Structure of ionic liquids with amino acid anions via neutron diffraction

Abstract

The liquid structures of the ionic liquids 1-ethyl-3-methylimidazolium alaninate and 1-ethyl-3-methylimidazolium serinate are fully elucidated through the application of neutron diffraction techniques. We observe significant direct interaction between anions, particularly in the case of the serinate ionic liquid which is strongly hydrogen bonding between its hydroxyl and carboxylate groups, and is attributed the significant increase in viscosity of the neat liquid to this structural feature. Minor differences in the elucidated interactions are present between the R and S forms of the anions.

---

Introduction

Over the last decade interest in the use of room temperature ionic liquids for metal extractions, organic synthesis and electrochemical applications has grown significantly. Interest in this class of solvent stems from the properties exhibited by the liquids (including effectively zero vapour pressure) and the ease by which many of these properties may be varied. Many of these materials are based around the imidazolium cation, 1-alkyl-3-methylimidazolium, and by simply changing the anion, for example [BF₄]⁻, [PF₆]⁻, [CF₃CO₂]⁻, [(CF₃SO₂)₂N]⁻, or the alkyl chain on the cation, a wide range of properties can be tuned, for instance the hydrophobicity, viscosity and density.

Amino acid ionic liquids have been shown to be an interesting subsection of ionic liquids.^1–3 The amino acids are one of the most abundant biomaterials in nature and are readily available at reasonable cost, enabling ionic liquids based upon them to be prepared in large quantities, and in high purity. In addition, since an amino acid contains both an amino group and a carboxylic acid residue they may be used as either the anion or cation in a given ionic liquid. The physical and chemical properties of some these derivatives have been extensively investigated by Ohno, who detailed how the properties of the ionic liquids can vary depending on the amino acid side chain.⁴ Amino acid ionic liquids containing side chains with no functionality (i.e. they are simple alkyl tails) display a clear relationship between their ionic conductivity and glass transition temperature. Once the side chain is functionalised to some degree the nature of this relationship changes, implying the manifestation of intra- and/or inter-ion interactions.

The potential for this specific family of these ionic liquids is considerable – to date these ionic liquids have been investigated as solvents for many processes ranging from lignocellulose pre-treatment and, catalysts for biodiesel production, to CO₂ capture, and chiral solvents and separations. In addition, with careful choice of the counterion these ionic liquids are biodegradable, and hence can be considered as greener alternatives to many other families of ionic liquids. Whilst the physical and chemical properties of these types of ionic liquids have been examined, if a degree of predictability for processes within the family of ionic liquids is to be achieved it is important, in the first instance, to fully understand the interactions occurring between the solvent ions.

Neutron diffraction with isotopic substitution has been applied extensively to the determination of the local structure of neat ionic liquids,^5–12 and has proven to be an invaluable tool in establishing the nature of the core interactions that exist within these interesting materials. In recent years the technique has further been applied to the study of solutes and mixtures, including those with glucose,¹³ methylnaphthalene,¹⁴ phenol,¹⁵ ethanol,¹⁶ nitrate salts,¹⁷ glycerol,¹⁸ methanol,¹⁹ and amines.²⁰ Herein we use the technique to determine the liquid structure of ionic liquids based on the 1-ethyl-3-methylimidazolium cation coupled with amino acid anions. Two liquids, 1-ethyl-3-methylimidazolium alaninate ([C₂mim][Ala]) and 1-ethyl-3-methylimidazolium serinate ([C₂mim][Ser]) are fully resolved while a third, 1-ethyl-3-methylimidazolium glycinate ([C₂mim][Gly]) is partially resolved and included for comparison.

Experimental

Synthesis of amino acid ionic liquids

Starting materials were purchased from Sigma Aldrich and used as received, except for partially-deuterated amino acids (no substitution at heteroatoms) which were purchased from Qmx. Target amino acid ionic liquids were synthesised following the procedure of Ohno.² Characterisation of the protiated derivatives was consistent with the reported literature.

Heteroatom-deuterated amino acid analogues were prepared in a similar manner using either D₂O or H₂O depending on the required substitution at the exchangeable centres. All ionic liquids were dried in vacuo overnight before use. The chemical structures of the ionic liquids are shown in Fig. 1.

Fig. 1 Structures of the studied ionic liquids.

Neutron diffraction

Neutron diffraction measurements were made on the SANDALS instrument at ISIS, STFC Rutherford Appleton Laboratory, Harwell Oxford. SANDALS is an amorphous materials diffractometer with Q range 0.1 < Q < 50 Å⁻¹, offering access to atomic correlations between 0.125–30 Å. Ionic liquid samples were loaded into null-scattering Ti_0.676Zr_0.324 flat-plate cans with 1 mm slot width and sealed with a PTFE o-ring. All measurements were made at 25 °C. For the ionic liquids [C₂mim][Ala] and [C₂mim][Ser] nine isotopically-substituted samples were considered, Table 1. For [C₂mim][Gly] only the fully protiated sample was considered owing to material constraints.

Table 1 Isotopic substitutions considered. For the cation, ring hydrogens are always protiated

Sample	Cation	Anion CH	Anion NH/OH
1	H	H	H
2	D	H	H
3	D	H	D
4	D	D	H
5	D	D	D
6	D	H:D	D
7	D	D	H:D
8	D	H:D	H:D
9	H:D	H:D	H:D

---

Collected data were processed with the Gudrun software in order to account for multiple scattering and attenuation effects, to remove residual inelasticity arising from the presence of hydrogen, and to normalise the structure factors to a vanadium standard, thus placing the data on an absolute scale.²¹ The observed mean scattering levels of each sample were consistent with their expected composition. The basic quantity thus obtained from such a scattering experiment, after these corrections is the total incoherent structure factor F(Q):

where c_i, c_j, b_i, and b_j are, respectively, the atomic fractions and bound coherent scattering lengths of atom types i and j in the system, and S_ij(Q) is the partial structure factor arising from correlations between the two components, which is the Fourier transform of the partial radial distribution function, g_ij(r), between the same:

Analysis of the data was made using the Empirical Potential Structure Refinement (EPSR) technique.²² Briefly, the EPSR procedure is as follows: a classical molecular Monte Carlo algorithm is applied to a configuration of atoms representative of the system under study. After equilibration of the energy of this starting configuration has been performed, the refinement procedure begins. Given a set of isotopically-distinct experimental F(Q) (in this case the set of all datasets measured for a particular ionic liquid), a corresponding set of simulated F(Q) is calculated from the current atomic configuration. Differences between the experimental and simulated F(Q) are calculated and used to form an empirical interatomic potential that is applied to the simulation box in conjunction with the reference potential. This has the effect of driving the Monte Carlo simulation towards reproduction of the experimental F(Q). Once satisfactory agreement between the experimental and calculated F(Q) have been achieved, properties of interest may be calculated at will based on atomic configurations generated by EPSR using the adjusted interatomic potential, including any partial g(r) of interest. Simulation boxes consisted of 300 ion pairs in each case, corresponding to cubic box lengths of 43.130, 44.491, and 44.837 Å for [C₂mim][Gly], [C₂mim][Ala], and [C₂mim][Ser] respectively, and reflecting the experimentally-determined densities.¹ In line with the purchased amino acid starting materials, a racemic mixture of the R- and S-forms of the anions was assumed in the simulation (i.e. 150 ions of each were present in the box). Reference potential parameters for the cation were taken from ref. 13, while those for the anions were taken from the OPLS-AA force field,²³ with atomic charges based on restricted electrostatic potential fits acquired from quantum mechanical calculations at the HF/6-31+G2dp level, performed using the GAMESS-US package.²⁴ All quantities presented in the following sections were averaged over at least 10⁴ configurations for each ionic liquid under consideration.

