Ultrafast dynamics of ionic liquids
J. Chem. Phys. 122, 184512 ͑2005͒
Much progress has been made in understanding the
properties of room temperature ionic liquids using molecular
dynamics simulations. To date, only dialkylimidazolium-
cation liquids have been studied. Nonpolarizable potential
force fields have been developed using both CHARMM99
and AMBER100,101 models, with one work reporting en-
hanced accuracy by use of a polarizable potential.102 Funda-
mental physical properties can be effectively studied with
molecular dynamics simulations,99–101,103,104 as well as solva-
tion energies105–108 and dynamics.64,65,109 It will be quite ben-
eficial to compare experiments and theory for optical Kerr
dynamics in ionic liquids, as well as to parametrize standard
potentials for both nonpolarizable and polarizable potentials
for nonaromatic cations and anions such as pyrrolidiniums
partment of Energy and supported by its Division of Chemi-
cal Sciences, Office of Basic Energy Sciences. Support from
the BNL LDRD Program ͑Project No. 03-118͒ is acknowl-
edged. The authors thank Mark Kobrak, Elina Trofimovsky,
and Kimberly Odynocki for helpful discussions and assis-
tance in the preparation of the ionic liquids.
1S. A. Forsyth, J. M. Pringle, and D. R. MacFarlane, Aust. J. Chem. 57,
113 ͑2004͒.
2T. Welton, Chem. Rev. ͑Washington, D.C.͒ 99, 2071 ͑1999͒.
3M. J. Earle and K. R. Seddon, Pure Appl. Chem. 72, 1398 ͑2000͒.
4P. Wasserscheid and W. Keim, Angew. Chem., Int. Ed. 39, 3772 ͑2000͒.
5T. Kitazume, J. Fluorine Chem. 105, 265 ͑2000͒.
6G. W. V. Cave, C. L. Raston, and J. L. Scott, Chem. Commun. ͑Cam-
bridge͒ 2001, 2159.
7J. Dupont, R. F. de Souza, and P. A. Z. Suarez, Chem. Rev. ͑Washington,
D.C.͒ 102, 3667 ͑2002͒.
−
and NTf2 .
8H. Zhao and S. V. Malhotra, Aldrichimica Acta 35, 75 ͑2002͒.
9J. S. Willkes, Green Chem. 4, 73 ͑2002͒.
10J. H. Davis, Jr., Chem. Lett. 33, 1072 ͑2004͒.
V. CONCLUSIONS
11Y. Chauvin, L. Mußmann, and H. Olivier, Angew. Chem., Int. Ed. Engl.
34, 2698 ͑1995͒.
The ultrafast dynamical aspects of the five pyrrolidinium
ionic liquids have been investigated using femtosecond
optical-heterodyne-detected Kerr effect spectroscopy. The pi-
cosecond relaxation processes of the five pyrrolidinium ionic
liquids show nonexponential behavior, with the best fits ob-
tained using a triexponential model. The fast and intermedi-
ate relaxation time constants are about 2 and 20 ps and are
almost same for each of the five ionic liquids. The slow
relaxation time constant for the ionic liquids correlates with
the product of the shear viscosity times the sum of anion and
cation volumes, so is roughly consistent with the Stokes–
Einstein–Debye hydrodynamic model for orientational fric-
12P. A. Z. Suarez, J. E. L. Dullius, S. Einloft, R. F. de Souza, and J. Dupont,
Polyhedron 15, 1217 ͑1996͒.
13C. E. Song and E. J. Roh, Chem. Commun. ͑Cambridge͒ 2000, 837.
14W. Chen, L. Xu, C. Chatterton, and J. Xiao, Chem. Commun. ͑Cambridge͒
1999, 1247.
15C. de Bellefon, E. Pollet, and P. Grenouillet, J. Mol. Catal. A: Chem. 145,
121 ͑1999͒.
