Chemical Science EDGE ARTICLE Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article Online View Journal | View IssueaGZG, Abteilung Kristallographie, Georg-A 37077, Germany. E-mail: ffabbia@gwdg.d 393935 bSchool of Chemistry and Chemical Engineer Belfast, Belfast BT9 5AG, UK. E-mail: c.h 974687; Tel: +44 (0)28 90 974592 † Electronic supplementary information processing and structure renement parameters describing cation dimers, ca and environment, short H/F contacts a three polymorphs. CCDC reference nu crystallographic data in CIF or o 10.1039/c2sc21959j Cite this: Chem. Sci., 2013, 4, 1270 Received 11th November 2012 Accepted 21st December 2012 DOI: 10.1039/c2sc21959j www.rsc.org/chemicalscience 1270 | Chem. Sci., 2013, 4, 1270–128Pinning down the solid-state polymorphism of the ionic liquid [bmim][PF6]† Sofiane Saouane,a Sarah E. Norman,b Christopher Hardacre*b and Francesca P. A. Fabbiani*a The solid-state polymorphism of the ionic liquid 1-butyl-3-methylimidazolium hexafluorophosphate, [bmim][PF6], has been investigated via low-temperature and high-pressure crystallisation experiments. The samples have been characterised by single-crystal X-ray diffraction, optical microscopy and Raman spectroscopy. The solid-state phase behaviour of the compound is confirmed and clarified with respect to previous phase diagrams. The structures of the previously reported g-form, which essentially exhibits a G0T cation conformation, as well as those of the elusive b- and a-forms, are reported. Crystals of the b- phase are twinned and the structure is heavily disordered; the cation conformation in this form is predominantly TT, though significant contributions from other less frequently encountered conformers are also observed at low temperature and high pressure. The cation conformation in the a-form is GT; the presence of the G0T conformer at 193 K in this phase can be eliminated on cooling to 100 K. Whilst X-ray structural data are overall in good agreement with previous interpretations based on Raman and NMR studies, they also reveal a more subtle interplay of intermolecular interactions, which give rise to a wider range of conformers than previously considered.Introduction Due to their unique physicochemical properties, room- temperature ionic liquids (ILs) have attracted considerable interest in several branches of chemistry and chemical engi- neering. A low melting point, large liquid range and high thermal stability are some of the properties that make them attractive green solvents for synthetic and catalytic processes.1–3 More recently, ILs have found application as crystallisation media. Although their nonvolatility makes them poor solvents for crystallisation by evaporation, they can be used in sol- vothermal, thermal shi, co-solvent, slow diffusion and electro- crystallisation of a variety of chemicals, including pharmaceu- ticals and biomolecules.4–6 The liquid structure of ILs has been the subject of numerous theoretical and experimental investigations, the latter focusingugust-Universita¨t Go¨ttingen, Go¨ttingen, e; Fax: +49 551 399521; Tel: +49 551 ing, The QUILL Centre, Queen's University ardacre@qub.ac.uk; Fax: +44 (0)28 90 (ESI) available: Crystallographic data details, anion disorder, geometric tion and anion coordination numbers nd simulated powder patterns for the mbers 909171–909175. For ESI and ther electronic format see DOI: 0on X-ray and neutron scattering experiments.7–9 The X-ray crystal structures of the most common ILs have also been reported in the literature: the Cambridge Structural Database10 (the CSD, V. 5.33 including updates to May 2012 was searched for structure with 3-D coordinates, an R-factor below 10% and no errors) contains 212 structures of 1-methylimidazolium- based salts, of which 90 have a conrmed melting point below 373 K, which is the commonly accepted maximum melting temperature for dening an IL. Crystal polymorphism in 1,3- dialkylimidazolium-based ILs has been reported on the basis of spectroscopic, diffraction and calorimetric data. To the best of our knowledge, the rst example also conrmed by single- crystal X-ray diffraction was that of [bmim]Cl.11 1-Butyl-3-methylimidazolium hexauorophosphate, [bmim] [PF6], (Fig. 1), is one of the most studied ionic liquids. A wide range of simulations have been undertaken on this compound and show signicant probability of nding anions around the C2 hydrogen position as well as above and below the plane of the imidazolium ring.12,13 As found with other ionic liquids, such as the analogous chloride based systems,14,15 steric hindrance from the long alkyl chain, in this case butyl, has anFig. 1 Chemical diagram of the [bmim]+ cation with atom numbering scheme. This journal is ª The Royal Society of Chemistry 2013 Edge Article Chemical Science Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article Onlineeffect on the structure with the probability of anion interaction with the other ring hydrogens being dominated by the one closest to the methyl group. The solid-state behaviour of [bmim] [PF6] at ambient-pressure conditions has been extensively studied by a variety of techniques, including Raman and NMR spectroscopy, calorimetry, single-crystal X-ray diffraction, as well as wide-angle X-ray scattering. [bmim][PFB6] has also been the subject of high-pressure Raman and calorimetry experi- ments. Despite the evidence for crystal polymorphism, only one crystal structure has been reported in the literature to date.16,17 A brief summary of the ndings to date is given in Table 1, emphasising the conditions under which different phases have been obtained and using the scheme introduced by Endo et al.18 as reference to harmonising the polymorph nomenclature. Endo et al.18,19 characterised three crystalline forms, a, b and g, by calorimetry, Raman and NMR spectroscopy, pointing out that the endothermic transition to the g-phase is difficult toTable 1 Summary of crystallisation conditions and observed polymorphs of [bmim Author Experimentsa Cry Endo et al.18 LT Raman, DSC a-form Cool to 192 and crystallise at Triolo et al.9 LT DSC, WAXS cr-II Heat supercool and crystallise at Note: In the rst pa sin Choudhry et al.17 LT DSC, SC-XRD Cool to 123 K, crys 233 K (evidence fr curve) Dibrov and Kochi.16 LT SC-XRD n.o. Russina et al.21 HP Raman to 0.9 GPa S1 Keep at 0.44 GP hours; crystallise on pressurisatio Su et al.23,24 HP Raman to 2 GPa, DTA to 1 GPa Only one phase rep Raman spectra by R Yoshimura et al.22 HP Raman to 2 GPa Identied as S1 by et al. This work HP Raman and XRD to 0.8 GPa; LT Raman and SC-XRD; optical microscopy a-form Flash-coo glass transition, h crystallise at 22 Compression of b above 0.4 GP a LT ¼ low temperature; HP ¼ high pressure; DSC ¼ differential scanning X-ray diffraction; DTA ¼ differential thermal analysis. b Bold-face phas d Current interpretation of DSC signal. This journal is ª The Royal Society of Chemistry 2013detect by DSC measurements. The authors assigned the cation structure in the three phases by comparing experimental Raman spectra with calculated ones obtained by full geometry optimisation, using DFT methods, of the three most stable rotational isomers of the [bmim]+ cation, GT, TT and G0T (see Results and discussion section for details). The isomers were identied on the basis of the conformations found in experi- mental crystal structures and in gas-phase quantum chemical calculations.20 According to their Raman spectra, Endo et al. assigned the conformers of the a-, b- and g-phases as GT, TT and G0T, respectively. The most recent and comprehensive high-pressure investi- gation into the polymorphism of [bmim][PF6] was carried out by means of Raman spectroscopy up to ca. 1 GPa by Russina et al.21 Raman work was also carried out by Yoshimura et al.22 and by Su et al. up to 2 GPa;23 Su et al. have also carried out differential thermal analysis experiments up to 1 GPa.24 Russina et al.][PF6] reported in the literature to date stallisation conditions and observed crystalline phasesb K, heat 226.5 K b-form Phase transition on heating a to 250.3 K g-form Phase transition on heating b to 276 K ed liquid 220 K n.o.c cr-I Phase transition on heating cr-II to 245–252 K, or cooling to 260 K and waiting a few hours. WAXS pattern matches structures of Dibrov et al. and Choudhry et al. rt of the paper the phase assignment on the basis of comparison with gle-crystal structures is reversed to the one given later tallise at om DSC Evidence of a/ b transition at 263 K from DSCd Cool to 123 K, crystallise at 233 K and grow at 243 K. b / g transition at 276 K from DSCd n.o. Shock-induced crystallisation aer supercooling to 243 K a for 12 further n S2 Depressurise S1 to 0.04 GPa n.o. orted in both papers; re-analysis of the ussina et al. points out evidence for S1/ S2 n.o. Russina n.o. n.o. l below eat and 8 K. -form a b-form Phase transition on heating a to 248 K. Crystallise at high pressure below 0.4 GPa g-form Phase transition on heating a or b between 248 and 258 K; isothermal crystallisation of the liquid at 277 K. Crystallise from an aq. mix at 0.8 GPa calorimetry; WAXS ¼ wide-angle X-ray scattering; SC-XRD: single-crystal es identied in original publications. c n.o. ¼ not observed/reported. Chem. Sci., 2013, 4, 1270–1280 | 1271 Fig. 2 Stages of isothermal in situ crystal growth of the b-phase of [bmim][PF6] in the DAC at 293 K. (a) and (b): melting at ca. 0.09 GPa; (c)–(g): slow growth on increasing pressure; (h): final crystal at 0.35 GPa (crystal 1); (i): final crystal at 0.07 GPa (crystal 2) in equilibrium with the liquid phase. Chemical Science Edge Article Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article Onlineidentied two high-pressure phases by Raman spectroscopy and constructed an elegant phase diagram; at ambient temperature, they determined a crystallisation pressure of ca. 0.048 GPa for the phase corresponding to the b-polymorph of Endo et al., in excellent agreement with a value of ca. 0.04 GPa reported by acoustic measurements.25 A second high-pressure phase, which the authors obtained by ambient-temperature crystallisation above ca. 0.44 GPa, was assigned the GT conformer and identied as the a-phase of Endo et al. The preference of this conformer at high pressure was conrmed in a very recent molecular dynamics simulation study, albeit in the liquid state.26 Intrigued by the observations that polymorphism of [bmim] [PF6] was reported at both low-temperature and high-pressure conditions yet only one crystal structure is available in the CSD, our approach has been to explore how the high-pressure and low-temperature crystallisation of [bmim][PF6] might lead to the formation of new polymorphs and to elucidate the crystal structures of these polymorphs by single-crystal X-ray diffrac- tion, complementing the results with Raman spectroscopy.Experimental methods Material The title compound was prepared in house. Bromobutane (1.1 eq.) was added to a stirred solution of methyl imidazole (1 eq.) in acetonitrile. The resulting mixture was stirred at 343 K overnight. Aer cooling, the solvent was concentrated and the crude product was recrystallised from ethyl acetate to yield a pale yellow solid. The bromide salt was then redissolved in acetonitrile and potassium hexuorophosphate (1.1 eq.) was added and the suspension was stirred vigorously overnight. The mixture was then ltered and the solvent concentrated in vacuo. The resulting oil was then redissolved in DCM and reltered to remove the KBr byproduct. Finally, the DCM solution was ushed through a pad of silica/alumina and charcoal and concentrated in vacuo to yield a colourless liquid. The 1H, 31P and 13C-NMR were consistent with previous reports. The ionic liquid was dried for 4 h at 333 K under high vacuum before use.High-pressure crystallisation: a- and b-phases A square-shaped beryllium-free diamond-anvil cell (DAC) of the Ahsbahs type27 (45 half-cell opening angle) was used for the high-pressure experiments; the cell was equipped with 600 mm culet diamonds of low uorescence grade and Inconel gaskets with a starting diameter hole of 300 mm. The pressure inside the sample chamber was measured according to the ruby uores- cence method28 using an in-house built kit that has a precision of 0.05 GPa. Crystallisation was induced by increasing the pressure progressively to ca. 0.7 GPa. A single crystal was obtained by slowmelting of the resulting polycrystalline powder when releasing the pressure to ca. 0.09 GPa (isothermal crystal growth) and subsequently increasing the pressure until the crystal was as big as the gasket hole (Fig. 2). The nal pressure inside the sample chamber was 0.35 GPa. A second single crystal was obtained by cooling the loaded DAC at 0.4 GPa for 121272 | Chem. Sci., 2013, 4, 1270–1280h at 277 K and subsequently warming it to room temperature and adjusting the pressure for optimal crystal growth. The nal crystal at ca. 0.07 GPa was in equilibrium with the liquid and X- ray diffraction data were collected at phase boundary conditions (Fig. 2). Both single crystals were attributed to the b-form. Despite the larger standard uncertainty, our value for the crys- tallisation pressure is in good agreement with that of 0.048(25) GPa by Russina et al.21 On compressing the single crystals to pressures above ca. 0.4 GPa, a phase transition accompanied by the disintegration into polycrystalline material was observed. Analysis of this phase at 0.8 GPa by Raman spectroscopy revealed the presence of the a-phase.Crystallisation in capillaries: a- and b-phases Initial crystallisation experiments were performed by loading the sample in 0.3 mm glass capillaries. The capillaries were sealed to keep the sample dry. The sample was rapidly ash cooled by immersing the capillary in liquid nitrogen to induce the crystal- lisation of a glass. On warming, a liquid-mediated transition to needle-like crystals was observed; on further warming these nee- dles transformed to plate-like crystals. A single crystal was then grown by temperature cycling; the capillary was subsequently transferred to a Bruker Apex diffractometer equipped with a low- temperature device set to a temperature of 193 K, which was chosen on the basis of the data collection temperature reported for the known crystal form. Warming similarly grown crystals to room temperature did not result in the formation of another phase. Since the sample was cooled by contact with a cotton bud soaked in liquid nitrogen, heating and cooling rates could not be monitored but were generally fast.Crystallisation using a Linkam stage: a- b- and g-phases To improve the temperature control, we switched to a Linkam THMS600 heating and freezing stage, which enables temperatureThis journal is ª The Royal Society of Chemistry 2013 Edge Article Chemical Science Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article Onlinecontrol within a precision of 0.1 K while still being able to view the sample through a quartz glass window (Fig. 3). Prior to sample cooling, the stage was heated to 373 K to remove any moisture trapped inside the sample chamber. Phase assignment was performed by a combination of single-crystal X-ray diffrac- tion and Raman spectroscopy. Over 50 crystallisation trials were performed. Depending on the starting crystalline phase, several transition paths are possible: herein, we detail ve crystallisation protocols. (1) The a-form was reproducibly obtained by cooling the sample below its glass transition temperature and then heating it up to a temperature in the range of 228–248 K, irre- spective of the heating rate. Once formed, fast heating rates suppressed any further transition. (2) Starting from the a-phase and using a slow heating rate (0.5 K min1) a transition to the b-phase occurred at ca. 248 K, in good agreement with the value of 250.3 K reported by Endo et al.18 (3) Starting from a fully crystalline sample of the a-phase (i.e. in the absence of any liquid), following an isothermal run at 258 K for 10–20 min, a transition to the g-phase was sometimes observed. (4) The g- phase was, at times, also obtained directly from the b-form using a similar isothermal run. (5) The g-phase, which has the lowest crystal growth rate, was reproducibly obtained by leaving the sample at 277 K for a few days. As reported by Endo et al.,19 the activation energy of the b / g phase change is considerable; hence, a long isothermal phase may be a key factor for either direct nucleation of the g-phase from the liquid state or for the b / g transition. Considerations on the reproducibility of the crystallisation protocols are further explored in the Results and discussion section. Single-crystal X-ray diffraction Full details on the programs used for data reduction and structure renement can be found in the ESI.† Structure renement strategies are discussed in detail in the ESI† and only the most salient features are reported below. Crystallo- graphic data are reported in Table 2. a-Phase Sample transfer to the diffractometer proved to be particularly difficult for this phase as any mechanical interference with theFig. 3 Optical images of crystals of [bmim][PF6] grown on a Linkam stage. (a) Single crystals of the a-phase growing from the liquid at 262 K; (b) single crystal of the a-phase at 260 K grown on a glass shard (coexistence of liquid and solid), and (c) crystal after cooling to 203 K; (d) b-phase melting at 279 K; (e) initial and (f) final stages of the a/ b phase transition at 248 K; (g) nucleation of the g-phase from the b-phase at 258 K and (h) g-phase coexisting with the melt at 282 K. This journal is ª The Royal Society of Chemistry 2013crystal at high enough temperatures where the crystal could be manipulated resulted in a phase transition. A suitable crystal for diffraction was hence grown on a secondary glass shard support placed on the Linkam stage and the shard was subse- quently mounted on a Bruker SMART 6000 Apex II CCD diffractometer (Fig. 3). Diffraction data were collected using Cu Ka radiation of l ¼ 1.54178 A˚ from a rotating anode at 100(2) and 193(2) K. The structure at 100 K was found to be ordered. At 193 K disorder could be modelled for the [PF6]  anion and for C8 and C9 of the butyl side chain using two-site split models. b-Phase Diffraction data were collected using a Bruker AXS SMART Apex II CCD diffractometer equipped with Mo-Ka sealed-tube radia- tion of l ¼ 0.71073 A˚ for an in situ grown crystal contained in a capillary at 193(2) K. A similar diffractometer equipped with a Ag microsource (Incoatec) of l ¼ 0.56085 A˚ was used for the 293(2) K high-pressure DAC experiment. Both low-temperature and high-pressure structures were found to be pseudo-mer- ohedral triclinic twins, with unit-cell constants emulating a monoclinic Cmetric cell and a [1 1 0] twin direction. Both [PF6]  anions were found to be equally disordered over two positions. Disorder was additionally found in the butyl side chain of the cations: for C10 at low temperature and for C9, C10 and C20 at high pressure. g-Phase Diffraction data were collected on a single crystal grown on the Linkam stage and transferred to a Bruker AXS SMART Apex II CCD diffractometer with Mo-Ka sealed-tube radiation of l ¼ 0.71073 A˚ at 263(2) K. A minor but signicant disordered component of the butyl side chain, atoms C8 and C9, could be successfully rened using a split-site model. Raman spectroscopy Raman spectra were recorded with a Horiba Jobin Yvon HR800 UV Micro-Raman spectrometer equipped with an air-cooled 20 mW 488 nm Ar-laser. Raman spectra were collected in the 200– 1200 cm1 range with a spectral resolution of ca. 2.2 cm1 using a grating of 600 grooves per mm and a Peltier-cooled CCD detector (Andor, 1024  256 pixels). The spectra were calibrated with the Raman scattering frequency of Si before and aer each measurement. Results and discussion Conformation of the [bmim]+ cation in the solid state In the present discussion, the rotamer denomination follows the scheme outlined by Tsuzuki et al.,20 which is based on the C2–N1–C7–C8 torsion angle (Fig. 1) being positive. The conformations of the N1–C7–C8–C9 and C7–C8–C9–C10 torsion angles, q1 and q2, respectively, are then assigned to the following angle ranges: T (trans, 150 to +150), G (gauche, +30 to +90) and G0 (gauche0, 30 to 90), see Fig. 4 for details. As noted by Endo et al.,18 the three polymorphs of [bmim][PF6] can be identied on the basis of characteristic Raman bands: for theChem. Sci., 2013, 4, 1270–1280 | 1273 Table 2 Crystallographic data for the three polymorphs of [bmim][PF6] discussed in this paper Structure a a b b g g Pressure 0.1 MPa 0.1 MPa 0.1 MPa 0.07 GPa 0.1 MPa 0.1 MPa T/K 100(2) 193(2) 193(2) 293(2) 180–193a 263(2) Ref. This work This work This work This work MAZXOB0117 This work Space group Pbca Pbca P1 P1 P1 P1 a/A˚ 9.3855(3) 9.4924(5) 9.3869(8) 9.5818(13) 8.774(5) 8.8215(8) b/A˚ 9.7769(3) 9.8406(5) 9.5879(8) 9.5826(9) 8.944(9) 9.0796(9) c/A˚ 26.7170(7) 26.8817(13) 14.4964(12) 14.5801(10) 9.032(6) 9.0381(8) a/ 90 90 98.492(5) 99.219(10) 95.95(2) 96.671(7) b/ 90 90 98.354(6) 99.252(6) 114.93(1) 114.768(6) g/ 90 90 101.089(6) 99.667(10) 103.01(3) 103.071(7) V/A˚3 2451.58(13) 2511.0(2) 1245.85(18) 1278.2(2) 610.2(8) 621.82(10) Z0 1 1 2 2 1 1 Dc/g cm 3 1.540 1.503 1.515 1.477 1.55 1.518 Measured reections 24 529 28 075 18 072 10 400 6680 7824 Unique reections 2380 2427 3544 1063 3484 2291 Observed reectionsb 2254 2054 2711 815 n.a. 1624 Parameters/restraints 156/0 229/255 349/618 341/809 154/0 174/57 Rint 0.05 0.06 0.05 0.04 n.a. 0.02 R 0.037 0.048 0.084 0.074 0.045 0.050 wR 0.100 0.155 0.275 0.2113 0.128 0.145 Drmin/max/e  A˚3 0.31/0.33 0.39/0.27 0.41/0.52 0.14/0.18 0.29/0.33 0.34/0.42 a Three different temperatures are given in the paper (193 K), ESI (180 K) and CIF le (183 K). b Criteria for observed reections: I > 2s(I). Chemical Science Edge Article Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article Onlinea-phase a band at 338 cm1, associated with ring wagging and chain deformation;29 for the b-phase a band at 624 cm1, which is associated with a number of group vibrational modes;30 for the g-phase a band at 326 cm1, presumably also associated with ring wagging and chain deformation. Endo et al.18 note that these bands are characteristic ngerprints for the GT, TT and G0T conformers, respectively. Our experimental Raman spectra, shown in Fig. 5, are in very good agreement with those reported at low temperature by Endo et al.18 and at high pres- sure by Russina et al.21 Whilst Raman spectroscopy enables a rapid and unambig- uous phase assignment, single-crystal X-ray diffraction allows the determination of a more detailed description of the cation conformation in the solid state. Gas-phase calculations on the cation, in isolation and in the presence of counterions have shown that the nine lowest-energy rotamers adopt a combina- tion of the T, G and G0 torsion angles, and that the energy of all rotamers are within 5 kJ mol1 of each other;20 however, our crystallographic study points out to the existence of other conformers, which in the previous calculations correspond toFig. 4 Dihedral angles for N1–C7–C8–C9 and C7–C8–C9–C10 for defining the [bmim]+ cation conformation. 1274 | Chem. Sci., 2013, 4, 1270–1280saddle points on the torsional potentials, so that additional torsion angles are introduced here: S (syn, 30 to +30), E (eclipsed, +90 to +150) and E0 (eclipsed0, 90 to 150). TheFig. 5 Raman spectra of [bmim][PF6] polymorphs collected at different condi- tions of temperature (top) and pressure (bottom). Solid lines highlight the frequencies of the characteristic bands below 800 cm1 for the a-, b- and g-forms at 338, 624 and 326 cm1, respectively. This journal is ª The Royal Society of Chemistry 2013 Fig. 6 Distribution of the conformers of [bmim]+-containing structures found in the CSD. Edge Article Chemical Science Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article Onlinevalues for the torsion angles q1 and q2 observed from single- crystal diffraction are summarised in Table 3. The conformers of the a- and g-phases are conrmed to be GT and G0T, respectively. The coexistence of the GT and G0T conformers in the a-phase, suggested by Endo et al.18 on the basis of variable-temperature Raman spectroscopy can also be rationalised: the single-crystal structure at 193 K indicates the presence of a minor (25%) component attributed to the G0T conformer. On further cooling the same crystal, the disorder disappears and the observed conformation is entirely GT. This observation points towards the presence of dynamic disorder in the cation, though ideally more data points should be collected to conrm this. Dynamic disorder or enhanced thermal motion in the cation is, to a lesser extent, also present in the g-form: whilst the structures reported by Dibrov and Kochi16 and Choudhury et al.17 determined below 200 K indicate that the cation conformer is G0T, our 263 K structure additionally shows the presence of a minor GT component (8%). Multiple confor- mations are also observed for the two molecules in the asym- metric unit of the b-phase at 193 K/0.1 MPa and at 293 K/0.07 GPa. When considering the main disordered components only, the conformations of the [bmim]+ cation at 193 K/0.1 MPa and at 293 K/0.07 GPa are best described as TG0/TT and TT/TT, respectively. When the minor disordered components are also taken into account, the description becomes TG0 (30%TE0)/TT at low temperature and TT (30%TE0)/TT (30%TE0) at high pressure. Crystal structures of ionic liquids crystallising with more than one molecule in the asymmetric unit do in fact oen contain different conformers; the E and E0 conformers are considerably less common compared to the T, G and G0 counterparts, but not unknown. A search for [bmim]+ containing structures in the current version of the Cambridge Structural Database (the CSD, V. 5.33 including updates to August 2012, was searched for structure containing the [bmim]+ cation with available 3-D coordinates, an R-factor below 10%, no errors and excluding powder structures) reveals that out of a total of 96 structures (90 unique), 1 contains twelve [bmim]+ cations in the asymmetric unit, another contains ve, 9 contain four, 2 contain three and 19 contain two. The distribution of the observed conformers isTable 3 Torsion angles and cation conformation for the polymorphs of [bmim][PF a(100 K) a(193 K) q1(N1–C7–C8–C9)/ (molecule 1) 63.65(19) 63.5(5) a [54.1(14)] q1(N11–C17–C18–C19)/ (molecule 2 b) q2(C7–C8–C9–C10)/ (molecule 1) 176.76(14) 176.2(3) [176.0(8)] q2(C17–C18–C19–C20)/ (molecule 2) Conformation molecule 1 GT GT [25% G0T] Conformation molecule 2 a Values in square brackets refer to the minor disordered component. b T This journal is ª The Royal Society of Chemistry 2013shown in Fig. 6. The wide range of different conformations observed in different polymorphs and in the same phase at different conditions of pressure and temperature is a further indication of the low barrier to rotational isomerism. In a very recent solid-state 1H NMR relaxation experiment study Endo et al.19 reported that the averaged mobility of the cation follows the order g < b # a and that the same trend is also followed by the slow segmented motion of the butyl group. These ndings are in general good agreement with our observations; moreover, in most cases the motions seem to be larger than librational motions and are associated with conformational changes of the butyl group. The TT conformer assignment for the b-phase by Endo et al. and Russina et al. on the basis of Raman data is essentially correct when considering the major disorder component; however, the availability of single-crystal data provides a higher level of detail. The model presented, herein, still represents an averaged model: the data do not allow a deconvolution of electron density and thermal motion and the given values should be taken as guidelines; in reality it is possible that a range of conformations between TG0 and TE0 exist in the crystal, in particular as a function of varying temperature and pressure.6] at different experimental conditions b(193 K) b(0.07 GPa) g(193 K)17 g(263 K) 179.8(12) 168(2) 60.80(17) 60.3(5) [164(3)] [56(4)] 167.5(8) 168(2) 64.4(19) 166(3) 178.05(17) 176.4(3) [104(2)] [135(3)] [173(3)] 180.0(8) 178(3) [122(4)] TG0 TT G0T G0T [30% TE0] [30% TE0] [8% GT] TT TT [30% TE0] wo molecules in the asymmetric unit for the b-form. Chem. Sci., 2013, 4, 1270–1280 | 1275 Chemical Science Edge Article Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article OnlineDisorder of the [PF6]  anion in the solid state Rotational disorder of the [PF6]  group is oen observed in the solid state, as expected for a highly symmetric molecule. At 193 K the only crystalline phase which does not show evidence of anion disorder is that of g, which incidentally is also the most dense phase at this temperature. In the a-phase the nature of the anion disorder is most likely to be dynamic, as exemplied by the ordering observed on cooling the single-crystal from 193 to 100 K. In the b-phase the two crystallographically indepen- dent anions are both extensively disordered about axial F–P–F bonds. The trends in the anion disorder in the crystalline state are in agreement with the results of the spin-lattice rotational dynamics study of Endo et al.19 A more detailed description of anion disorder can be found the ESI.†Fig. 8 Crystal packing of [bmim][PF6], a-form (top) and of 1-n-dodecyl-3- methylimidazolium hexafluorophosphate,32 CSD Ref. code HIWNOQ (bottom).The crystal structures of the three polymorphs The structure of the g-phase determined at 263 K is in very good agreement with the low-temperature structure reported previ- ously. The crystal packing in this form is based on two primary building blocks of centrosymmetric cation dimers. In the rst, here termed “closed” dimer, the butyl groups are pointing inwards, i.e. towards the space between two imidazolium rings, and in the second, “open” dimer, they are pointing outwards (Fig. 7). Both types of packing arrangements are found in short- and long-chain ionic liquids.11,31,32 In long alkyl structures, “closed” dimers favour hydrophobic alkyl–alkyl interactions. The b-phase also exhibits both types of packing arrangements: the “open” dimer is formed between symmetry equivalent molecules, whilst the “closed” one, a pseudo-centrosymmetric dimer, is formed between symmetry-independent molecules. In contrast to the g- and b-forms, the a-phase only exhibits one type of centrosymmetric dimeric arrangement, namely the “open” dimer; in addition, a different, less compact closed dimer, withmolecules related by glide symmetry, is formed. The nature of the “open” dimer also changes across the three structures: in the a- and b-phases the methyl groups of the imidazolium ring are aligned, whereas in the g-form the butyl groups are. When considering the entire crystal packing, an overall change from a mixed short-alkyl/long-alkyl to a long-Fig. 7 Dimeric arrangements in the three polymorphs of [bmim][PF6] viewed along have been omitted for clarity. oi ¼ centrosymmetric “open” dimer, ci ¼ centrosym “closed” dimer. Symmetry-independent molecules in the b-phase are labelled 1 and 1276 | Chem. Sci., 2013, 4, 1270–1280alkyl structure-type packing arrangement is observed when going from the g- through the b- to the a-phase. The phase that crystallises at the lowest temperature from the supercooled liquid state adopts a packing arrangement that is more akin to that encountered in long-alkyl chain hexauorophosphate ionic liquids (Fig. 8), with layers of interacting butyl side chains separated by c/2, similar to what was previously predicted for this compound.9 This is in line with the thermodynamic data of Triolo et al.,9 who determined that the supercooled liquid- fragility index of [bmim][PF6] is similar to that of longer side chain 1-alkyl-3-methyl imidazolium-based salts.33 The geometric parameters describing the centrosymmetric dimeric arrangements are detailed in Table S1 (see ESI for details†). p–p stacking of cation dimers is observed between dimers in the “open” arrangement, and the smallest offset to perfect stacking is found in the b-phase [2.342 A˚ at 193 K compared to 3.543 and 3.394 A˚ in the a- and g-phases, respec- tively, at the same temperature], although the interplanartwo different stacking directions. Anions, H-atoms andminor disorder components metric “closed” dimer, cpi ¼ pseudo-centrosymmetric “closed” dimer, cg ¼ glide 2. This journal is ª The Royal Society of Chemistry 2013 Edge Article Chemical Science Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article Onlinedistance is also the longest. At 193 K, the interplanar distance for the “closed” dimer is shortest in the g-phase [4.3688(9) A˚ compared to 5.452(4) A˚ in the b-phase]. From Table S1† the more pronounced pseudo-symmetry in the b-form at high pressure is evident: all parameters describing the “open” dimers between symmetry inequivalent molecules are more similar at high pressure than they are at low temperature. The [PF6]  anions contribute to structural stability by linking the dimeric arrangements via several F/H contacts. In the structures of the g- and b-phases, the anions occupy the space between the stacked cation columns, whilst in the structure of the a-phase the structure motifs formed by cations and anions are intertwined. Void analysis (performed with the program Mercury34) reveals that whilst this space forms continuous columns along the b-axis in the g-form, it gradually moves into a double-pocket arrangement and nally into segregated indi- vidual pockets that t a single anion on going from g to b and from b to a (Fig. 9). Interestingly, at 0.07 GPa the density of the b-phase is ca. 2.5% less than the corresponding density of the same phase at 193 K. Isothermal compression at 293 K to ca. 0.35 GPa, leads to a density increase of ca. 3.4% from 1.477 to 1.527 g cm3 (full structural data not reported here). The effect of decreasing temperature on the crystals of the g- and a-phases is to form denser structures. The considerable variation of density as a function of temperature is indicative of the “soness” of the intermolecular interactions that govern crystal packing. The attractive energy between imidazolium-based cations and anions is predominantly attributable to Coulombic forces. The contact involving the acidic C2–H2 bond and the anion is generally accepted as a hydrogen bond, albeit weak in the case of [PF6] .35,36 Tsuzuki et al., have shown by computational methods that hydrogen bonds in ionic liquids involving imi- dazolium ring hydrogens are not directional and are mainly governed by the distance between the ring hydrogen and the anion. In a subsequent publication, the authors have shown that a contact between a Br anion and the C2–H2 group sta- bilises the GT [bmim]+ cation conformer, whilst close contacts to C4–H4 and C5–H5 do not alter the relative stability of the GT, G0T and TT conformers.20 H/F contacts less than the sum ofFig. 9 Crystal packing in the three polymorphs of [bmim][PF6]. H-atoms and diso depicted by solvent accessible void surfaces generated with the programMercury (p expressed as % unit cell volume: 5.1% in the a-, 5.3% in the b- and 4.4% in the g-p This journal is ª The Royal Society of Chemistry 2013the van der Waals radii are given in Table S2 (see ESI for details†). From this table it can be seen that the coordination number of the cation and anion varies across temperature, pressure and crystal phase. Within the same phase, the coor- dination numbers are inversely proportional to crystal density; the higher density observed at low temperature is achieved by an increase in the number of close H/F contacts, dened as contacts whose distance is smaller than the sum of the van der Waals radii. The crystal density at 193 K follows the order a < b  g; the same order is observed in the number of close H/F contacts, whilst the order is a g > bwhen considering both the coordination number of cations and anions. These results are in fairly good agreement with the strength of cation–anion interactions estimated from Raman spectroscopy that follows the trend g > b $ a.19 All three crystalline phases are characterised by C/F contacts involving H2 to two anions on either side of the imi- dazolium ring. The anion is dened to lie above the ring when it is on the same side of the butyl chain and below when on the opposite side. Details of the geometric parameters for these contacts are given in the ESI (Table S3†). At 193 K, short C2/F distances are observed for the a-form [two contacts to disor- dered F-atoms at 3.062(6) and 3.064(9) A˚ to the anion below the ring; note that at 2.993(10) and 3.029(8) A˚, C/F contacts to the anion above the ring are shorter but H/F distances are greater than the sum of the van der Waals radii]. In the g-form the corresponding distance is slightly longer [3.192(4) A˚]; in this form, the distance to the anion above the ring is comparatively longer [3.233(5) A˚] but more linear (C2–H2/F angle 152 and 120, respectively). In the b-form a wider range of distances to anions involving H2 are observed because of the presence of two molecules in the asymmetric unit and extensive anion disorder, thereby making contact analysis more complex. For instance, at 193 K the shortest distance has a value of 2.921(16) A˚ and involves molecule 2 and the anion below the imidazolium ring; the disordered counterpart is longer [3.031(17)] A˚ and at 2.84 A˚ the H/F distance is greater than the sum of the van der Waals radii; the associated contact angles are 99 and 92, respectively. At 0.07 GPa and ambient temperature, the corresponding distances and angles are 2.74(3) A˚, 112 and 3.19(3) A˚, 118,rdered atoms have been omitted for clarity. The space occupied by the anions is robe radius¼ 1.2 A˚, grid spacing ¼ 0.7 A˚). Solvent accessible void volume at 193 K olymorph, respectively. Chem. Sci., 2013, 4, 1270–1280 | 1277 Chemical Science Edge Article Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article Onlinerespectively, and all H/F distances are less than the sum of the van der Waals radii. At high pressure, the number of short H/F contacts is lower than at low temperature, in line with the lower density, and the environment around the symmetry indepen- dent molecules more similar when considering average values, in line with the higher pseudo-symmetry. All three polymorphs exhibit close contacts that involve not only H2 but all ring as well as themethyl hydrogen atoms; overall, the more linear contacts are associated with longer C/F distances. The former observation seems to suggest that close contacts between these atoms and anions are not directly responsible for the observed cation conformation in the solid state. On inspection of Tables S2 and S3 deposited as ESI† other more indicative trends emerge. The GT conformer appears to be stabilised by a contact to the hydrogen atom attached to C7 and no contacts to C8 or C10 (for which distances are >3.7 A˚); the G0T conformer is additionally stabilised to contacts to C10 (the contact to H8 observed at 193 K is long: H8/F distance ¼ 2.67 A˚ and C8/F distance¼ 3.662(5) A˚). The difference between the TT, TG0 and TE0 conformers is not immediately evident but all involve additional contacts to the terminal butyl carbon atom. Differ- ences in the number of contacts involving methyl hydrogen atoms are inconclusive given that their positions are not reliable.General discussion Our experiments indicate that the a-polymorph is very sensitive to temperature changes and should be kept well below the transition temperatures to the b- or g-phases. The temperature range used for in situ crystallisation followed by crystal growth on a diffractometer by zone melting, e.g. using an OHCD apparatus, is usually quite large and difficult to control: this could partially explain why single-crystal structures reported in the literature correspond to the g-phase despite the fact that the crystal was grown at temperatures where the a-phase is stable.17 The reproducibility of the crystallisation protocols outlined in the Experimental section seems to be strongly dependent on a number of factors, including sample volume, water contami- nation and coexistence of liquid and solid phases. For instance, nucleation of the a-phase is most reproducible with sample volumes greater than ca. 200 mL; similarly, crystallisation of the b-phase appears to be more reproducible when the sample is contained in a capillary and the heating rates are high. Proto- cols 3 and 4 outlined in the Experimental section are most successful when the whole sample is crystalline: when the liquid and crystalline phases coexist, the a to liquid phase transition can occur at ca. 263 K, without a prior solid–solid phase transition. The presence of water as a contaminant, either as a liquid or as ice, also seems to facilitate the formation of the g-form, though this has not been tested systematically. Inter- estingly, the presence of water also seems to be crucial for the crystallisation of the g-phase under high-pressure conditions: preliminary experiments indicate that crystallisation of [bmim] [PF6] from very diluted aqueous solutions in the 0.4–0.7 GPa pressure range results in the formation of the g-form, as conrmed by single-crystal X-ray diffraction. In line with other authors, on increasing pressure, we did not observe the1278 | Chem. Sci., 2013, 4, 1270–1280crystallisation of the g-phase from the pure liquid, suggesting that, based on density differences, the application of modest pressure favours crystallisation of the b-phase. The existence of multiple conformers in the liquid state of [bmim][PF6] is well documented.18,37 A recent elegant study by Jeon el al. reported the nanocrystallisation of [bmim][PF6] at the vapour–liquid interface at 310 K by grazing-angle synchrotron X-ray diffraction.38 The sharp Bragg diffraction peaks observed by the authors, arising from nanocrystallites at the interface and coexisting with the broad diffraction features of the liquid, were indexed and a quasi-2D long-range ordered structure was extracted which was similar to that of the g-form. Whilst we also observe coexistence of crystalline and liquid states at certain conditions of temperature and pressure, our discrete Bragg reections collected in transmission geometry are very starkly dominated by the crystalline bulk: hence, the cation and anion disorder were analysed in terms of crystalline disorder. Surface diffractionmethods could shedmore light into the nature of the crystal–liquid interfaces. It could be conceivable, in particular for the b-phase, where disorder is more severe, that different cooling rates would result in a different distribution of conformers – however, data collected on different crystals suggest that the differences are not signicant, so that the main trends reported are conrmed. As indicated in the Experimental section, crystals of the b-phase obtained at high-pressure or low-temperature condi- tions are characterised by twinning. The b-phase was found to be stable between ca. 0.07 and 0.4 GPa. As expected, the lowest deviation from perfect monoclinic C symmetry was observed for the high-pressure structure collected at 0.07 GPa, i.e. at phase- boundary conditions, and this is also the structure exhibiting the highest degree of pseudo-symmetry and perfect domain overlap. Further compression of the single crystal from 0.07 to 0.35 GPa resulted in a unit-cell with a higher degree of deviation from a perfect monoclinic C cell, see ESI† for details. This is a small pressure range to probe the effects of pressure on disorder; however, our data analysis indicates no signicant changes in disorder on increasing pressure. Whilst the study of conformational exibility and energetic aspects of the [bmim]+ cation in the gas phase are useful, it is clear that an extrapolation of the ndings to structures adopted in the solid state is not entirely justied since DFT gas phase calculations on isolated cations ignore the effects of anions and the stabilising energy contributions that arise from crystal packing. It should be noted that whilst Tsuzuki et al. have studied anion effects in their conformational analysis, the [PF6]  anion was not included in the calculations. Hence, whilst the three most stable conformers in the gas phase are GT (0.02 kcal mol1), TT (0.00 kcal mol1) and G0T (0.50 kcal mol1) (geometries optimised at the MP2/6-311G** level and relative energies estimated at the CCSD(T)/cc-pVTZ level), the TG0 conformer observed for one of the two independent molecules in the b-phase at 193 K is only ca. 4 kJ mol1 less stable than the G0T rotamer and such a small barrier to rotation can be easily overcome by other stabilising intermolecular interactions in the solid state. The phenomenon of conformational polymorphism in the solid state has been well documented in the literature.This journal is ª The Royal Society of Chemistry 2013 Edge Article Chemical Science Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article OnlineNangia summarises this very nicely: “organic molecules with exible torsions and low-energy conformers have a greater likelihood of exhibiting polymorphism because (1) different conformations lead to new hydrogen bonding and close- packing modes and (2) the tradeoff reduces the total energy difference between alternative crystal structures”.39 These observations are also applicable in the case of the title compound and conrmed by the values of the phase transition enthalpies from DSC measuraments.18 Whilst it would be interesting to compare the difference in lattice and interaction energies of the three polymorphs by means of computational calculations, the observed disorder would make the analysis more difficult, and such a study is beyond the scope of this paper. Despite the widespread conformational exibility of the alkyl side chain and the large number of structures available in the CSD, there are not many examples of polymorphs reported for ionic liquids, but those reported are conformational poly- morphs; the low incidence of polymorphic structures might be partially due to the fact that the polymorphism of these compounds has not been extensively investigated; for examples see refs. 11, 31, 32 and 40–42. Although the g- and a-phases exhibit very similar Raman spectra, their crystal structures are profoundly different. The crystal structures of the three polymorphs are substantially different from one another: hence, X-ray powder diffraction can be an effective tool for rapid phase identication. Two forms were previously identied by Triolo et al.9 using wide-angle X-ray scattering (cr-I and cr-II, see Table 1). Comparison of simulated patterns (see Fig. S4 in the ESI†) derived from the single-crystal structures of the three forms with the patterns reported in this previous paper indicates the following: the patterns labelled cr-I (see Fig. 8, 9 and 10 of the original paper) correspond to the a- phase, whilst the patterns labelled cr-II correspond to the g-phase (see Fig. 9 and 10 of the original paper). The b-form was not reported. It is interesting to note that the powder data clearly indicate the occurrence of the a / g transition at 250 K, also conrmed by thermal analysis, despite the fact that a similar temperature was reported for the a/ b transition by Endo et al. Our experiments conrm that sample thermal history, impurities and heating/cooling rates are important factors affecting the polymorphic phase transitions in [bmim][PF6].Conclusions The polymorphic behaviour of the ionic liquid [bmim][PF6] has been studied by Raman spectroscopy, single-crystal X-ray diffraction and optical microscopy. The existence of three polymorphs, a, b and g, which