The Madelung Constant of Organic Salts 2009, Vol

DOI: 10.1021/cg900656z

The Madelung Constant of Organic Salts 2009, Vol. 9 4834–4839 Ekaterina I. Izgorodina,* Uditha L. Bernard, Pamela M. Dean, Jennifer M. Pringle, and Douglas R. MacFarlane* School of Chemistry, Monash University, Clayton, Victoria 3800, Australia

Received June 11, 2009; Revised Manuscript Received August 18, 2009

ABSTRACT: The Madelung constant is a key feature determining the lattice energy of a crystal structure and hence its stability. However, the complexity of the calculation has meant that it has previously not been readily available for complex structures, for example for organic salts. We propose a new robust method for calculating Madelung constants of such structures based on a generalized numerical direct summation approach. The method is applied to various organic salts from the ionic liquid and pharmaceutical fields. The values calculated are seen to be a unique feature of the crystal structure, reflecting the positioning of the ions in the unit cell and being sensitive to ion pairing. The difference in Madelung constants between different polymorphs of a compound is also shown.

Introduction general method of calculating the Madelung constant directly from a known crystal structure. We describe here the basis of The much-discussed Madelung constant is an important the approach we have developed and a number of example fundamental aspect of our understanding of the solid state. It calculations for organic salts of interest in both the ionic liquid describes the total interaction energy of an ion in a crystalline and pharmaceutical fields. The electrostatic energy contribu- lattice, with all of the other ions, and hence provides us with an tion to the lattice energy derived from these Madelung con- understanding of the origins of the lattice energy of an ionic stants is compared with a variety of inorganic salts and the crystal. The Madelung constant of various simple inorganic relative importance of electrostatic interactions is discussed. salts has been the focus of much attention for nearly 100 years The idea of the Madelung constant as a measure of the net and a variety of approaches to its calculation have been electrostatic interaction energy, E , in ionic solids was intro- developed.1-14 However, these approaches are not easily es duced by Madelung in 1918.1 He summed all electrostatic generalized to provide a mechanism for the calculation of interactions experienced by an ion in the crystal structure and the Madelung constant of more complex ionic crystals, for then expressed the total as example, organic salts, and hence the detailed understanding that it provides is largely lacking in respect of such crystals. On Mqcationqanion Ees 1 the other hand, complex organic salts have become a major ¼ 4πε0dmin ð Þ area of interest in recent years for a variety of reasons. For example, about 50% of known pharmaceutical compounds where dmin is the distance to the nearest counterion. If one are produced as organic salts15 and their solid-state phase considers the electrostatic binding energy of a single ion pair IP IP behavior in terms of polymorphism is of very great concern to be Ees then Ees = MEes and hence M represents the degree from the point of view of stability and bioavailability.16 In a to which the lattice of ions is more stable than the isolated different arena, the field of ionic liquids has generated a huge collection of ion pairs would be. For stable crystal structures Downloaded via MONASH UNIV on September 26, 2019 at 01:43:25 (UTC). range of new organic salts17-20 and an intense interest in M > 1 and the larger M is, the more stable the crystal understanding and predicting the melting points of these structure will be. For simple inorganic salts Ees makes up a 21,22 salts. As an offshoot of that field, a number of these salts large part of the lattice energy, UL, of the salt: See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. exhibit plastic phases in their crystalline state and these are of UL Ees ESR 2 growing interest as solid-state ion conductors.23 ¼ þ ð Þ The phase behavior of these organic compounds is often a where ESR is the energy of short-range interactions including subtle combination of long-range and short-range interac- atom-atom repulsive interactions. In the case of organic salts, tions. The short-range interactions are relatively easy to these short-range interactions include a number of important identify and understand when a crystal structure is available. attractive components including, for example, π-π interac- Tools such as the Hirshfeld surface have been of great tions, hydrogen bonding, and van der Waals interactions. 