16K. Jin, X. Y. Huang, L. Pang, J. Li, A. Appel, and S. Wherland, Chem.
Commun. ͑Cambridge͒ 23, 2872 ͑2002͒.
17M. Eckstein, P. Wasserscheid, and U. Kragl, Biotechnol. Lett. 24, 763
͑2002͒.
18A. L. Monteiro, F. K. Zinn, R. F. de Souza, and J. Dupont, Tetrahedron:
Asymmetry 8, 177 ͑1997͒.
19A. Hu, H. L. Ngo, and W. Lin, Angew. Chem., Int. Ed. 43, 2501 ͑2004͒.
20A. Berthod, L. He, and D. W. Armstrong, Chromatographia 53, 63 ͑2001͒.
21D. W. Armstrong, L. K. Zhang, L. F. He, and M. L. Gross, Anal. Chem.
73, 3679 ͑2001͒.
−
tion. The observed viscosities of the pyrrolidinium NTf2
and DCA− ionic liquids indicate that these fluids are well
above the glass transition at ambient temperatures. It appears
that simple hydrodynamic scaling may be sufficient to ex-
22S. Carda-Broch, A. Berthod, and D. W. Armstrong, Rapid Commun. Mass
Spectrom. 17, 553 ͑2003͒.
plain the faster reorientation of P1EOE+/NTf2 and
−
23Z. B. Alfassi, R. E. Huie, B. L. Milman, and P. Neta, Anal. Bioanal. Chem.
377, 159 ͑2003͒.
P1EOM+/NTf2 relative to P14+/NTf2 because of the relative
−
−
24D. McMorrow, W. T. Lotshaw, and G. A. Kenney-Wallace, IEEE J. Quan-
tum Electron. 24, 443 ͑1988͒.
differences in viscosity.
The Fourier-transform Kerr spectra of the five pyrroli-
dinium ionic liquids have also been obtained. The spectra of
25W. T. Lotshaw, D. McMorrow, N. Thantu, J. S. Melinger, and R. Kitch-
enham, J. Raman Spectrosc. 26, 571 ͑1995͒.
26D. McMorrow and W. T. Lotshaw, Chem. Phys. Lett. 174, 85 ͑1990͒.
27D. McMorrow and W. T. Lotshaw, J. Phys. Chem. 95, 10395 ͑1991͒.
28S. Kinoshita, Y. Kai, T. Ariyoshi, and Y. Shimada, Int. J. Mod. Phys. B 10,
1229 ͑1996͒.
−
−
the ionic liquids with the NTf2 anion ͑P1EOE+/NTf2 ,
P1EOM+/NTf2 , and P14+/NTf2 ͒ are quite similar. The in-
−
−
+
tramolecular vibrational bands of P1EOE /Br− and
+
P1EOE /DCA− are similar except for the addition of the
29E. W. Castner, Jr. and M. Maroncelli, J. Mol. Liq. 77, 1 ͑1998͒.
30N. A. Smith and S. R. Meech, Int. Rev. Phys. Chem. 21, 75 ͑2002͒.
31B. J. Loughnane, R. A. Farrer, A. Scodinu, T. Reilly, and J. T. Fourkas, J.
Phys. Chem. B 104, 5421 ͑2000͒.
DCA− bending modes. The broad intermolecular spectra of
the ionic liquids have been analyzed by the multimode
Brownian oscillator model. The intermolecular spectral den-
32R. A. Farrer and J. T. Fourkas, Acc. Chem. Res. 36, 605 ͑2003͒.
33N. T. Hunt, A. A. Jaye, and S. R. Meech, J. Phys. Chem. B 107, 3405
͑2003͒.
sities of P1EOE+/NTf2 , P1EOM+/NTf2 , and P14+/NTf2
−
−
−
+
show maxima at lower frequencies than those of P1EOE /Br−
34N. T. Hunt, A. A. Jaye, A. Hellman, and S. R. Meech, J. Phys. Chem. B
108, 100 ͑2004͒.
