Structural properties of 1-alkyl-3-methylimidazolium bis{(trifluoromethyl) sulfonyl}amide ionic liquids: X-ray diffraction data and molecular dynamics simulations

Article abstract of DOI:10.1021/jp1093299

X-ray diffraction data for 1-alkyl-3-methylimidazolium bis{(trifluoromethyl)sulfonyl}amides are reported as a function of the length of the alkyl chain on the imidazolium ring. The measured diffraction patterns have been compared with the theoretical patterns calculated (from the geometries obtained) with molecular dynamics simulations. This provides a detailed description (at the atomistic level) of the morphology in the liquid state of these salts, highlighting the role played by the alkyl chain length. An analysis of the behavior of the hydrogen bonds that are formed between the imidazolium acidic protons and the anion is presented.

Full text of DOI:10.1021/jp1093299

16398
J. Phys. Chem. B 2010, 114, 16398–16407
Structural Properties of 1-Alkyl-3-methylimidazolium Bis{(trifluoromethyl)sulfonyl}amide
Ionic Liquids: X-ray Diffraction Data and Molecular Dynamics Simulations
Enrico Bodo,*,† Lorenzo Gontrani,‡ Ruggero Caminiti,† Natalia V. Plechkova,§
Kenneth R. Seddon,§,| and Alessandro Triolo⊥
Department of Chemistry and CNISM, UniVersity of Rome “Sapienza”, Italy, P. A. Moro 5, 00185, Rome,
Italy, Department of Chemistry, UniVersity of Rome “Sapienza”, Italy, QUILL, The Queen’s UniVersity of
Belfast, Belfast BT9 5AG, Northern Ireland U.K., and CNR-Istituto di Struttura della Materia, Area della
Ricerca di Roma 2, Tor Vergata, Rome, Italy
ReceiVed: September 29, 2010; ReVised Manuscript ReceiVed: NoVember 5, 2010
X-ray diffraction data for 1-alkyl-3-methylimidazolium bis{(trifluoromethyl)sulfonyl}amides are reported as
a function of the length of the alkyl chain on the imidazolium ring. The measured diffraction patterns have
been compared with the theoretical patterns calculated (from the geometries obtained) with molecular dynamics
simulations. This provides a detailed description (at the atomistic level) of the morphology in the liquid state
of these salts, highlighting the role played by the alkyl chain length. An analysis of the behavior of the
hydrogen bonds that are formed between the imidazolium acidic protons and the anion is presented.
Introduction
3-methylimidazoliumbis{(trifluoromethyl)sulfonyl}amide([C6mim]-
[NTf2]) as a standard for comparing the variability of experi-
mental measurements carried out by different research groups.16,17
In 2004, Canongia Lopes and Pa´dua18 proposed an optimized
potential for liquid simulations (OPLS-AA) derived force field
for the description of the anion that has proven to provide
reliable quantities when compared to experimental observables.
Ionic liquids1-3 with low volatilities and low melting points
have represented, in the last ten years, a rich field of research
because of their many applications in diverse technological
research areas.4 These substances contain both a cation (which
can, for example, be an asymmetric derivative of the 1-alky-
lpyridinium, tetraalkylphosphonium, 1,3-dialkylimidazolium, or
tetraalkylammonium ions) and an anion, which can be as simple
aschlorideoramorecomplexstructuresuchasbis{(trifluoromethyl)-
sulfonyl}amide [NTf2]-. Their versatility depends on the fact
that their properties can be tuned by varying the cation and the
anion.
Together with the very large number of experimental studies
of ionic liquids,5-8 molecular simulations based on classical
force fields have played a crucial role in providing the link
between the physical and chemical properties and the molecular
structure at the nanoscopic level.9 Since their introduction to
the scientific community,10 the force fields for molecular
dynamics (MD) simulations of ionic liquids have been refined,
expanded11,12 and reparametrized several times (see, for example,
ref 13 and references therein) and have now reached a point in
which molecular dynamics simulations are able to make reliable
predictions of many liquid properties. For example, we have
recently used such force fields to provide an interpretation of
the first X-ray diffraction studies of the [C8mim] Br (where
[Cnmim]+ ) 1-alkyl-3-methylimidazolium, alkyl ) CnH2n+1, 1
< n < 18) ionic liquid.14
Earlier density measurements for these compounds date back
to 1996.15 Various studies have then expanded the set of
measured observables and provided their dependence both on
temperature and in the homologous series of these compounds
obtained by varying the length of the alkyl chain on the
imidazolium cation. A list of the known properties for the title
ionic liquids is reported in Table 1.
It was found by Watanabe et al.26 that, in the series of
[Cnmim][NTf2] ionic liquids, the viscosity increases with
increasing n together with the self-diffusion coefficient. To
explain such counterintuitive experimental data, Urahata and
Ribeiro27 and Wang and Voth28 carried out united-atom and
coarse grained theoretical simulations of imidazolium based
ionic liquids concluding that the apolar, nearly neutral alkyl
chains tend to self-assemble. When the size of the chains is
large enough, both MD simulations and experiments highlighted
that this segregation effect leaves a clear imprint in the
diffraction patterns and in the radial distributions functions
(RDF).25,29,30 Hardacre and co-workers31 explored the structure
of the [C1mim][NTf2] by neutron diffraction and theoretical
simulations, concluding that the liquid phase had peculiar
geometric features: for example, the anion was found to be
predominantly in the trans orientation, whereas the solid phase
is known to have a cis preference; moreover, as opposed to
[C1mim]Cl or [C1mim][PF6], [C1mim][NTf2] is found to show
negligible long-range alternating counterion ordering. More
generally, as opposed to the other mentioned ionic liquids, this
material is characterized by a liquid structure that shows only
limited resemblance to the layered crystalline one. A reason
for such contrasting structures might be the high flexibility of
the [NTf2]- anion, which can be thermally activated in the liquid
state. Other geometrical features of these ionic liquids, such as
The bis{(trifluoromethyl)sulfonyl}amide anion was first
introduced into the ionic liquid literature in 1996 as a basis for
hydrophobic electrolytes for electrochemical studies.15 Its air
and water stability combined with its lack of chemical reactivity
have made it a popular ion for experimental and theoretical
studies to the extent that a IUPAC committee selected a 1-hexyl-
* To whom correspondence should be addressed. E-mail: bodo@caspur.it.