24 assistance in this respect. However, the long-range interac- The Madelung constant can in principle always be calcu- tions are electrostatic and to understand their relative impor- lated by choosing an ion as the origin and then summing all tance one effectively needs to be able to obtain the Madelung possible electrostatic interactions with surrounding ions at 21 constant for the crystal structure. Hence, as part of our increasing distances from the origin. An (effectively) infinite effort in designing and understanding the properties of or- series of contributions, the Madelung series, is generated in ganic salts for a variety of applications, we sought to develop a which each term represents electrostatic interactions with ions at a particular distance from the origin. Since Madelung’s original work, most of the calculations have followed the *Corresponding authors. E-mail: [email protected] (E.I.I.), [email protected] (D.R.M.). Tel.: 61-3-9905-8639 same direct summation approach aiming to generate a series (E.I.I.), 61-3-9905-4540 (D.R.M.). Fax: 61-3-9905-4597.þ expansion for the Madelung constant, which converges to a þ þ

pubs.acs.org/crystal Published on Web 09/18/2009 r 2009 American Chemical Society Article Crystal Growth & Design, Vol. 9, No. 11, 2009 4835

Table 1. Melting Points (Tm), Number of Ions, Calculated and Literature Madelung Constants (M) and Electrostatic Energy Densities (EED) of Traditional Crystals Together with Number of Ion Pairs in the Unit Cell (N), Crystal Structure Type and Space Group

Tm, number Ees, EED source of salta °C N crystal type space group of ionsb MM(literature) kJ mol-1 108, J/mÂ3 crystal structure ZnS 170037 4 cubic F43m 1.4 107 1.640 ( 0.006c 1.63812 -3899.2 1633 38 39 Â 7 c 2 CaF2 1418 4 cubic Fm3m 2.1 10 2.524 ( 0.010 2.5194 -2974.1 1213 38 39 Â 7 c 40 CaCl2 772 2 orthorhombic Pnnm 1.1 10 2.364 ( 0.005 2.365 -2432.6 477 40 NaCl 80139 4 cubic Fm3m 1.8 Â 104 1.7476 1.74762 -862.6 318 38 CsCl 64539 1 cubic Pm3m 3.5 Â 106 1.768 ( 0.007c 1.76272 -690.0 163 38 41 Â 5 [NMe4][BF4] 443 2 tetragonal P4/nmm 8.0 10 1.6859 n/a -489.6 40.8 42 d 43 Â 6 [C3mpyr][Cl] 245 8 orthorhombic Pbcn 1.7 10 1.5358 n/a -549.9 38.0 44 45 Â 6 [Mep][Sacc] 188 4 orthorhombic P212121 2.0 10 1.1841 n/a -406.0 10.9 45 [Prop][Br] 16745 8 tetragonal P4/n 5.4 Â 104 1.4857 n/a -451.7 13.2 45 46 Â 5 [C1mpyr][NTf2] 132 4 monoclinic P21/c 8.9 10 1.3583 n/a -416.6 18.6 47 48 Â 6 c [C2mpyr][tos] 120 2 monoclinic Pn 2.4 10 1.408 ( 0.004 n/a -519.5 22.4 49 48 Â 6 [C4mpyr][tos] 115 2 triclinic P1 1.0 10 1.2963 n/a -504.1 19.1 49 41 Â 6 c [C4mim][Cl] 66 4 orthorhombic Pna21 4.7 10 1.379 ( 0.004 n/a -474.2 32.8 50 41,e Â 6 [C4mim][Cl] 41 4 monoclinic P21/c 2.4 10 1.3447 n/a -450.3 31.0 50 d 51 Â 5 [C1mpyr][NMes2] 40 4 monoclinic P21/n 3.4 10 1.1932 n/a -413.5 21.7 52 d 52 Â 6 [C4mpyr][NMes2] 36 8 orthorhombic Pbca 3.5 10 1.1639 n/a -424.9 17.9 52 [C mpyr][NTf ] -1846 4 orthorhombic P 2 2 2 5.6 Â 105 1.2854 n/a -423.1 15.5 53 4 2 1 1 1 Â a Abbreviations have the following meanings: Cnmpyr = (Cn-alkyl)methylpyrrolidinium, C4mim = butylmethylimidazolium, Mep = mepenzolate, - - Prop = propanthalate, Sacc = saccharinate, tos = tosylate, Mes = CH3SO3 , Tf = CF3SO3 . Structures of the propanthalate and mepenzolate ions are shown in Figures 2 and 3. b Number of ions needed in the calculation to reach convergence of the lattice energy using the EUGEN method. c Cases where the unit cell has a nonzero dipole moment and therefore convergence reached using the Harrison method. d DSC measurements carried out from -150 to 150 °C at a scan rate of 10 °C min-1. Onset temperatures are reported in all cases. e A broad peak characteristic of plastic crystal transitions. limit. Emersleben developed two approaches to these direct also calculate, for the first time, Madelung constants of summation calculations: (1) by considering ever expand- selected examples of organic salts of interest in the ionic ing cubes and (2) by considering ever expanding spheres with liquid/plastic crystal field and the pharmaceutical salt field. respect to the ion at the origin.3 However, the direct summation approach becomes intractable for all but the simplest of Results inorganic salt structures. There are two ways to overcome this problem: the Ewald method4 and the Evjen method.5 How- Accuracy and Convergence. The Madelung constants cal- ever, in both methods the sum representing the net electro- culated for a number of simple inorganic and organic salts static interactions is conditionally convergent, which means are compared in Table 1. The crystal structures of the that the result depends on the order of summation if the unit two tosylate salts are described here for the first time 25 cell possesses a nonzero dipole moment. A discussion of (CCDC Refcodes: 713020 for [C2mpyr][tos] and 713021 for such calculations and the various corrections required can be [C4mpyr][tos]). The comparison in Table 1 shows that the 6 found in the work of Wood. Neither of these methods has values calculated for the Madelung constants of ZnS, CaF2, made the evaluation of the Madelung constant straightfor- CaCl2, NaCl, and CsCl using the EUGEN method are in ward and able to be carried out in a generic way, with a excellent agreement with values reported in the literature. In number of factors affecting the outcome of the calculations. the case of the organic salts convergence typically requires To improve convergence, modifications to both methods have 24 iterations on average. The selected convergence threshold been thoroughly studied and introduced.7-14,26,27 Among the corresponds to only 10-3 kJ mol-1 in the lattice energy. simplest and easily achievable technique is the method intro- Loosening the threshold to 1 kcal mol-1 (traditionally duced by Harrison.28 Importantly, this method does not suffer considered to be “chemical accuracy”) further shortens the from conditional convergence like the Ewald and Evjen calculation time. Nonetheless, the calculation for even the methods; that is, it always converges to the correct Madelung most complex crystal structures in Table 1 still only requires constant. a few minutes on a laptop computer. In this work, we describe a generic approach to calculating Organic Salt Madelung Constants. The organic salts stu- the Madelung constant that involves a direct numerical sum- died here are comprised of singly charged ions and, therefore, mation, and which can therefore be used even with the a direct comparison with the Madelung constants of NaCl complex crystal structures that occur with organic salts. In and CsCl can be made. The Madelung constants of all of the principle, it can be used to obtain the Madelung constant for organic salts studied here are lower, in some cases consider- any known crystal structure including nonorthogonal unit cell ably so, than those of NaCl and CsCl, which is in accord with structures and those containing more than one ion pair in their much lower melting points. For example, salts such as the asymmetric unit, all very common circumstances. The [C1mpyr][NMes2], [C4mpyr][NMes2], and [C4mpyr][NTf2], approach could be integrated into standard crystallography with melting points below 100 °C, have Madelung constants or crystal structure display software. The unit cells can be of at least 25% lower than that of NaCl. The arrangement of any shape: cubic, orthorhombic, monoclinic, etc., and contain ions in the unit cell with respect to one another is a major an unlimited number of ion pairs. In this sense, the proposed factor in determining the Madelung constant of the organic method can be considered as an expanding unit-cells general- salts, much more so than is the case for the inorganic salts. ized numerical method, which we refer to as the EUGEN For example, [NMe4][BF4], Figure 1, crystallizes in a cubic method. We demonstrate the convergence and accuracy of the system with very uniformly arranged positions of the ions EUGEN calculation for known crystal structures such as with respect to all of their neighbors. To simplify inspection of NaCl, body-centered cubic CsCl, CaF2, ZnS, and CaCl2 and the structure, we show in Figure 1b a version in which only the 4836 Crystal Growth & Design, Vol. 9, No. 11, 2009 Izgorodina et al.

Figure 1. Packing diagrams for [NMe4][BF4] viewed down the c-axis showing (a) full unit cell contents (the BF4 ions are shown with their measured disorder) and (b) charged atoms only.

Figure 2. Packing diagrams for [Prop][Br] as viewed down the c-axis showing (a) full unit cell contents and (b) charged atoms only. charged atoms are visible. As a result of the highly symme- mepenzolate cation is a skeletal muscle relaxant sold com- trical packing the Madelung constant, at 1.69, is only very mercially as the Br- salt under the trade name Cantil. In this slightly lower than that of NaCl (M = 1.7476) or CsCl (M = structure, we see more distinct ion pairs and a correspond- 1.7627). On the other hand, Figure 2 shows the structure of ingly lower Madelung constant of 1.18. The fact that this propanthaline bromide ([Prop][Br]); this compound is a compound has a melting point as high as 188 °C, despite this muscarinic acetylcholine receptor antagonist sold under the very low Madelung constant, is a clear sign of the strong trade name Pro-Banthine. Figure 2b shows that the charged anion-cation π-π stacking that is present in this salt. atoms are positioned in clusters in the unit cell, with distinct For ionic liquids containing the same anion and the cation ion-pairing also present (i.e., one counterion in much closer with the alkyl chains of different length we can easily discern proximity than any other). This lowers the Madelung con- that the melting point decreases in accord with the Madelung stant considerably, to 1.49. To put this effect in context, a constant decrease. For example, this trend is observed for the tight ion pair that is well isolated in a large neutral species following pairs of ILs: (1) [C1mpyr][[NMes2] and [C4mpyr]- from neighboring ion pairs could have a Madelung constant [[NMes2] and (2) [C1mpyr][[NTf2] and [C4mpyr][[NTf2]. approaching 1.0. This situation is illustrated to an extent by However, comparison of the two pairs with each other shows mepenzolate saccharinate ([Mep][Sacc]) in Figure 3; the that the Madelung constant of [C4mpyr][[NTf2] is larger than Article Crystal Growth & Design, Vol. 9, No. 11, 2009 4837

Figure 3. Packing diagrams for [Mep][Sacc] as viewed down the a-axis showing (a) full unit cell contents and (b) charged atoms only (the negatively charged center is shown as a larger symbol to indicate the fact that the charge is likely to be delocalized to some extent in this anion). that of [C4mpyr][[NMes2] and yet the melting point of the three oxygens simultaneously and we will address this ap- former is lower by about 50 degrees. This finding further proach in a future paper. The fact that the effective Made- reinforces the importance of additional specific interactions lung constants of the three oxygens in the lattice are so (such as hydrogen bonding, van der Waals interactions, different indicates that they experience quite different elec- π-π stacking interactions etc.) in the lattice energy, UL, of trostatic environments and one of the oxygens in particular the salt. Therefore, accurate estimation of these interactions (O3 in Figure S3, Supporting Information) experiences more is needed before a correlation between UL and the melting attraction than the others. This may itself have an impact on point can be established. the actual electron distribution in the ion within the crystal and Also listed in Table 1 is the electrostatic contribution, Ees, the electric field vector, which is implicitly obtainable from the to the lattice energy calculated from the Madelung constant EUGEN method, could be used in an advanced quantum using eq 1. The values for the inorganic structures range from chemical calculation to further investigate this effect. more than 1000 kJ/mol for the divalent metal salts, to values Polymorphism. The Madelung constant itself is effectively around 800 kJ/mol for NaCl. The values for the organic salts a geometric parameter. While each space group will have a are typically much lower than these, reflecting the lower unique Madelung constant associated with it when the electrostatic interaction involved. structure involves close packed ions, this will not necessarily Charge Delocalization. Charge delocalization in molecular be true where the ions are molecular species. Two salts ions potentially places fractional charges on several atoms. crystallizing in the same space group may have different This effect is relatively easy to deal within the EUGEN Madelung constants if the charged atoms have different method if the details of the charge delocalization are known, locations in the molecular ions. For organic salts, then, the or can be estimated. The case of the tosylate anion illustrates Madelung constant is very much a unique feature of each the simplest approach; the value listed in Table 1 is the individual compound. Where polymorphism exists for an average of the Madelung constants calculated by assuming organic salt, each polymorph will also very likely have a that the unit negative charge is present on each of the oxygens different Madelung constant. The change in lattice energy in turn, producing values: 1.387, 1.500, and 1.338. A more that the change in Madelung constant describes therefore sophisticated approach would place partial charges on the becomes an important aspect of understanding the phase 4838 Crystal Growth & Design, Vol. 9, No. 11, 2009 Izgorodina et al. transformation. A good example of this can be seen in the take place through regions of, for example, hydrocarbon case of the polymorphs of 1-butyl,3-methyl imidazolium chains; the assumption that the material is isotropic also does chloride (C4mimCl) listed in Table 1. The Madelung con- not necessary hold. Measurements and investigations of the stant of the monoclinic form is distinctly lower, and in energy components of dielectric constants in organic liquids and terms this amounts to about 24 kJ/mol difference in lattice solids are ongoing in several groups32-36 and will assist in energy (other factors being equal); as a result it would be exploring this issue further. However, the EUGEN approach expected that this difference in lattice energy could drive a allows this issue to be explicitly accounted for in the calcula- solid-solid transformation of this form to the more stable tion by inclusion of different values for the average dielectric orthorhombic form. As observed by Holbrey et al.41 the constant, or potentially a spatially varying dielectric constant. orthorhombic form is indeed more thermodynamically stable than the monoclinic form. Conclusions

Discussion In this paper, we have introduced an expanding unit-cells generalized numerical (EUGEN) method for the calculation of The Madelung constant allows us to calculate an electro- the Madelung constants of salts with a known crystal structure. static energy density (EED) for the structure: The approach has been validated against Madelung constants available in the literature for inorganic salts. Madelung constants E N EED es J=m3 3 of a number of organic salts of interest in the ionic liquid/plastic ¼ V ð ÞðÞ crystal and pharmaceutical fields are reported here for the first time and it was shown that the values were lower than those of the where V is the unit cell volume and N is the number of ion pairs inorganic salts, falling in the range of 1.16 to 1.69. Organic salts in the unit cell. The EED values are also reported in Table 1. with highly symmetrical packing, such as [NMe4][BF4], can dis- For most salts the EED makes up a large part of the lattice play Madelung constants almost as high as that of NaCl. On the energy (expressed in unit volume terms) of the structure, and other hand, distinct ion pairing in a crystal, as observed in the therefore it is a useful quantity for direct comparison of crystal structure of [Mep][Sacc], results in a considerable decrease different compounds or different polymorphs of the same of the Madelung constant, approaching 1.0. The Madelung compound. This quantity is a more comprehensive measure of constants calculated are currently only a first approximation, electrostatic interactions in a material, because it takes ac- however the EUGEN approach paves the way for more advanced count of the very different concentrations of charged species calculations, which might properly account for any charge delo- (i.e., number of ions per unit volume) that can occur when calization in the ions or the effective screening of electrostatic comparing inorganic and various, much larger, organic ions. interaction by other regions of the molecule. The EED values of the organic salts in Table 1 are as much as 2ordersofmagnitudelowerthanthoseoftheinorganicsalts, Methods whereas Ees,whichisamolarquantity,variesbyonlyabout Expanding Unit-Cells Generalized Numerical (EUGEN) Method. 1orderofmagnitude.Ingeneral,theEEDtrendalsoshowsa A detailed discussion of the method and the implemented code are fairly good correlation with melting point until it drops to around provided in the Supporting Information. The method requires a fully 20 108 J/m3,whereundoubtedlyshort-rangeinteractionsbegin solved crystal structure as input data, from which a list of the Â fractional coordinates of all of the charged atoms in the unit cell to make an important contribution to the lattice energy. has been produced. The calculation then proceeds in a series of In general, due to the nature of most organic salts, such iterations involving a larger number of ions at each step. At the nth short-range interactions make a more significant contribution iteration, the calculation constructs an expanded cell containing to the total lattice energy than in inorganic salts. For example, n replicates of the unit cell along each of the unit cell directions. It the attraction due to electron correlation between imidazo- then sums all of the pairwise interactions between each ion in the lium rings is about 28 kJ mol-1 (as estimated at the CCSD(T) expanded cell and the central ion. It continues expanding the calcula- level of theory29), which is a non-negligible contribution to the tion cell until convergence is reached. As is well-known, such a calculation is conditionally convergent if the unit cell possesses a total lattice energy of salts based on this cation, for example, nonzero dipole moment;25 this occurs occasionally in organic salt the C4mim salts in Table 1. Since the type of interactions crystal structures. The calculation checks for this property and, when present in a particular organic salt depends very much on the present, adopts the Harrison approach to achieving convergence. nature of the constituting ions, approaches to their quantifi- Since the result in this case tends to produce a slow oscillation toward cation, of the type being intensely investigated in the compu- convergence, the software calculates a moving average (and standard tational chemistry field, are needed if a complete picture of the deviation) of the five most recent iterations and reports this value 30,31 when the convergence condition is reached. If the convergence is not lattice energy is to be achieved. reached after 60 iterations, the calculation stops and the final Made- A “second order” issue that exists in all approaches to lung constant is calculated as an average of the 5 last iterations. calculating long-range electrostatic interactions can also be dealt with by an appropriate expansion of the EUGEN code Acknowledgment. We gratefully acknowledge generous presented here. The Madelung constant is normally calculated allocations of computing time from the National Facility of on the basis that the electrostatic interaction between two ions the National Computational Infrastructure. E.I.I. gratefully occurs through a medium of dielectric constant equal to 1. In a acknowledges the support of the Australian Research Council real ionic material, including NaCl, this is plainly not the case, for her postdoctoral fellowship, as does D.R.M. for his since ions/atoms of varying polarizability may lie between the Federation Fellowship. interacting ions. In the classical Madelung analysis and all of the more modern approaches to the calculation, including the Supporting Information Available: The supporting information Ewald summation, this is ignored, partly on the basis that the contains a thorough description of the EUGEN method as well as the program that implements this method. Examples of the input attractive and repulsive interactions are equally affected. In a files needed to run the program are also given. In addition, the material consisting of mainly organic cations and anions this supporting information provides details of the crystal structures, issue becomes more acute since some of the interactions may [C2mpyr][tos] and [C4mpyr][tos], reported here for the first time. Article Crystal Growth & Design, Vol. 9, No. 11, 2009 4839

The program code (written in Fortran 95) for the EUGEN method (30) Jurecka, P.; Sponer, J.; Cerny, J.; Hobza, P. Phys. Chem. Chem. is available for download from http://www.chem.monash.edu/ Phys. 2006, 8, 1985–1993. ionicliquids. This material is available free of charge via the Internet (31) Khaliullin, R. Z.; Cobar, E. A.; Lochan, R. C.; Bell, A. T.; at http://pubs.acs.org. Head-Gordon, M. J. Phys. Chem. A 2007, 111, 8753–8765. (32) Schrodle, S.; Annat, G.; MacFarlane, D. R.; Forsyth, M.; Buchner, References R.; Hefter, G. Chem. Commun. 2006, 1748–1750. (33) Stoppa, A.; Hunger, J.; Buchner, R.; Hefter, G.; Thoman, A.; (1) Madelung, E. Phys. Zs. 1918, 19, 524–533. Helm, H. J. Phys. Chem. B 2008, 112, 4854–4858. (2) Housecroft, C. E.; Sharp, A. G. Inorganic Chemistry, 2nd ed.; (34) Schroder, C.; Haberler, M.; Steinhauser, O. J. Chem. Phys. 2008, Pearson Prentice Hall: New York, 2005. 128, 134501. (3) Emersleben, O. Naturwissenschaften 1959, 46, 64–65. (35) Weingartner, H. Z. Phys. Chem. 2006, 220, 1395–1405. (4) Ewald, P. P. Ann. Phys. 1921, 369, 253–287. (36) Izgorodina, E. I.; Forsyth, M.; MacFarlane, D. R. Phys. Chem. (5) Evjen, H. M. Phys. Rev. 1932, 39, 675–687. Chem. Phys. 2009, 11, 2452–2458. (6) Wood, R. H. J. Chem. Phys. 1960, 32, 1690–1692. (37) Lide, D. R. CRC Handbook of Chemistry and Physics, 86 ed.; CRC (7) Harris, F. E.; Monkhorst, H. J. Phys. Rev. B 1970, 2, 4400. Press: Taylor & Francis, Boca Raton, 2005. (8) Marshall, S. L. J. Phys.: Condens. Matter 2000, 12, 4575–4601. (38) Rohrer, G. S. Structure and Bonding in Crystalline Materials; (9) Yakub, E.; Ronchi, C. J. Low Temp. Phys. 2005, 139, 633–643. Cambridge University Press: Cambridge, 2001. (10) Johnson, E. J. Chem. Phys. 1996, 105, 5529–5531. (39) Chase, M. W.; Davies, C. A.; Downey, J. R.; Frurip, D. J.; (11) Fennell, C.; Gezelter, D. J. Chem. Phys. 2006, 124, 234104. McDonald, R. A.; Syverud, A. N. J. Phys. Chem. Ref. Data (12) Zucker, I. J. J. Phys. A 1975, 8, 1734–1745. 1985, 14. (13) Lowery, M. D.; House, J. E. Comput. Chem. 1984, 8, 249–254. (40) Johnson, Q. C.; Templeton, D. H. J. Chem. Phys. 1961, 34, 2004– (14) Lundeen, J. E.; House, J. E. J. Thermochim. Acta 1979, 31, 251–257. 2007. (15) Handbook of Pharmaceutical Salts: Properties, Selection, and Use; (41) Zabinska, G.; Ferloni, P.; Sanesi, M. Thermochim. Acta 1987, 122, Stahl, P. H., Wermuth, C. Eds.; Wiley VCN: Zurich, 2002. 87–94. (16) (a) Polymorphism in the Pharmaceutical Industry; Hilfiker, R., Ed.; (42) Giuseppetti, G.; Mazzi, F.; Tadini, C.; Ferloni, P.; Torre, S. Z. Wiley VCH Weinheim: Germany,(b) Polymorphism in Pharmaceuti- Kristallogr. 1992, 202, 81. cal Solids; Brittain, H. G., Ed.; Marcel Dekker, Inc., New York, 1999. (43) Laus, G.; Bentivoglio, G.; Kahlenberg, V.; Griesser, U. J.; Schottenberger, (17) Rogers, R. D.; Seddon, K. R. Science 2003, 302, 792–793. H.; Nauer, G. CrystEngComm 2008, 10,748–752. (18) Forsyth, S. A.; Pringle, J. M.; MacFarlane, D. R. Aust. J. Chem. (44) Dean, P. M.; Pringle, J. M.; MacFarlane, D. R. Acta Crystallogr. 2004, 57, 113–119. 2008, E64, o637. (19) Hough, W. L.; Smiglak, M.; Rodriguez, H.; Swatloski, R. P.; (45) Dean, P. M.; Turanjanin, J.; Yoshizawa-Fujita, M.; MacFarlane, Spear, S. K.; Daly, D. T.; Pernak, J.; Grisel, J. E.; Carliss, R. D.; D. R.; Scott, J. L. Cryst. Growth Des. 2009, 9, 1137–1145. Soutullo, M. D.; Davis, J. H.; Rogers, R. D. New J. Chem. 2007, 31, (46) MacFarlane, D. R.; Meakin, P.; Sun, J.; Amini, N.; Forsyth, M. 1429–1436. J. Phys. Chem. B 1999, 103, 4164–4170. (20) Rogers, R. D.; Voth, G. A. Acc. Chem. Res. 2007, 40, 1077–1078. (47) Forsyth, C. M.; MacFarlane, D. R.; Golding, J.; Huang, J.; Sun, J.; (21) Physicochemical properties: physicochemical properties of ionic Forsyth, M. Chem. Mater. 2002, 14, 2103. liquids: melting points and phase diagrams. Holbrey, J. D.; Rogers, (48) Golding, J.; Forsyth, S.; MacFarlane, D. R.; Forsyth, M.; Deacon, R. D. In Ionic Liquids in Synthesis, 2nd ed.; Wiley-VCH : New York, G. B. Green Chem 2002, 4, 223. 2008); Vol. 1, pp 57-72. (49) Dean, P. M.; Pringle, J. M.; MacFarlane, D. R. Both tosylate- (22) Krossing, I.; Slattery, J. M.; Daguenet, C.; Dyson, P. J.; Oleinikova, based ionic liquids were deposited at the Cambridge Crystallo- A.; Weingartner, H. J. Am. Chem. Soc. 2006, 128, 13427–13434. graphic Data Centre, the reference codes are CCDC 713020 & (23) MacFarlane, D. R.; Huang, J. H.; Forsyth, M. Nature 1999, 402, 713021, 2008. 792–794. (50) Holbrey, J. D.; Reichert, W. M.; Nieuwenhuyzen, M.; Johnston, S.; (24) Dean, P. M.; Pringle, J. M.; Forsyth, C. M.; Scott, J. L.; MacFarlane, Seddon, K. R.; Rogers, R. D. Chem. Commun. 2003, 1636–1637. D. R. New J. Chem. 2008, 32,2121–2126. (51) Pringle, J. M.; Golding, J.; Baranyai, K.; Forsyth, C. M.; (25) Smith, E. R. Proc. R. Soc., Ser. A 1981, 375, 475–505. Deacon, G. B.; Scott, J. L.; MacFarlane, D. R. New J. Chem. (26) Hoppe, R. Z. Anorg. Allg. Chem. 1956, 283, 196–206. 2003, 27, 1504. (27) Hoppe, R. Angew. Chem., Int. Ed. 1966, 5, 95–106. (52) Dean, P. M.; Pringle, J. M.; Forsyth, C. M.; Scott, J. L.; MacFar- (28) Harrison, W. A. Phys. Rev. B 2006, 73, 212103–212102. lane, D. R. New J. Chem. 2008, 32, 2121–2126. (29) Li, H.; Boatz, J. A.; Gordon, M. S. J. Am. Chem. Soc. 2008, 130, (53) Choudhury, A. R.; Winterton, N.; Steiner, A.; Cooper, A. I.; 392–393. Johnson, K. A. J. Am. Chem. Soc. 2005, 127, 16792–16793.