+
and P1EOE /DCA−. A sum of three Brownian oscillators pro-
35G. Giraud and K. Wynne, J. Am. Chem. Soc. 124, 12110 ͑2002͒.
36G. Giraud, J. Karolin, and K. Wynne, Biophys. J. 85, 1903 ͑2003͒.
37J. D. Eaves, C. J. Fecko, A. L. Stevens, P. Peng, and A. Tokmakoff, Chem.
Phys. Lett. 376, 20 ͑2003͒.
vides a good fit to the intermolecular line shapes for each of
the five ionic liquids studied, though we note that this model
may underestimate the complexity of the interactions occur-
ring in such a binary ionic system.
38H. Shirota and E. W. Castner, Jr., J. Am. Chem. Soc. 123, 12877 ͑2001͒.
39B. R. Hyun, S. V. Dzyuba, R. A. Bartsch, and E. L. Quitevis, J. Phys.
Chem. A 106, 7579 ͑2002͒.
ACKNOWLEDGMENTS
40J. Rajian, S. Li, R. Bartsch, and E. Quitevis, Chem. Phys. Lett. 393, 372
͑2004͒.
Work done at Rutgers was supported by the National
Science Foundation ͑Grant No. CHE-0239390͒. Part of this
work was carried out at Brookhaven National Laboratory
under Contract No. DE-AC02-98CH10886 with the U.S. De-
41G. Giraud, C. M. Gordon, I. R. Dunkin, and K. Wynne, J. Chem. Phys.
119, 464 ͑2003͒.
42H. Cang, J. Li, and M. D. Fayer, J. Chem. Phys. 119, 13017 ͑2003͒.
43A. M. Funston, E. Trofimovsky, K. Odynocki, M. N. Kobrak, and J. F.
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

184512-12
Shirota et al.
J. Chem. Phys. 122, 184512 ͑2005͒
Wishart ͑unpublished͒.
75B. J. Loughnane, A. Scodinu, R. A. Farrer, J. T. Fourkas, and U. Mohanty,
J. Chem. Phys. 111, 2686 ͑1999͒.
44P. Wasserscheid and T. Welton, Ionic Liquids in Synthesis ͑Wiley-VCH,
Weinheim, 2003͒.
76D. McMorrow, N. Thantu, V. Kleiman, J. S. Melinger, and W. T. Lotshaw,
J. Phys. Chem. A 105, 7960 ͑2001͒.
45D. R. MacFarlane, P. Meakin, J. Sun, N. Amini, and M. Forsyth, J. Phys.
Chem. B 103, 4164 ͑1999͒.
77D. McMorrow, W. T. Lotshaw, J. S. Melinger, and V. Kleiman, ACS
Symp. Ser. 820, 14 ͑2002͒.
46N. L. Lancaster, P. A. Salter, T. Welton, and G. B. Young, J. Org. Chem.
67, 8855 ͑2002͒.
78Y. Tanimura and S. Mukamel, J. Chem. Phys. 99, 9496 ͑1993͒.
79S. Mukamel, Principles of Nonlinear Optical Spectroscopy ͑Oxford Uni-
versity Press, New York, 1995͒.
47D. R. MacFarlane, J. Sun, J. Golding, P. Meakin, and M. Forsyth, Elec-
trochim. Acta 45, 1271 ͑2000͒.
48D. R. MacFarlane, S. A. Forsyth, J. Golding, and G. B. Deacon, Green
Chem. 4, 444 ͑2002͒.
80Y. J. Chang and E. W. Castner, J. Phys. Chem. 98, 9712 ͑1994͒.
81Y. J. Chang and E. W. Castner, Jr., J. Phys. Chem. 100, 3330 ͑1996͒.
82N. A. Smith, S. Lin, S. R. Meech, and K. Yoshihara, J. Phys. Chem. A
101, 3641 ͑1997͒.