can be crystallised as a function of low temperature or high pressure, has been conrmed and all three have been structurally characterised. The conformational exibility of the [bmim]+ cation, together with the capability of rotational disorder of the [PF6]  anion, give rise to different conformations in the solid state not only as a function of crys- tallisation conditions but also of changes in temperature and pressure to the same phase. The range of cation conformations identied provides a basis by which to validate simulation and modelling data for both fundamental studies and processThis journal is ª The Royal Society of Chemistry 2013design. The GT, TT and G0T conformations of the cation in the a-, b- and g-phases, respectively, inferred from previous analysis of Raman stretching frequency, is essentially conrmed; however, the availability for the rst time of single-crystal X-ray data for all polymorphs reveals a wider range of conformers than previously thought. Across the three phases and under different experimental conditions, a large variation in crystal packing densities and in structural disorder is observed. These variations are likely contributing driving forces for the poly- morphic interconversions. Whilst phase transitions do not occur in a single-crystal-to-single-crystal fashion and the molecular packing arrangements of the three forms are very distinct, some continuity in the structural motifs of the building blocks is observed.Acknowledgements This work was carried out with the nancial support of the Deutsche Forschungsgemeinscha (DFG Emmy Noether Grant FA 964/1-1), the EPSRC (EP/G012156/1) and Merck GMbH. We would like to thank: Prof. Dietmar Stalke and his group (Go¨ttingen) for access to the Ag microsource, Prof. George M. Sheldrick and Dr Birger Dittrich (Go¨ttingen) for access to the Cu rotating anode, Dr Junfeng Qin (Go¨ttingen) for initial assistance with Raman measurements, Drs Heidrun Sowa and Regine Herbst-Irmer (Go¨ttingen) as well as Michael Ruf (Madison) for useful discussions, Ulf Kahmann and Heiner Bartels for technical assistance and Prof. Simon Parsons (Edinburgh) for a copy of SHADE.Notes and references 1 M. J. Earle and K. R. Seddon, Pure Appl. Chem., 2000, 72(7), 1391. 2 P. Wasserscheid and T. Welton, Ionic Liquids in Synthesis, Wiley-VCH, Weinheim, 2007. 3 V. I. Parvulescu and C. Hardacre, Chem. Rev., 2007, 107(6), 2615. 4 W. M. Reichert, J. D. Holbrey, K. B. Vigour, T. D. Morgan, G. A. Broker and R. D. Rogers, Chem. Commun., 2006, 4767. 5 R. A. Judge, S. Takahashi, K. L. Longenecker, E. H. Fry, C. Abad-Zapatero and M. L. Chiu, Cryst. Growth Des., 2009, 9(8), 3463. 6 J.-H. An, J.-M. Kim, S.-M. Chang andW.-S. Kim, Cryst. Growth Des., 2010, 10, 3044. 7 T. G. A. Youngs, C. Hardacre and C. L. Mullan, Ionic Liquids in Chemical Analysis, CRC Press, Boca Raton, FL, 2008, p. 73. 8 C. Hardacre, J. D. Holbrey, M. Nieuwenhuyzen and T. G. A. Youngs, Acc. Chem. Res., 2007, 40, 1146. 9 A. Triolo, O. Russina, C. Hardacre, M. Nieuwenhuyzen, M. A. Gonzalez and H. Grimm, J. Phys. Chem. B, 2005, 109, 22061. 10 F. H. Allen, Acta Crystallogr., Sect. B: Struct. Sci., 2002, 58, 380. 11 J. D. Holbrey, W. M. Reichert, M. Nieuwenhuyzen, S. Johnston, K. R. Seddon and R. D. Rogers, Chem. Commun., 2003, 1636.Chem. Sci., 2013, 4, 1270–1280 | 1279 Chemical Science Edge Article Pu bl ish ed o n 02 Ja nu ar y 20 13 . D ow nl oa de d by U ni ve rs ity o f G oe tti ng en o n 22 /0 7/ 20 14 1 1: 52 :3 3. View Article Online12 B. L. Bhargava and S. Balasubramanian, J. Chem. Phys., 2007, 127, 114510. 13 B. L. Bhargava and S. Balasubramanian, J. Phys. Chem. B, 2007, 111, 4477. 14 M. Bu¨hl, A. Chaumont, R. Schurhammer and G. Wipff, J. Phys. Chem. B, 2005, 109, 18591. 15 M. G. Del Po´polo, R. M. Lynden-Bell and J. Kohanoff, J. Phys. Chem. B, 2005, 109, 5895. 16 S. M. Dibrov and J. K. Kochi, Acta Crystallogr., Sect. C: Cryst. Struct. Commun., 2005, 62, o19. 17 A. R. Choudhury, N. Winterton, A. Steiner, A. I. Cooper and K. A. Johnson, J. Am. Chem. Soc., 2005, 127, 16792. 18 T. Endo, T. Kato, K.-i. Tozaki and K. Nishikawa, J. Phys. Chem. B, 2010, 114, 407. 19 T. Endo, H. Murata, M. Imanari, N. Mizushima, H. Seki and K. Nishikawa, J. Phys. Chem. B, 2012, 116, 3780. 20 S. Tsuzuki, A. A. Arai and K. Nishikawa, J. Phys. Chem. B, 2008, 112, 7739. 21 O. Russina, B. Fazio, C. Schmidt and A. Triolo, Phys. Chem. Chem. Phys., 2011, 13, 12067. 22 Y. Yoshimura, T. Takekiyo, Y. Imai and H. Abe, in Ionic Liquids – Classes and Properties, ed. S. T. Handy, InTech, 2011, ch. 8. 23 L. Su, M. Li, X. Zhu, Z. Wang, Z. Chen, F. Li, Q. Zhou and S. Hong, J. Phys. Chem. B, 2010, 114, 5061. 24 L. Su, L. Li, Y. Hu, C. Yuan, C. Shao and S. Hong, J. Chem. Phys., 2009, 130, 184503. 25 R. G. de Azevedo, J. Esperanca, V. Najdanovic-Visak, Z. P. Visak, H. J. R. Guedes, M. N. da Ponte and L. P. N. Rebelo, J. Chem. Eng. Data, 2005, 50, 997. 26 Y. Zhao, X. Liu, X. Lu, S. Zhang, J. Wang, H. Wang, G. Gurau, R. D. Rogers, L. Su andH. Li, J. Phys. Chem. B, 2012, 116, 10876. 27 F. P. A. Fabbiani, D. C. Levendis, G. Buth, W. F. Kuhs, N. Shankland and H. Sowa, CrystEngComm, 2010, 12, 2354.1280 | Chem. Sci., 2013, 4, 1270–128028 G. J. Piermarini, S. Block, J. D. Barnett and R. A. Forman, J. Appl. Phys., 1975, 46, 2774. 29 R. W. Berg, Ionic Liquids in Chemical Analysis, CRC Press, Boca Raton, FL, 2008, p. 307. 30 R. W. Berg, M. Deetlefs, K. R. Seddon, I. Shim and J. M. Thompson, J. Phys. Chem. B, 2005, 109, 19018. 31 A. Downard, M. J. Earle, C. Hardacre, S. E. J. McMath, M. Nieuwenhuyzen and S. Teat, J. Mater. Chem., 1998, 8, 2627. 32 C. M. Gordon, J. D. Holbrey, A. R. Kennedy and K. R. Seddon, J. Mater. Chem., 1998, 8, 2627. 33 J. J. Moura Ramos, C. A. M. Afonso and L. C. J. Branco, J. Therm. Anal. Calorim., 2003, 71, 659. 34 C. F. Macrae, I. J. Bruno, J. A. Chisholm, P. R. Edgington, P. McCabe, E. Pidcock, L. Rodriguez-Monge, R. Taylor, J. van de Streek and P. A. Wood, J. Appl. Crystallogr., 2008, 41, 466. 35 C. Hardacre, S. E. J. McMath, D. T. Bowron and A. K. Soper, J. Chem. Phys., 2003, 118, 273. 36 S. Tsuzuki, H. Tokuda and M. Mikami, Phys. Chem. Chem. Phys., 2007, 9, 4780. 37 M. Macchiagodena, L. Gontrani, F. Ramondo, A. Triolo and R. Caminiti, J. Chem. Phys., 2011, 134, 114521. 38 Y. Jeon, D. Vaknin, W. Bu, J. Sung, Y. Ouchi, W. Sung and D. Kim, Phys. Rev. Lett., 2012, 108, 055502. 39 A. Nangia, Acc. Chem. Res., 2008, 41(5), 595. 40 J. Backer, S. Mihm, B. Mallick, M. Yang, G. Meyer and A.-V. Mudring, Eur. J. Inorg. Chem., 2011, 4089. 41 M. Kawahata, T. Endo, H. Seki, K. Nishikawa and K. Yamaguchi, Chem. Lett., 2009, 38, 1136. 42 Y. U. Paulechka, G. J. Kabo, A. V. Blokhin, A. S. Shaplov, E. I. Lozinskaya, D. G. Golovanov, K. A. Lyssenko, A. A. Korlyukov and Y. S. Vygodskii, J. Phys. Chem. B, 2009, 113, 9538.This journal is ª The Royal Society of Chemistry 2013