† Department of Chemistry and CNISM, University of Rome “Sapienza”.
‡ Department of Chemistry, University of Rome “Sapienza”, Italy.
§ QUILL, The Queen’s University of Belfast.
| Universidade Nova de Lisboa.
⊥ CNR-Istituto di Struttura della Materia.
10.1021/jp1093299  2010 American Chemical Society
Published on Web 11/22/2010

1-Alkyl-3-methylimidazolium [NTf2]-
J. Phys. Chem. B, Vol. 114, No. 49, 2010 16399
TABLE 1: Measured Properties of [Cnmim][NTf2]
side chain length n
property
reference
1-3
2-10
2-3
density, melting point, viscosity, refractive index, thermal stability
phase transition, density, viscosity, surface tension
density, melting temperatures, phase transition, thermal stability,
heat capacity
15
19
20
1-3, 6,8
3
density, viscosity, diffusion coeff., ionic conductivity
phase transition, density, refractive index, surface tension,
solvatochromic effect
21
22
1-10
small/large angle X-ray scattering
23-25
the relative conformations of the side alkyl chains, were also
analyzed and reported previously.12 Canongia Lopes and Pa´dua32
in 2006 investigated the above-mentioned side chain segregation
effects in [Cnmim][NTf2] ionic liquid and showed the presence
of a network of polar domains (composed of the charged heads
of imidazolium cation and of the anions) and a “continuous
microphase” of nonpolar domains made by the alkyl chains
when n > 2, and attributed to the experimental viscosity behavior
to the presence of these microdomains. While the force field
model of Canongia Lopes and co-workers18 was able to provide
a robust theoretical framework for this kind of topological
model, it failed in reproducing other physicochemical properties
such as heats of vaporization. From reparametrization of the
force field in ref 33 where the authors provided a new set of
Lennard-Jones parameters, the accuracy in the prediction of
several properties, especially densities and enthalpies of vapor-
ization, improved significantly.34 More recently, a theoretical
study35 has provided ab initio optimized gas-phase structures
of several ion pairs in [Cnmim][NTf2] ionic liquids with n ) 4,
6, or 8 and has shown that the size of the lateral alkyl chain
has little effect on the RDF. Elsewhere,23 X-ray diffraction data
have been compared with MD simulations to unravel the
geometric structure of the [C2mim][NTf2] ionic liquid and have
shown the presence of a mixture of two anion conformers (see
Figure 1). Recently, some of us reported a joined small angle
X-ray scattering/Kerr effect spectroscopic study on a series of
[C2mim][NTf2] salts (with 1 < n < 10) highlighting structural
and dynamic features of this family of ionic liquids.24 In
particular, it was shown that only n > 3 alkyl chains tend to
self-assemble into apolar domains. Joint Raman spectroscopic
data and DFT calculations were applied to explore the confor-
mational equilibrium of the anion in [C2mim][NTf2], suggesting
a restricted rotation about the S-N bond leading to a facile
interconversion between cis and trans conformers.37,38
relative radial functions for a series of 1-alkyl-3-methylimida-
zolium bistriflamide salts, where n ) 1, 4, 6, or 8. This study
provided a detailed atomistic level description of the morphology
of these salts in the liquid state, highlighting the role played by
the alkyl chain length.
Experimental Section
1-Alkyl-3-methylimidazoliumbis{(trifluoromethyl)sulfonyl}amides
(alkyl ) butyl, hexyl, octyl) were purchased from Iolitec.39 1,3-
Dimethylimidazolium bis{(trifluoromethyl)sulfonyl}amide was
synthesized according to the following procedure: 1,3-Dimeth-
ylimidazolium methyl sulfate (30.85 g, 0.148 mol) and Li[NTf2]
(43.39 g, 0.151 mol) were dissolved in water (250 cm3) and
stirred for 10 h, during which the mixture separated into two
phases. Dichloromethane (300 cm3) was added to this mixture.
The upper aqueous phase was discarded and the lower dichlo-
romethane layer was washed with water (6 × 200 cm3) and
dried over anhydrous magnesium sulfate (20 g), which was then
removed by filtration. The filtrate was heated in vacuo at 40 °C
to remove the excess solvent. The product was obtained as a
1
transparent paste (46.20 g, 0.122 mol, 81% yield). H NMR
spectra (300 MHz, trichloromethane-d1) showed chemical shifts
δ 3.86 (s, 6H, NCH3), 7.31 (s, 2H, NCHCH), and 8.84 ppm (s,
1H, NCHN) and a melting point (DSC) of 23.35 °C.