49D. R. MacFarlane, J. Golding, S. Forsyth, M. Forsyth, and G. B. Deacon,
Chem. Commun. ͑Cambridge͒ 2001, 1430.
50P. P. Wiewior, H. Shirota, and E. W. Castner, Jr., J. Chem. Phys. 116, 4643
͑2002͒.
83N. A. Smith, S. Lin, S. R. Meech, H. Shirota, and K. Yoshihara, J. Phys.
Chem. A 101, 9578 ͑1997͒.
51M. J. Frisch, G. W. Trucks, H. B. Schlegel et al., GAUSSIAN 03, Gaussian,
Inc., Pittsburgh, PA, 2003.
84H. Shirota, K. Yoshihara, N. A. Smith, S. Lin, and S. R. Meech, Chem.
Phys. Lett. 281, 27 ͑1997͒.
52A. D. Becke, J. Chem. Phys. 98, 5648 ͑1993͒.
53C. Lee, W. Yang, and R. G. Parr, Phys. Rev. B 37, 785 ͑1988͒.
54N. Ito, S. Arzhantsev, M. Heitz, and M. Maroncelli, J. Phys. Chem. B
108, 5771 ͑2004͒.
85N. A. Smith and S. R. Meech, J. Phys. Chem. A 104, 4223 ͑2000͒.
86N. T. Hunt and S. R. Meech, J. Chem. Phys. 120, 10828 ͑2004͒.
87S. Palese, J. T. Buontempo, L. Schlling, W. T. Lotshaw, Y. Tamimura, S.
Mukamel, and R. J. Dwayne Miller, J. Phys. Chem. 98, 12466 ͑1994͒.
88S. Palese, L. Schilling, R. J. D. Miller, P. R. Staver, and W. T. Lotshaw, J.
Phys. Chem. 98, 6308 ͑1994͒.
55B. Mennucci and J. Tomasi, J. Chem. Phys. 106, 5151 ͑1997͒.
56B. Mennucci, E. Cances, and J. Tomasi, J. Phys. Chem. B 101, 10506
͑1997͒.
89M. Cho, M. Du, N. F. Scherer, G. R. Fleming, and S. Mukamel, J. Chem.
Phys. 99, 2410 ͑1993͒.
57See EPAPS Document No. E-JCPSA6-122-506519 for details of the
GAUSSIAN 03 calculation results. Optimized geometries, electrostatic param-
eters, and rotational constants, and CHelpG charges are tabulated. This
document can be reached via a direct link in the online article’s HTML
reference section or via the EPAPS homepage ͑http://www.aip.org/
pubservs/epaps.html͒.
90P. Bartolini, M. Ricci, R. Trroe, R. Righini, and I. Santa, J. Chem. Phys.
110, 8653 ͑1999͒.
91H. Shirota, J. Chem. Phys. 122, 044514 ͑2005͒.
92P. A. Madden and T. I. Cox, Mol. Phys. 43, 287 ͑1981͒.
93P. A. Madden, in Ultrafast Phenomena IV, edited by A. H. Auston and K.
B. Eisenthal ͑Springer, Berlin, 1984͒, Vol. 38, p. 244.
94L. C. Geiger and B. M. Ladanyi, J. Chem. Phys. 89, 6588 ͑1988͒.
95L. C. Geiger and B. M. Ladanyi, Chem. Phys. Lett. 159, 413 ͑1989͒.
96S. Ryu and R. M. Stratt, J. Phys. Chem. B 108, 6782 ͑2004͒.
97P. K. Chowdhury, M. Halder, L. Sanders, T. Calhoun, J. L. Anderson, D.
W. Armstrong, X. Song, and J. W. Petrich, J. Phys. Chem. B 108, 10245
͑2004͒.
58B. J. Loughnane and J. T. Fourkas, J. Phys. Chem. B 102, 10288 ͑1998͒.