Results and discussion

Experimental data and fits

Collected experimental F(Q) and the corresponding data simulated and refined with EPSR are shown in Fig. 2, along with the difference functions illustrating discrepancies between the two. The refined simulation data show excellent agreement with the experimental curves, and indicate that the atomic configurations so generated can confidently be mined for structural information on these systems. Generally, the observed differences in the datasets between isotopic substitutions of a given ionic liquid are small, but nevertheless some initial information may be extracted from these raw data. For instance, it can be seen that the primary feature at Q ≈ 1.5 Å⁻¹ corresponding to correlations between molecules increases slightly in intensity on moving from [C₂mim][Gly] to [C₂mim][Ala], and then becomes significantly sharper in the case of [C₂mim][Ser]. On a basic level this suggests that the serinate system exhibits considerably stronger ordering between ions than either the glycinate or the alaninate.

Fig. 2 Experimental (dashed lines), simulated (solid lines), and fit residual (dotted lines) F(Q) for all three ionic liquids studied. Note that in the case of [C₂mim][Gly] only one experimental dataset, H(HH), is available. Thin lines represent differences between experiment and simulated data. Notation used to identify the isotopic substitution of the samples is X(YZ), where ‘X’ represents the cation ethyl and methyl chains, ‘Y’ represents the C–H of the anion, and ‘Z’ represents the N–H and O–H (if present) of the anion. In all cases, ‘H : D’ indicates a 50 : 50 mix of H and D.

General liquid structure

Centre-of-mass radial distribution functions (RDFs) show the prototypical features found for many previously-studied ionic liquids, Fig. 3. A strong correlation in the cation–anion RDF at around 5 to 5.5 Å reflecting the basic ionic interaction between the oppositely charged species is clear, but it is notable that further correlations reflecting second-neighbours and beyond are weak, if not completely absent. This contrasts strongly with 1,3-dimethylimidazolium chloride⁶ or 1-ethyl-3-methylimidazolium acetate⁷ where clear second-order peaks are found, and reflects reduced potential for formation of oscillating ‘shells’ of cations/anions by the present anions.

Fig. 3 Ionic radial distribution functions, calculated using the centroid of the imidazole nitrogens for the cation and the centres-of-mass of the anions for glycinate (black), alaninate (red), and serinate (green) systems.

The cation–cation structuring is weak, as might be expected, but the anion–anion RDFs display distinct peaks between 4 and 6 Å which indicate the presence of non-negligible anion–anion close contacts. This primary peak is followed by a secondary broad correlation around 9 Å, which is attributed to anions interacting with the same imidazolium ring of a given cation. Cation–anion and anion–anion coordination numbers calculated from integration of the primary peaks up to the first minimum in the RDFs are listed in Table 2. On average seven anions are found around a central cation, although clearly not all are strongly interacting at once, with some in close proximity simply in order to satisfy charge balance within the liquid. For the anion–anion coordination numbers, a general increase is seen on moving from [Gly]⁻ to the more ‘complex’ anions [Ala]⁻ and [Ser]⁻, with the highest number of neighbours seen for the latter. The origins of this increase in anion association will be discussed in the following sections.

Table 2 Ion–ion coordination numbers

	Cation–anion		Anion–anion
	r (Å)	CN	r (Å)	CN
[C2mim][Gly]	7.70	7.56	6.24	2.57
[C2mim][Ala]	7.81	7.27	6.77	3.28
[C2mim][Ser]	8.08	7.78	6.94	3.63

---

Cation–anion interaction

As is usually the case with ionic liquids based on imidazolium cations, the primary directional interaction between the cation and the amino acid anions in the present study is between the imidazolium ring hydrogens H2, H4, and H5 and the carboxylate group of the anion. Indeed, many parallels can be drawn with the related acetate system in this regard.⁹ Fig. 4 shows the partial RDFs between H2, H4, and H5 and the carboxylate oxygens of the anions. For each hydrogen a sharp peak is seen at 2.3 Å, indicating a C–H⋯O hydrogen bond of reasonable strength, with the strongest and most prevalent being for the slightly more delta-positive H2 position as is to be expected. Furthermore, the H4 proton exhibits stronger association with the carboxylate oxygens than does H5 as a result of the nearby ethyl chain in the case of the latter, restricting anion approach to the cation. In the serinate system, ring proton association is reduced when compared with the alaninate – peaks are still evident in the RDF but their intensity is significantly reduced. These observations are consistent with the increased capacity for interionic interactions of the serinate anion.

Fig. 4 Partial radial distribution functions between imidazolium ring hydrogens and anion carboxylate oxygens for glycinate (black), alaninate (red), and serinate (green) systems.

Table 3 details the contact numbers between imidazolium ring protons and the carboxylate oxygens of the anion for the two primary binding geometries – monodentate and bidentate modes. For all three ionic liquids the preference is for monodentate interaction over bidentate interaction, in approximately a 3 : 1 ratio (regardless of the proton considered). In total, the cation is involved in between 1.9 and 2.5 hydrogen bonds of this nature at any point in time, and we observe a general decrease in the number of contacts in the order [Gly]⁻ > [Ala]⁻ > [Ser]⁻. The more accessible and acidic proton of the cation, H2, is found to be involved in more of this type of contact than the weaker H4 proton, and the slightly hindered H5 proton (nearest the ethyl side chain).

Table 3 Contact numbers between imidazolium ring hydrogens and carboxylate groups of the anion, within a maximum H⋯O distance of 3.0 Å. Total numbers given effectively represent the number of hydrogen bonds with carboxylate oxygens that each ring hydrogen takes part in at any moment in time. Other possible contact types between the species are not shown, and account for less than 1% of H2⋯O contacts and around 10% of H4⋯O and H5⋯O contacts (where most of these can be associated with the carboxylate group bridging over H4 and H5 simultaneously)

	Total				Monodentate			Bidentate
	H2	H4	H5	All	H2	H4	H5	H2	H4	H5
Gly	0.97	0.79	0.64	2.39	0.73	0.56	0.46	0.24	0.16	0.11
Ala	0.90	0.73	0.60	2.23	0.65	0.49	0.41	0.25	0.16	0.11
Ser	0.78	0.63	0.55	1.96	0.56	0.41	0.36	0.22	0.13	0.10

---

For all three ionic liquids considered there are further sites beyond the carboxylate group which may also attract interactions from the cation, namely the NH₂ group and, for [Ser]⁻, its OH group. Fig. 5 shows partial RDFs between the H2, H4, and H5 protons of the cation and these groups, where it can be seen that a small interaction with the amine groups exists, and follows the expected trend based on acidity/accessibility of the ring protons (H2 > H4 > H5), with contact numbers reaching a maximum of 0.32 (for H2⋯N(H₂) in [C₂mim][Gly], integrating up to 3.0 Å). Interactions with the oxygen of the hydroxyl group in the serinate system prove to be slightly stronger, but otherwise show a similar trend with ring proton.

Fig. 5 Partial radial distribution functions between imidazolium ring hydrogens and amine nitrogens (left) and (where present) the hydroxyl group oxygen (right) for glycinate (black), alaninate (red), and serinate (green) systems.

Anion–anion interactions

As noted earlier, the coordination number of anions around anions increased in the order [Gly]⁻ < [Ala]⁻ < [Ser]⁻, and follows the trend of increasing viscosity.¹ Firstly, we find no significant clustering or association of aliphatic groups for any of the anions studied (see ESI†), contrary to [C₂mim][OAc] where a slight clustering of methyl groups on the anion was observed, reflecting somewhat the ‘polar head, apolar tail’ of the acetate. It might, however, be expected that the nearby NH₂ group would show some self-association, and perhaps drive the clustering of the aliphatic backbones. Fig. 6 reveals that there is in fact very little correlation between the amine hydrogens and the amine nitrogen on separate anions. Again it is clear that some short contacts exist (evident in all three ILs) but they cannot be regarded as a significant driver for anion association. More pronounced interactions are found in the form of hydrogen bonds between the amine hydrogens and the carboxylate groups on different anions, with the association strongest for the [Gly]⁻, and similar between [Ala]⁻ and [Ser]⁻.

Fig. 6 Partial radial distribution functions between amine hydrogens and carboxylate oxygens (top) and amine nitrogens (bottom) for glycinate (black), alaninate (red), and serinate (green) systems.

This difference is attributed to the smaller side chain on glycinate, permitting more facile approach to the carboxylate group on a second anion (and hence stronger interaction), while in the case of alaninate and serinate the CH₃ and CH₂OH groups hinder this somewhat, reducing its effectiveness.

In the case of [C₂mim][Ser] a second hydrogen bond donor/acceptor group exists in the form of the hydroxyl. Fig. 7 shows that there are no preferential interactions with the amine group as the donor and hydroxyl as the acceptor, and that a reasonably strong H-bond can occur between hydroxyl groups. However the strongest and most prevalent interaction is between the hydroxyl hydrogen and the carboxylate oxygen, displaying a sharp peak of large magnitude at 1.64 Å (compared to 1.74 Å for the other two H-bonding interactions considered here). The occurrence of each interaction in terms of contact numbers is summarised in Table 4. We see that, overall, the number of short contacts that may be considered hydrogen bonds is slightly greater between ions for [C₂mim][Gly] than for [C₂mim][Ala], which reflects the additional methyl group in the latter, frustrating packing between molecules and thus reducing the number of favourable contacts found. Cation–anion interactions account for approximately 75% of these, as one would expect, but even in the case of the [Gly]⁻ and [Ala]⁻ based ionic liquids there are distinct hydrogen bonding interactions between the anions. The addition of the OH group in [Ser]⁻, however, allows the formation of further favourable contacts between both cation–anion and anion–anion pairs. While the number of short contacts between the imidazolium ring protons and carboxylate oxygens is decreased in [C₂mim][Ser], this is more than compensated for by additional contacts formed with the oxygen of the hydroxyl group on the anion. Moreover, while the hydroxyl group itself does not form strong contacts with other hydroxyls or indeed the amine group nitrogen, there is significant interaction with carboxylate oxygens, providing a further 0.53 hydrogen bonds per carboxylate in the system. Overall, then, the serinate system displays the highest number of favourable hydrogen bonding interactions between both cation–anion and anion–anion pairs, accounting for the change in thermophysical properties of this particular system.

Fig. 7 Partial intermolecular radial distribution functions involving the hydroxyl group in [C₂mim][Ser].

Table 4 Site–site coordination numbers between potential hydrogen bond donor sites on the cation and anion, and specific hydrogen bond acceptor sites on the anion. Values are calculated using the following distance cutoffs: H2/4/5 to O(COO⁻), N(NH₂), and O(OH) – 3.0 Å; N(NH₂) to O(COO⁻), N(NH₂), and O(OH) – 2.6, 2.6, and 2.4 Å respectively; H(OH) to O(COO^–), N(NH₂), and O(OH) – 2.2 Å respectively

	Site	[Gly]^−			[Ala]^−		[Ser]^−
		O(COO^−)		N(NH2)	O(COO^−)	N(NH2)	O(COO^−)	N(NH2)	O(OH)
[C2mim]^+	H2	0.97		0.32	0.90	0.30	0.78	0.28	0.31
	H4	0.79		0.25	0.73	0.23	0.63	0.22	0.26
	H5	0.64		0.20	0.60	0.19	0.55	0.17	0.21
[Gly]^−	H(NH2)	0.52		0.66
[Ala]^−	H(NH2)				0.46	0.53
[Ser]^−	H(NH2)						0.51	0.51	0.13
	H(OH)						0.53	0.08	0.13
Total (cation–anion)		3.17			2.95		3.41
Total (anion–anion)		1.18			0.99		1.89

---

Chiral effects on liquid structure

In this section we focus in on any structural differences that may arise between the R and S forms of the [Ala]⁻ and [Ser]⁻ anions, which in all analysis up to this point have been treated as equivalent. Centre-of-mass RDFs for the alaninate and serinate systems, Fig. 8, shows that the general cation–anion packing does not reflect any bias towards a particular conformer – coordination numbers are 3.622 and 3.644 for R and S respectively in the case of alaninate, and 3.887 and 3.895 in the case of serinate. Between anions there are more pronounced differences, but effects are still rather small. For alaninate anions the data suggest that the association of R-anions with other R anions is slightly preferred to that of S–S and R–S (coordination numbers of 1.696, 1.667, and 1.596 respectively). In the case of serinate the coordination between R forms is somewhat decreased relative to S–S and R–S (1.774, 1.798, and 1.847 respectively) but the difference functions do suggest that the R forms can approach at slightly shorter distances compared to S–S and R–S, as evidenced by the positive peak at around 4.1 Å.

Fig. 8 Original ion–ion radial distribution functions (black lines) and differences observed when considering individual conformers: red = R (or R–R), blue = S (or S–S), and magenta = R–S (or S–R).

Turning to the partial RDFs between cation ring hydrogens and anion carboxylate oxygens (see ESI†) for alaninate the association of H2, H4, and H5 is almost identical for R and S. For serinate, however, there does appear to be a slight preference for the H2 proton to interact with R over S (coordination numbers of 0.40 and 0.38 respectively). For interatomic interactions between the anions, again several small differences can be observed, with the largest changes occurring around the OH group of the serinate system (see ESI†). These may be indicative of preferential association between anion conformers, but again the changes are subtle.

Conclusions

The results detailed above show that, as has been revealed in many other cases, the primary interaction between cation and anion is via hydrogen bonding interactions involving the H2, H4, and H5 imidazolium hydrogens (albeit the last two to a lesser extent owing to their reduced acidity relative to H2). The anions studied in the present case all contain a carboxylate functional group, leading to the overriding directional cation–anion interaction being hydrogen bonding to these oxygen atoms. However, we also observe a non-negligible and highly specific interaction between anions in the case of the serinate ionic liquid. The hydroxyl group is shown to hydrogen bond strongly to the carboxylate functionality, permitting almost double the number of favourable hydrogen-bonding contacts between anions, compared to alaninate or glycinate, and is entirely consistent with the increased viscosity of the serinate system.

The anion amine group is shown not to be involved in any significant interactions in the system, except for the corresponding carboxylate group, and is more pronounced in the glycinate system than for alaninate or serinate. Nevertheless, caution must be employed regarding any inferences on the fine structure of the glycinate system, owing to the reduced number of neutron datasets available.

Acknowledgements

Experiments at the ISIS Pulsed Neutron and Muon Source were supported by a beamtime allocation from the Science and Technology Facilities Council, experiment RB1410649.

Notes and references

1. N. Muhammad, Z. B. Man, M. A. Bustam, M. I. Abdul Mutalib, C. D. Wilfred and S. Rafiq, J. Chem. Eng. Data, 2011, 56, 3157.
2. K. Fukumoto, M. Yoshizawa and H. Ohno, J. Am. Chem. Soc., 2005, 127, 2398.
3. G.-H. Tao, L. He, N. Sun and Y. Kou, Chem. Commun., 2005, 3562.
4. H. Ohno and K. Fukumoto, Acc. Chem. Res., 2007, 40, 1122.
5. C. Hardacre, J. D. Holbrey, M. Niewenhuyzen and T. G. A. Youngs, Acc. Chem. Res., 2007, 40, 1146.
6. C. Hardacre, J. D. Holbrey, S. E. J. McMath, D. T. Bowron and A. K. Soper, J. Chem. Phys., 2003, 118, 273.
7. D. T. Bowron, C. D'Agostino, L. F. Gladden, C. Hardacre, J. D. Holbrey, M. C. Lagunas, J. McGregor, M. D. Mantle, C. L. Mullan and T. G. A. Youngs, J. Phys. Chem. B, 2010, 114, 7760.
8. R. Atkin and G. G. Warr, J. Phys. Chem. B, 2008, 112, 4164.
9. C. Hardacre, J. D. Holbrey, C. L. Mullan, M. Nieuwenhuyzen, T. G. A. Youngs and D. T. Bowron, J. Phys. Chem. B, 2008, 112, 8049.
10. R. Hayes, S. Imberti, G. G. Warr and R. Atkin, Angew. Chem., Int. Ed., 2013, 52, 4623.
11. R. Hayes, S. Imberti, G. G. Warr and R. Atkin, J. Phys. Chem. C, 2014, 118, 13998.
12. C. Hardacre, J. D. Holbrey, C. L. Mullan, T. G. A. Youngs and D. T. Bowron, J. Chem. Phys., 2010, 133, 074510.
13. T. G. A. Youngs, J. D. Holbrey, C. L. Mullan, S. E. Norman, C. M. Lagunas, C. D’Agostino, M. D. Mantle, L. F. Gladden, D. T. Bowron and C. Hardacre, Chem. Sci., 2011, 2, 1594.
14. C. Hardacre, J. D. Holbrey, C. L. Mullan, M. Nieuwenhuyzen, T. G. A. Youngs, D. T. Bowron and S. J. Teat, Phys. Chem. Chem. Phys., 2010, 12, 1842.
15. A. Turner and J. D. Holbrey, J. Solution Chem., 2015, 44, 621.
16. O. Russina, A. Mariani, R. Caminiti and A. Triolo, J. Solution Chem., 2015, 44, 669.
17. R. Hayes, S. A. Bernard, S. Imberti, G. G. Warr and R. Atkin, J. Phys. Chem. C, 2014, 118, 21215.
18. T. Murphy, R. Hayes, S. Imberti, G. G. Warr and R. Atkin, Phys. Chem. Chem. Phys., 2014, 16, 13182.
19. O. Russina, A. Sferrazza, R. Caminiti and A. Triolo, J. Phys. Chem. Lett., 2014, 5, 1738.
20. J. Jacquemin, M. Bendova, Z. Sedlakova, M. Blesic, J. D. Holbrey, C. L. Mullan, T. G. A. Youngs, L. Pison, Z. Wagner, K. Aim, M. F. Costa-Gomes and C. Hardacre, ChemPhysChem, 2012, 13, 1825.
21. A. K. Soper, S. W. Howells and A. C. Hannon, ATLAS-Analysis of Time-of-Flight Diffraction Data from Liquid and Amorphous Samples, Report RAL-89–046, Rutherford Appleton Laboratory, 1989.
22. A. K. Soper, Chem. Phys., 1996, 202, 295; A. K. Soper, Chem. Phys., 2000, 258, 121; A. K. Soper, Mol. Phys., 2001, 99, 1503.
23. W. L. Jorgensen and J. Tirado-Rives, J. Am. Chem. Soc., 1996, 110, 1657.
24. M. W. Schmidt, K. K. Baldridge, J. A. Boatz, S. T. Elbert, M. S. Gordon, J. H. Jensen, S. Koseki, N. Matsunaga, K. A. Nguyen, S. Su, T. L. Windus, M. Dupuis and J. A. Montgomery, J. Comput. Chem., 1993, 14, 1347; Theory and Applications of Computational Chemistry: the first forty years Advances in electronic structure theory, ed. M. S. Gordon, M. W. Schmidt, C. E. Dykstra, G. Frenking, K. S. Kim and G. E. Scuseria, GAMESS a decade later Elsevier, Amsterdam, 2005, p. 1167.

---

Footnotes

† Electronic supplementary information (ESI) available: Additional partial radial distribution functions for individual R/S conformers of the anions, and EPSR molecular definition files for the ionic liquid ions. See DOI: 10.1039/c5ra06785e

‡ Current address: ISIS Facility, STFC Rutherford Appleton Laboratory, Harwell Oxford, Didcot, OX11 0QX, UK

---