The four samples, after drying for 24 h in vacuo at 60 °C, to
reduce the moisture content, were rapidly put into a cell sealed
with Mylar windows. Sample and windows thicknesses were 3
mm. The large angle X-ray scattering (LAXS) experiments were
conducted at room temperature using the noncommercial energy-
scanning diffractometer built in the Department of Chemistry,
Rome University “Sapienza” (Italian Pat. Appl. no. 1126484 -
23 June 1993, Caminiti et al.)40,41 White Bremsstrahlung
radiation emitted by a tungsten tube (50 kV, 40 mA) was used.
The expression for the scattering variable s (transferred mo-
mentum) is
Given the large number of studies on this class of ionic
liquids, we undertook the first combined experimental and
theoretical studies of X-ray diffraction structure factors and the
Figure 1. Cisoid (left) and transoid (right) isomers of the bistriflamide anion, [CF3SO2NSO2CF3]-. Both of these forms were identified
crystallographically.37 In solution there is a rapid interchange between these two.

16400 J. Phys. Chem. B, Vol. 114, No. 49, 2010
4π sin(θ)
Bodo et al.
s )
= E·1.014 sin(θ)
(1)
λ
when E is expressed in keV and s in Å-1
.
The primary beam intensity I0(E) was measured directly, by
reducing the tube current to 10 mA without the sample.
Transmission of the sample was measured under the same
conditions. Both quantities are needed to carry out the necessary
absorption corrections to experimental data. The ultrathin Mylar
cell windows contribution to the diffraction intensity is less than
1/10 000 of the total. After correction of experimental data for
escape peak suppression,42 the various angular data were
normalized to a stoichiometric unit of volume containing one
ion pair and merged to yield the total static structure factor,
I(s), in momentum-space. I(s) is equal to
Figure 2. Sketch of the template structures and the labels of the atoms
used in the FF.
n
TABLE 2: Calculated and Measured Densities in g cm-3 for
[Cnmim][NTf2]a
I(s) ) Ieu
-
x f (s)2
(2)
∑
i i
i)1
n
Canongia Lopes FF
Ko¨ddermann FF
expt
where fi(s) are the interpolated atomic scattering factors,43,44 xi
are the number concentrations of i-type atoms in the stoichio-
metric unit, and Ieu is the observed intensity in electron units.
The function was multiplied by sM(s) where M is an s-dependent
sharpening factor defined in eq 3
1
4
6
8
1.63
1.49
1.40
1.37
1.557
1.438
1.364
1.320
1.43
1.37
a Densities from ref 26.
2
an additional mixing was provided by a short (300 ps) NVT
run at 300 K where the electrostatic interactions were turned
off. The resulting configurations were finally left to evolve in a
constant volume NVT production run of about 5 ns with T )
300 K after an equilibration period. The theoretical densities
are compared with the experimental ones in Table 2.
All the bonds involving the hydrogen atoms have been kept
constrained using the Shake algorithm while no other degrees
of freedom were restricted. The system was treated in a periodic
box of around 703 Å3 containing 550 ionic couples. Nonbonded
interactions were computed up to 9 Å. To test the quality of
this choice, a few hundred additional picosecond runs with with
the cutoff value set to 12.0 Å were performed and we found
that the results were largely insensitive to this choice. The long-
range electrostatic interactions were accounted for by means of
the smoothed particle mesh Ewald (SPME) solver with a
precision of 10-6. The integration time step for the leapfrog
algorithm was 1 fs in the production run and a snapshot of the
configuration was saved each 20 ps to be used in the geometric
analysis.
fN (0)
M(s) )
exp(-0.01s2)
(3)
2
fN (s)
having chosen nitrogen as the sharpening atom. Fourier
transformation of I(s) led to RDF in distance space, D(r):
smax
2r
π
D(r) ) 4πr2F0 +
sI(s) M(s) sin(rs) ds
(4)
∫
0
In this equation, F0 is the bulk number density of stoichiometric
units. We used the value of 19.56 Å -1 (last experimental point)
as the upper limit of integration. Subtraction of the 4πr2F0 term
(corresponding to a uniform distribution) from this formula
allows the isolation of the structural contribution to the
distribution function (Diff(r)). The latter RDF is more suited to
highlight intermolecular (medium-long-range) interactions than
the more commonly used G(r). Detailed description of proce-
dures and formulas used can be found in refs 42, 45, and 46.
Each production run produced a few hundred snapshots that
were then employed in the calculation of the structure function
following the Debye scattering equations:49
Calculations
All-atom MD simulations have been carried out on pure
[Cnmim][NTf2] ionic liquids with n ) 1, 4, 6, or 8 using the
DLPOLY247 package. We have employed two force fields: the
one from Canongia Lopes and Pa´dua11,18,48 and the one from
Ko¨ddermann et al.33 The latter was used only for n ) 4 and 6.
A sketch of the template structures and the labels of the atoms
used in the FF are displayed in Figure 2.
The initial configurations have been generated by randomly
distributing 550 ionic couples in very large simulation cells. A
short (100 ps) isobaric-isothermal (NPT) run at 500 K and with
103 atmospheres pressure was carried out to provide densely
packed, mixed initial configurations. These initial configurations
were left to relax during another NPT run of 300-500 ps at 1
atm and 300 K up to the point in which an approximately
constant density was obtained. The volume was then fixed and
sin(srij)
I(s) )
f (s) f (s) x x
i j
(5)
∑ ∑
i
j
srij
i
j
where fi(s) and fj(s) are the atomic scattering factors and xi and
aj are the number concentrations of i- and j-type atoms in the
stoichiometric unit. From I(s), it is possible to extract the RDF
Diff(r) by means of eq 4. It is important to point out that in the
very low s range (s < 1.2 Å-1) the theoretical I(s) is unreliable
due both to the limited box size and to the Debye formula we
have used. We therefore cut our simulated values at s ∼ 1 Å-1
and smoothly joined them with the experimental low s sets of
data before performing the integration of eq 4.

1-Alkyl-3-methylimidazolium [NTf2]-
J. Phys. Chem. B, Vol. 114, No. 49, 2010 16401
Figure 3. sM(s)I(s) for the four ionic liquids. Left: measured quantities. Right: simulated structure factors. The numbers in the captions are the
densities in g cm-3. The simulations reported here are obtained with Canongia Lopes force field.
Results and Discussion
Canongia Lopes force field and are in fairly good agreement
with the experiments: they reproduce both the intensity and
positions of most of the experimental peaks.
Table 2 reports the MD-computed densities compared with
the experimentally determined values, as a validation of the
simulation. The densities obtained with the Canongia Lopes
force field are larger that the experimental ones, while those
calculated with the Ko¨dderman force field are in much better
agreement.33
As shown in the Introduction, many key structural features
of the present ionic liquids have already been described in the
literature:
1. With a side chain longer than four carbon atoms, chain
aggregation in nonpolar domains is visible in the experi-
mental24 as well as in the theoretical structure factor I(s)
as low-s distinctive peaks.28,32 This aggregation effect may
explain the behavior of experimental quantities such as
viscosity, diffusion coefficients and ionic conductivity;32
2. In the anion, the trans conformation is dominant with
respect to the cis31,37,38 one (see Figure 1);
3. In the side chain, the conformation of the carbon-carbon
dihedral angle becomes more and more sharply defined
as one moves from the carbon near the ring to the end of
the chain: the C-C dihedral angles are predominantly in
the anti conformation;12
In particular, the peaks at 6 and 10 Å-1 are accurately
duplicated, as is the shoulder between 6.5 and 8 Å-1. These are
due to structural correlations in the intramolecular range. Below
4 Å-1 the agreement between the theory and the experiment
deteriorates but it is still possible to match all the peaks positions
although not their intensities. The total I(s) has been presented
in ref 23 for the [C2mim][NTf2] liquid. By visual comparison
of Figure 3 with Figure 2 of ref 23, [C1mim][NTf2] and
[C2mim][NTf2] can be seen to present a similar profile in which
the first two peaks (those centered at ca. 0.9 and 1.5 Å-1) have
similar heights, while for [Cnmim][NTf2] (n g 4) the peak at
0.9 Å-1 is weaker. The low-s region of Figure 3 is reported in
Figure 4 with an expanded s-scale. These experimental data are
in very good agreement with those reported in ref 24. In the
low-s region, we see three distinct peaks: 2.6-2.8, 1.4, and 0.8
Å-1. These peaks are due to structures in the range 2-8 Å in
the direct space. As can be seen, the force fields and the
simulations are accurate enough to reproduce the 2.6-2.8 Å
peak, though the different ionic liquids show a somewhat more
marked difference in the simulated data than in the experimental
counterpart. The second peak at 1.4 Å-1 is well reproduced by
the simulation, which accounts correctly for the different
intensities of the various ionic liquids. Since this peak is due to
structures around 5 Å, it is a good test of the reliability of the
employed force fields in reproducing the structural correlations.
The intensity of this peak is proportional to the length of the
side chain and, therefore, can be probably attributed to the
characteristic 5 Å feature that can be seen in the RDFs already
reported by other authors (see refs 28 and 32) and that has been
attributed to the side chain aggregation effect.27,29
4. The length of the side chain has relatively little effect on
the RDFs. The liquid local structure is due to the
coexistence of various conformers that can be identified
by in vacuo ab initio calculations.35
5. A comparison between LAXS experiments and MD
simulations23 has established a few important features of
the liquid structure of [C2mim][NTf2]. In particular, it has
been shown that the imidazolium ring carbon atoms are
all mainly correlated with the oxygen atoms of the sulfonyl
groups in the anion. The strongest interaction is attributed
to the CR atom (see Figure 2). Whether this interaction
has to be attributed to hydrogen bonding or not depends
on the definition of hydrogen bonding, which is rather
fuzzy.50,51
At around 1 Å-1, there is a third low-s peak that should point
out the presence of long-range structuring. This structure is more
prominent for the shortest chain, [C1mim][NTf2]. Unfortunately,
in the simulation of [C1mim][NTf2], this feature, though present,
is not as prominent as in the experimental profile and this
prevented a clear identification of a geometric feature explaining
such a result.
The experimental (left) and theoretical (right) structure factors,
as measured and calculated by eq 5, are illustrated in Figure 3.
The theoretical data reported here were obtained using the

16402 J. Phys. Chem. B, Vol. 114, No. 49, 2010
Bodo et al.
Figure 4. sM(s)I(s) for the four ionic liquids, [Cnmim][NTf2] (where n ) 1, 4, 6, or 8). Left: measured quantities. Right: simulated structure
factors. The numbers in the captions are the densities in g cm-3. The simulations reported here are obtained with the Canongia Lopes force field.
imidazolium based ionic liquids have been carried out in the
aforementioned papers by Urahata and Ribeiro,27 Wand and
Voth,28 and by Canongia Lopes and Pa´dua.32 Experimental work
on these low-s features and a more detailed discussion on their
nature can be found in the work by Triolo et al.29
Some of the above calculations, namely for the [C4mim]-
[NTf2] and [C6mim][NTf2] systems, have been repeated with
the more recent force fields by Ko¨ddermann et al. and produced
better densities.33 The sM(s)I(s) calculated with the latter force
field are reported in Figure 6. As can be seen, above 3 Å-1
(below 2 Å) the Ko¨ddermann and Canongia Lopes force fields
provide substantially the same results and this is a rather obvious
result since the intramolecular part of the force field is the same.
The two force fields show small differences at long range, i.e.,
below 3 Å-1, but it is hard to say which one scores better in
predicting the peak positions, while for both the agreement on
Figure 5. I(s) calculated by activating the contribution of the carbon/
the intensities remains uncertain.
hydrogen groups indicated in the legend. (See Figure 2 for notation.)
From the I(s), the total radial distribution function or, to better
visualize the various contributions, the Diff(r) of eq 4 can be
easily obtained. Figure 7 represents the Diff(r) for the four ionic
To investigate the presence of this peak, we have calculated
the static structure factors setting to zero the scattering factors
liquids compared with those coming from the Fourier trans-
of some carbon/hydrogen atoms in the imidazolium ring, while
form of the experimental structure factors. For all the ionic
leaving the others unchanged. In particular, Figure 5 illustrates
liquids the results were obtained from the calculations performed
the structure factors obtained by activating the contribution of
with the Canongia Lopes force field, while for [C4mim][NTf2]
one of the C1/H1, CW/HCW, and CR/HCR groups and setting
and [C6mim][NTf2] the results are also presented for the
to zero the other two. As evident from Figure 5, the exclusion
Ko¨ddermann one. Again, the intramolecular radial distribution
of C1/H1 and CW/HCW groups in the calculation (i.e., only
function, i.e., the spectral range below 3 Å has been accurately
reproduced. The Canongia Lopes force field, however, seems
to perform better at long-range, especially in reproducing the
amplitudes of the oscillations between 5 and 15 Å. In general,
both force fields are quantitatively in good agreement with the
experimental data and describe correctly the shape of the long-
range RDF in the longer side chain compounds.
The experimental result (Figure 7 black circles) clearly
indicates a substantial difference between [C1mim][NTf2] and
[Cnmim][NTf2] (n > 1). Pronounced oscillations are, in fact,
clearly visible on the long-range scale while they are suppressed
for [C6mim][NTf2] and [C8mim][NTf2]. Unfortunately the results
reported in Figure 5 of ref 23 are limited to distances below 20
CR/HCR is active among the carbon ring atoms) makes the peak
considerably more pronounced than in the case where methyl
and CW-HCW groups are activated. The role played by CR/
HCR in the intermolecular interactions complies with the
structural pattern that has been found in crystalline [C1mim]-
[PF6],36 where each HCR interacts with two PF6 anions, and
suggests that a certain degree of local order might be preserved
in liquid [C1mim][NTf2].
Starting with [C8mim][NTf2], a prepeak located at 0.3 Å-1
appears, which can be observed also for [C6mim][NTf2], albeit
with more difficulty; but it is absent for shorter chain lengths.
The distances involved in that signal are around 25 Å.
Theoretical studies of these long-range ordering effects in

1-Alkyl-3-methylimidazolium [NTf2]-
J. Phys. Chem. B, Vol. 114, No. 49, 2010 16403
Figure 6. sI(s)M(s) for the [C4mim] (above) and [C6mim] (below) ionic liquids with two force fields compared to the experimental determination.
Figure 7. Diff(r) for [Cnmim][NTf2] with n ) 1 (upper-left corner), n ) 4 (upper-right), n ) 6 (lower-left), n ) 8 (lower right). Both experimental
and theoretical data are reported (two force fields where available).
Å and it is difficult to say whether [C2mim][NTf2] shows long-
range structuring effects as for the [C1mim][NTf2].
The relative conformation of the anion has already been
shown to be predominantly of trans type.31 It is very simple to
provide a further confirmation of this aspect of the local
geometric structure by looking at the CBT-CBT RDFs (see
Figure 2 and Figure 9). It can be seen that the second peak
(corresponding to the larger C-C distance) is always dominant
in our simulations. Unfortunately, in the experimental data, this
information is hidden in the Diff(r) because it is superimposed
onto the peaks coming from the intermolecular interactions.
The local structure of the anion can be also characterized by
the analysis of its torsional angles.12,35 The probability distribu-
tion of the dihedral angles for two representative choices of the
atoms in the chain: the S-N-S-C angle and the O-S· · ·S-O
angle are shown in Figure 10. The former distribution is rather
broad and slightly irregular. It is in agreement with that shown
in ref 35 and clearly denotes the existence of two conformers,
It is possible to assign most of the short-range peaks using
our simulation: the first peak is due to a coalescence of C-F,
S-O, and S-N pair distributions in the [NTf2]- anion and to
C-C distributions in the imidazolium cation. The small second
peak at 2.1-2.2 Å is mainly due to F · · ·F pairs in the anion
and to C-C interactions in the imidazolium ring. The third
feature between 2.5 and 3.0 Å is due to various pair contribu-
tions in the anion superimposed on those coming from alkyl
chains of the cation. The peak at 4 Å is due to F· · ·N and F· · ·O
contributions. The anion-cation intermolecular contributions
form an almost continuous background starting at 3 Å and this
is what we see in the Diff(r) between 4.5 and 7 Å. In Figure 8,
these findings are summerized for the special case of
[C6mim][NTf2].

16404 J. Phys. Chem. B, Vol. 114, No. 49, 2010
Bodo et al.
Figure 8. Lower panel: experimental (circles) and theoretical (line) Diff(r) for [C6mim][NTf2]. Upper panel: selected RDFs for various intra- and
intermolecular pairs.
are described. In the first row, the probability distribution for
the angle formed by the first two carbon atoms with respect to
the CR-NA direction is reported. The distribution is rather
broad although it shows a certain preference for the torsional
energy minima,12 thus confirming that for side chains longer
than ethyl, there is a considerable freedom of rotation around
the NA-C1 bond. The second row reports the dihedral for the
NA-C1-C2-CS sequence. The anti and gauche conformers that
are known to have a comparable energy clearly predominate.12 It
is interesting to note that the rotational mobility around the C1-C2
bond is drastically reduced with respect to that around the NA-C1
bond. As the chain gets longer, it behaves as in liquid hydrocarbons,
and all its torsional angles prefer the anti conformation with little
preference for the gauche (see the third row of Figure 11). We
conclude that in these liquids, the alkyl chain is relatively free to
rotate around its very first connections to the ring while its relative
conformation becomes more rigid the further from the ring the
carbon atoms reside.
The hydrogen atoms on the imidazolium ring are slightly
acidic51,52 and can partake in a special form of hydrogen bond
with the negatively charged atoms of the anion. Figure 12 shows
an analysis of the RDF of the two kind of hydrogen atoms on
the imidazolium ring and the various negatively charged terminal
Figure 9. C-C RDF in the [NTf2]- anion for the four ionic liquids
atoms of the anion. Although both the nitrogen and the fluorine
analyzed here.
atoms do present some correlation with the protons, it is evident
that the hydrogen bond is formed with the oxygen of the sulfonyl
groups. Despite having a larger partial charge than oxygen, the
one with dihedral angles around 90 and 270° and another with
nitrogen seems unable to form a stable “hydrogen bond” with
the imidazolium ring, probably because of the repulsive steric
effect due to the presence of the two sulfonyl groups. The
“hydrogen bond” with the HCR atom (which is also the most
acidic) is the strongest and leads to a distinctive feature in the
RDF at 2.5 Å. The situation for the other three ionic liquids is
quite similar and is illustrated by Figure 13. The strength of
the hydrogen bonding is larger for the HCR atom and increases
slightly with the size of the alkyl chain. In the same Figure, the
RDF for the CR-OBT distance (compared with those reported
in Figure 6 of ref 23) are shown, and hence we can easily
the angles around 120 and 240°. The first conformer is the most
stable12 and also the most populated. The O-S· · ·S-O angles
show instead that the highly polarized oxygen atoms in the anion
clearly prefer an anti conformation. It is notable that the structure
of the anion depends very little on the size of the alkyl chain
on imidazolium.
The conformation of the alkyl side chain on imidazolium can
be characterized by an analogous analysis of the dihedral angles
between carbon atoms. The probability distribution for such
angles is shown in Figure 11. In each of three columns, the
results for [C4mim][NTf2], [C6mim][NTf2], and [C8mim][NTf2]

1-Alkyl-3-methylimidazolium [NTf2]-
J. Phys. Chem. B, Vol. 114, No. 49, 2010 16405
Figure 10. Dihedral angles probability distribution for the [NTf2]- anion for the four ionic liquids.
Figure 11. Dihedral angles probability distribution for the imidazolium alkyl chain in [Cnmim][NTf2] with n ) 4, 6, or 8. Labeling is defined in Figure 2.
evaluate the average CR-HCR-OBT angle that turns out to
be ca. 130°. This means that this kind of hydrogen bonding in
these liquids is slightly directional though the H atom lies off
the internuclear C-O distance by about 20°.
Figure 14 represents the correlation diagram for the two RDFs
for CR-OBT and HCR-OBT for the [C1mim][NTf2] ionic liquid.
Each point of the diagram represents a possible value of the
CR-HCR-OBT angle (except those points that do not fulfill
the triangular condition such as those in the lower right half of
the square). Three contours are plotted, each enclosing the zone
where the angle has the values: 30, 90, and 120°. On top of
these regions we have plotted the contours of the angular
distribution function obtained as a product of the two RDFs.
This distribution shows a sharp peak at the point with
coordinates (3.5, 2.5), which means an angle around 120°. This
sharp distribution is compatible with a rather directional
hydrogen bonding feature between the OBT atom and the CR
atom of the imidazolium ring.

16406 J. Phys. Chem. B, Vol. 114, No. 49, 2010
Bodo et al.
Figure 12. HCW/HCR RDF with various negatively charged atoms in the [C1mim][NTf2] ionic liquid. Labeling is defined in Figure 2.
Figure 13. HCW/HCR-OBT RDF for the [Cnmim][NTf2] ionic liquids with n ) 1, 4, 6, or 8.
Conclusions
the imidazolium ring have also been analyzed. In the liquid
phase, the anion is found in the preferred trans conformation,
and the alkyl side chains are characterized by a sequence of
dihedral angles in the anti conformation. The conformation of
the side chain is less rigid close to the imidazolium ring, where
there is a certain rotational mobility. In [C1mim][NTf2], a new
feature shows up that is due to an intermolecular long-range
ordering effect, which is probably allowed by the absence of a
lateral chain and involves predominantly the carbon atoms on
the ring. A global RDF has been extracted from experimental
and simulated data, allowing the interpretation of the main
features of the experimental peaks in terms of atomic pair
contributions. Even in this case, the two force fields employed
here have not shown substantial differences. We have also
analyzed the behavior of the hydrogen bonds that are formed
between the acidic imidazolium protons and the anion: the
electron donors are almost exclusively the oxygen atoms of the
Measurements of the X-ray diffraction data for the pure ionic
liquids [Cnmim][NTf2] with n ) 1, 4, 6, or 8, have been reported
for the first time, and interpreted by extracting the corresponding
simulated structure factor from all-atom MD calculations. In
particular, the current force fields provide a local geometric
structure of the pure liquid, which is in very good agreement
with the experimental data. Both force fields are able to provide
an accurate description of the intramolecular structure and a
fairly good description of the intermolecular local structure. In
terms of static structure factors, the two force fields do not show
substantial differences, and it would be difficult to judge which
one scored better in this particular aspect of the simulation.
The simulation confirms once again some of the structural
features of these ionic liquids, such as the chain-chain
segregation effect on a scale of about 5-6 Å. The geometric
conformations of the anion and of the alkyl chains attached to

1-Alkyl-3-methylimidazolium [NTf2]-
J. Phys. Chem. B, Vol. 114, No. 49, 2010 16407
(15) Bonhoˆte, P.; Dias, A. P.; Papageorgiou, N.; Kalyanasundaram, K.;
Gra¨tzel, M. Inorg. Chem. 1996, 35, 1168–1178.
(16) Marsh, K. N.; Brennecke, J. F.; Chirico, R. D.; Frenkel, M.; Heintz,
A.; Magee, J. W.; Peters, C. J.; Rebelo, L. P. N.; Seddon, K. R. Pure Appl.
Chem. 2009, 81, 781.
(17) Chirico, R. D.; Diky, V.; Magee, J. W.; Frenkel, M.; Marsh, K. N.
Pure Appl. Chem. 2009, 81, 791.
(18) Lopes, J. N. C.; Deschamps, J.; Pa´dua, A. A. H. J. Phys. Chem. B
2004, 108, 2038.
(19) Dzyuba, S. V.; Bartsch, R. A. ChemPhysChem 2002, 3, 161–166.
(20) Fredlake, C. P.; Crosthwaite, J. M.; Hert, D. G.; Aki, S. N. V. K.;
Brennecke, J. F. J. Chem. Eng. Data 2004, 49, 954–964.
(21) Tokuda, H.; Hayamizu, K.; Ishii, K.; Susan, M. A. B. H.; Watanabe,
M. J. Phys. Chem. B 2005, 109, 6103–6110.
(22) Jin, H.; O’Hare, B.; Dong, J.; Arzhantsev, S.; Baker, G. A.; Wishart,
J. F.; Benesi, A. J.; Maroncelli, M. J. Phys. Chem. B 2008, 112, 81–92.
(23) Fujii, K.; Soejima, Y.; Kyoshoin, Y.; Fukuda, S.; Kanzaki, R.;
Umebayashi, Y.; Yam- aguchi, T.; Ishiguro, S.-i.; Takamuku, T. J. Phys.
Chem. B 2008, 112, 4329–4336.
(24) Russina, O.; Triolo, A.; Gontrani, L.; Caminiti, R.; Xiao, D.; Hines,
L. G., Jr.; Bartsch, R. A.; Quitevis, E. L.; Pleckhova, N.; Seddon, K. R. J.
Phys. Condens. Matter 2009, 21, 424121.
(25) Xiao, D.; Hines, L. G., Jr.; Li, S.; Bartsch, R. A.; Quitevis, E. L.;
Russina, O.; Triolo, A. J. Phys. Chem. B 2009, 113, 6426–6433.
(26) Tokuda, H.; Hayamizu, K.; Ishii, K.; Susan, M. A. B. H.; Watanabe,
M. J. Phys. Chem. B 2004, 108, 16593–16600.
(27) Urahata, S. M.; Ribeiro, M. C. C. J. Chem. Phys. 2004, 120, 1855–
1863.
Figure 14. Radial correlation plot for the OBT-HCR and CR-OBT
distances in the [C1mim][NTf2] ionic liquid. See text for details.
(28) Wang, Y.; Voth, G. A. J. Am. Chem. Soc. 2005, 127, 12192–12193.
(29) Triolo, A.; Russina, O.; Bleif, H.-J.; Cola, E. D. J. Phys. Chem. B
2007, 111, 4641.
(30) Triolo, A.; Russina, O.; Fazio, B.; Appetecchi, G. B.; Carewska,
M.; Passerini, S. J. Chem. Phys. 2009, 130, 164521.
(31) Deetlefs, M.; Hardacre, C.; Nieuwenhuyzen, M.; Pa´dua, A. A. H.;
Sheppard, O.; Soper, A. K. J. Phys. Chem. B 2006, 110, 12055–12061.
(32) Canongia Lopes, J. N. A.; Pa´dua, A. A. H. J. Phys. Chem. B 2006,
110, 3330–3335.
sulfonyl groups, despite not being the most negatively charged
atoms. The hydrogen bonding is found here to be rather
directional and stronger with the HCR hydrogen, with this
strength increasing, albeit only slightly, with the size of the side
chain.
(33) Ko¨ddermann, T.; Paschek, D.; Ludwig, R. ChemPhysChem 2007,
Acknowledgment. Financial support of the Scientific Com-
mittee of the University of Rome is gratefully acknowledged.
Computational support of the CASPUR and CINECA Super-
computing Centers is also acknowledged. A.T. acknowledges
support from the FIRB “Futuro in Ricerca” research project
(RBFR086BOQ_001, “Structure and dynamics of ionic liq-
uids”). K.R.S. and N.V.P. thank Dr. David Wassell for his input
to the manuscript.
8, 2464–2470.
(34) Ko¨ddermann, T.; Paschek, D.; Ludwig, R. ChemPhysChem 2008,
9, 549–555.
(35) Logothetit, G.-E.; Ramos, J.; Economou, I. G. J. Phys. Chem. B
2009, 113, 7211–7224.
(36) Holbrey, J. D.; Reichert, W. M.; Nieuwenhuyzen, M.; Sheppard,
O.; Hardacre, C.; Rogers, R. D. Chem. Commun. 2003, 476.
(37) Fujii, K.; Fujimori, T.; Takamuku, T.; Kanzaki, R.; Umebayashi,
Y.; Ishiguro, S. I. J. Phys. Chem. B 2006, 110, 8179–8183.
(38) Lassegues, J. C.; Grondin, J.; Holomb, R.; Johansson, P. J. Raman
Spectrosc. 2007, 38, 551.
References and Notes
(39) http://www.iolitec.de.
(40) Caminiti, R.; Sadun, C.; Rossi, V.; Cilloco, F.; Felici, R. Italian
Pat. Appl. No. 01261484, 1993.
(41) Gontrani, L.; Russina, O.; Marincola, F. C.; Caminiti, R. J. Chem.
Phys. 2009, 131, 244503.
(42) Murata, Y.; Nishikawa, K. Bull. Chem. Soc. Jpn. 1978, 51, 411–
418.
(43) Cromer, D.; Waber, J. Atomic Scattering Factors for X-Rays. In
International Tables for X- Ray Crystallography; Kynoch Press: Birming-
ham, U.K., 1974; Vol. 4, pp 71-147.
(1) Ionic Liquid Synthesis, 2nd ed.; Welton, T., Wasserscheid, P., Eds.;
Wiley VCH: Weinheim, Germany, 2008; p 721.
(2) Stark, A.; Seddon, K. R. Ionic Liquids. In Kirk-Othmer Encyclo-
paedia of Chemical Technology, 5th ed.; Seidel, A., Ed.; John Wiley &
Sons, Inc.: Hoboken, NJ, U.S., 2007; Vol. 26, pp 836-920.
(3) Freemantle, M. An Introduction to Ionic Liquids; RSC Publishing:
London, 2009; p 281.
(4) Plechkova, N. V.; Seddon, K. R. Chem. Soc. ReV. 2008, 37, 123–
150.
(44) Dietrich, H. J. Appl. Crystallogr. 1976, 9, 238–239.
(45) Nishikawa, K.; Iijima, T. Bull. Chem. Soc. Jpn. 1984, 57, 1750–
1759.
(46) Fritsch, G.; Wagner, C. N. J. Z. Phys. B: Condens. Matter 1986,
62, 189–194.
(47) Smith; Forester, T. R. DL POLY, a package of molecular simulation
routines; The Council for the Central Laboratory of the Research Councils,
Daresbury Laboratory at Daresbury, North Warrington, U.K., 1996.
(48) Canongia Lopes, J. N.; Pa´dua, A. A. H. J. Phys. Chem. B 2004,
108, 16893–16898.
(49) Wieder, T.; Fu¨ss, H. Z.; Naturforsch, A. Phys. Sci. 1997, 52, 386–
392.
(50) Gilli, G.; Gilli, P. The Nature of the Hydrogen Bond: Outline of a
ComprehensiVe Hydrogen Bond Theory, 1st ed.; International Union of
Crystallography Monographs on Crystallography; Oxford University Press:
Oxford, U.K., 2009; p 336.
(51) Kempter, V.; Kirchner, B. J. Mol. Struct. 2010, 972, 22–34.
(52) Thar, J.; Brehm, M.; Seitsonen, A. P.; Kirchner, B. J. Phys. Chem.
B 2009, 113, 15129–15132.
(5) Ionic Liquids IIIA: Fundamentals, Progress, Challenges, and
Opportunities: Properties and Structure; Rogers, R. D., Seddon, K. R. Eds.;
American Chemical Society: Washington, DC, 2005; Vol. 901, p 356.
(6) Ionic Liquids IIIB: Fundamentals, Progress, Challenges, and
Opportunities: Transformations and Processes; Rogers, R., Seddon, K., Eds.;
American Chemical Society: Washington, DC, 2005; Vol. 902.
(7) Ionic Liquids: From Knowledge to Application; Rogers, R. D.,
Plechkova, N. V., Seddon, K. R., Eds.; American Chemical Society:
Washington, DC, 2009; Vol. 1030, p 464.
(8) Molten Salts and Ionic Liquids: NeVer the Twain?; Gaune-Escard,
M., Seddon, K. R., Eds.; Wiley: New York, NY, U.S., 2010; p 441.
(9) Maginn, E. J. J. Phys. Condens. Matter. 2009, 21, 373101.
(10) Lynden-Bell, R. M.; Atamass, N. A.; Vasilyuk, A.; Hanke, C. G.
Mol. Phys. 2002, 100, 3225–3229.
(11) Canongia Lopes, J. N.; Deschamps, J.; Pa´dua, A. A. H. ACS Symp.
Ser. 2005, 901, 134–149.
(12) Canongia Lopes, J. N. A.; Pa´dua, A. A. H. J. Phys. Chem. B 2006,
110, 7485–7489.
(13) Borodin, O. J. Phys. Chem. B 2009, 113, 11463–11478.
(14) Bodo, E.; Gontrani, L.; Triolo, A.; Caminiti, R. J. Phys. Chem.
Lett. 2010, 1, 1095–1100.
JP1093299