59D. Kivelson and P. A. Madden, Annu. Rev. Phys. Chem. 31, 523 ͑1980͒.
60G. R. Fleming, Chemical Applications of Ultrafast Spectroscopy ͑Oxford
University Press, New York, 1986͒.
61J. T. Edwards, J. Chem. Educ. 47, 261 ͑1970͒.
62Because the values for S of NTf2 and N- for NTf2 and DCA− are not
listed, the value for S is used as 2ϫSϭand the value for N is tentatively
used as ϾN-.
−
−
98M. N. Kobrak and N. Sandalow, in Molten Salts XIV, edited by R. A.
Mantz ͑The Electrochemical Society, Pennington, NJ, 2005͒.
99T. I. Morrow and E. J. Maginn, J. Phys. Chem. B 106, 12807 ͑2002͒.
100J. de Andrade, E. S. Boes, and H. Stassen, J. Phys. Chem. B 106, 13344
͑2002͒.
63J. N. Israelachvili, Intermolecular and Surface Forces, 2nd ed. ͑Academic,
London, 1992͒.
64M. N. Kobrak and V. Znamenskiy, Chem. Phys. Lett. 395, 127 ͑2004͒.
65C. J. Margulis, Mol. Phys. 102, 829 ͑2004͒.
66R. Karmakar and A. Samanta, J. Phys. Chem. A 106, 4447 ͑2002͒.
67R. Karmakar and A. Samanta, J. Phys. Chem. A 107, 7340 ͑2003͒.
68D. Chakravarty, P. Harza, A. Chakravarty, D. Seth, and N. Sarkar, Chem.
Phys. Lett. 381, 697 ͑2003͒.
101C. G. Hanke, S. L. Price, and R. M. Lynden-Bell, Mol. Phys. 99, 801
͑2001͒.
102T. Y. Yan, C. J. Burnham, M. G. Del Popolo, and G. A. Voth, J. Phys.
Chem. B 108, 11877 ͑2004͒.
69S. Arzhantsev, N. Ito, M. Heitz, and M. Maroncelli, Chem. Phys. Lett.
381, 278 ͑2003͒.
103M. G. Del Popolo and G. A. Voth, J. Phys. Chem. B 108, 1744 ͑2004͒.
104C. J. Margulis, H. A. Stern, and B. J. Berne, J. Phys. Chem. B 106,
12017 ͑2002͒.
70J. A. Ingram, R. S. Moog, N. Ito, R. Biswas, and M. Maroncelli, J. Phys.
Chem. B 107, 5926 ͑2003͒.
105V. Znamenskiy and M. N. Kobrak, J. Phys. Chem. B 108, 1072 ͑2004͒.
106C. G. Hanke, A. Johansson, J. B. Harper, and R. M. Lynden-Bell, Chem.
Phys. Lett. 374, 85 ͑2003͒.
71N. Ito, S. Arzhantsev, and M. Maroncelli, Chem. Phys. Lett. 396, 83
͑2004͒.
72J. D. Holbrey, W. M. Reichert, and R. D. Rogers, J. Chem. Soc. Dalton
Trans. 2004, 2267.
107C. G. Hanke and R. M. Lynden-Bell, J. Phys. Chem. B 107, 10873
͑2003͒.
73M. Castriota, T. Caruso, R. G. Agostino, E. Cazzanelli, W. A. Henderson,
and S. Passerini, J. Phys. Chem. A 109, 92 ͑2005͒.
74I. Rey, P. Johansson, J. Lindgren, J. C. Lassegues, J. Grondin, and L.
Servant, J. Phys. Chem. A 102, 3249 ͑1998͒.
108J. B. Harper and R. M. Lynden-Bell, Mol. Phys. 102, 85 ͑2004͒.
109Y. Shim, J. S. Duan, M. Y. Choi, and H. J. Kim, J. Chem. Phys. 119,
6411 ͑2003͒.
Downloaded 16 Dec 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions