Received 16 July 2013
Polymorphs, hydrates and solvates of a co-crystal of caffeine with anthranilic acid
A polymorph screen on a new 1:1 co-crystal of caffeine, C8H10N4O2, with anthranilic acid, C7H7NO2, has revealed a rich diversity of crystal forms (two polymorphs, two hydrates and seven solvates, including two sets of isostructural solvates). These forms were prepared by liquid-assisted grinding and solution crystallization, and the crystal structures of nine of these forms have been solved using either single-crystal or powder X-ray data. The structures contain O-HN and N-HO hydrogen bonds through which caffeine and anthranilic acid molecules assemble to form zigzag-type chains. These chains can interact in an anti-parallel and offset manner to form cage- or channel-type skeletons within which solvent molecules can be located, giving rise to the diversity of forms observed for this co-crystal. In contrast, an equivalent series of liquid-assisted grinding and solution crystallization experiments with the closely related system of theobromine, C7H8N4O2, and anthranilic acid resulted in the formation of only one 1:1 co-crystal form.
Co-crystal formation has received significant attention due to its relevance in the design and construction of solid-state multi-component systems, particularly in the area of pharmaceutical materials science (Almarson & Zaworotko, 2004). The preparation of co-crystals was traditionally achieved by solution crystallization, but it has been demonstrated that mechanochemical methods (dry grinding and liquid-assisted grinding) can be faster, cleaner and more successful in finding new crystal forms (Trask et al., 2004).
Significant attention has been devoted to the formation of caffeine co-crystals as a means of preventing caffeine converting to caffeine hydrate under humid conditions (Trask et al., 2005). Caffeine (3,7-dihydro-1,3,7-trimethyl-1H-purine-2,6-dione) is known to be a central nervous system stimulant (Waldvogel, 2003) which exists in two anhydrous polymorphic forms (Enright et al., 2007) and a hydrate form (Pirttimäki & Laine, 1994). Since caffeine is very weakly basic in nature, it is more suited to co-crystallization than salt formation and there are several reports on caffeine co-crystals in the literature (Trask et al., 2005; Eddleston, Lloyd & Jones, 2012; Friscic et al., 2008; Bucar et al., 2009, 2010; Karki et al., 2007; Schultheiss et al., 2011; Aitipamula et al., 2012). Trask et al. (2005) have studied co-crystals of caffeine with dicarboxylic acids such as maleic acid, malonic acid, oxalic acid and glutaric acid prepared by co-crystallization from solution and by liquid-assisted grinding. Eddleston, Arhangelskis et al. (2012) described the formation and dissociation of a co-crystal comprised of caffeine and the closely related compound theophylline. Friscic et al. (2008) have investigated co-crystals of caffeine and succinic acid in the presence of appropriate guests which can form both C-HO hydrogen bonds and XN (where X = Cl, Br) halogen bonds. Bucar et al. (2009) have reported nine co-crystals of caffeine with different hydroxybenzoic acids prepared via a solution-mediated phase transformation screening technique. The same group has also reported a caffeine 2-hydroxy-1-naphthoic acid co-crystal obtained via a slurrying screening method (Bucar et al., 2010). Karki et al. (2007) have studied the formation of caffeine-citric acid co-crystals by liquid-assisted grinding and Schultheiss and co-workers (Schultheiss et al., 2011) have demonstrated that caffeine forms two co-crystals with p-coumaric acid, with 1:1 and 1:2 stoichiometric ratios. Recently, Aitipamula et al. (2012) have reported a co-crystal hydrate of caffeine and 4-hydroxybenzoic acid. In all of these previously reported carboxylic acid co-crystals of caffeine, the carboxylic acid group interacts with the imidazole moiety of caffeine via an O-HN hydrogen bond. Interestingly, while the Cambridge Structural Database (CSD; Allen, 2002) contains over 75 caffeine co-crystals, 25 of which are saturated (with 19 of these being hydrates), only four have more than one polymorph. Furthermore, none of the solvated co-crystals also has a reported anhydrous co-crystal form (for some co-formers both solvated and anhydrous forms exist, but only at different stoichiometries).
In this study, co-crystal formation involving caffeine (caf) and anthranilic acid (ana) is described. In addition, the results of a co-crystal polymorph and solvate screen on this co-crystal are presented. Co-crystal screening was carried out using liquid-assisted grinding (LAG; Shan et al., 2002), and a series of solvates, including isostructural solvates, was obtained. In total, 11 crystal forms of the caf:ana co-crystal (polymorphs, solvates and hydrates) were prepared and characterized. In contrast, an equivalent series of LAG experiments with theobromine (tbn), a compound with a molecular structure very similar to that of caffeine, and anthranilic acid demonstrated a significantly different outcome in terms of the number of forms observed, with just one co-crystal form being obtained. This is rationalized in terms of crystal packing.
All chemicals were purchased from Sigma-Aldrich and used without purification.
Grinding was carried out using a Retsch MM200 mixer mill at a frequency of 30 Hz. For each grinding experiment, 200 mg of material (117 mg, 0.604 mmol of caffeine and 83 mg, 0.604 mmol of anthranilic acid) was ground in the presence of liquid in a 15 ml steel grinding jar containing two small steel balls (7 mm diameter, 1.4 g).
Samples of co-crystal obtained by grinding were transferred to a 25 ml sample vial to which the appropriate solvent was added. The resulting suspension was heated until a clear solution was obtained. The clear solution was filtered into a fresh 25 ml vial and allowed to evaporate slowly under ambient conditions.
A slurrying experiment was performed in order to determine the thermodynamically stable form of the co-crystal at ambient temperature. A slurry of a 1:1 mixture of caf:ana forms (I) and (II) in heptane was stirred for 7 days under ambient conditions, and the solid was isolated by filtration prior to analysis.
Single-crystal X-ray data were collected at 180 K on a Nonius KappaCCD diffractometer equipped with an Oxford Cryosystems cooling device, using Mo K radiation ( = 0.7107 Å). Non-H atoms were refined with anisotropic displacement parameters, except for the solvent molecule in o-xylene solvate, which had to be refined isotropically due to disorder. H atoms were placed geometrically and were allowed to ride during the refinement. Crystal data, data collection and structure refinement details for the single-crystal X-ray data are summarized in Table 1. For the toluene and o-xylene monosolvates, the high R factors probably reflect unresolved disorder/twinning, discussed further in §3.2.2.
Powder X-ray diffraction (PXRD) data for structure solution of caf:ana forms (I) and (II) were recorded using beamline I11 at the Diamond Light Source, Didcot, UK. The beamline comprises a transmission geometry X-ray instrument with a wide range position-sensitive detector. All patterns were collected over the 2 range 1-90° at ambient temperature with a wavelength of 1.0338 Å and a detector zero angle offset of 0.008 (3)°.
Laboratory PXRD data were collected on a Phillips PW3710 diffractometer with Ni-filtered Cu K radiation ( = 1.5418 Å) at 40 kV and 40 mA using a scanning RTMS X'Celerator detector. Prior to recording PXRD patterns, powders were gently pressed onto a glass slide to give a flat surface. The data were collected over the 2 range 3-50° at ambient temperature with a step size of 0.0334°. For powder X-ray structure solutions, each PXRD pattern was collected over a time period of 1 h.
Simulated annealing structure solution and Rietveld refinement were carried out using TOPAS Academic (Coelho, 2007). Crystal data, data collection and structure refinement details for the powder X-ray data are summarized in Table 2, and Rietveld difference curves are shown in the supporting information.1 The coordinates of all atoms (including H) were allowed to refine, with a comprehensive set of tight restraints applied to all intramolecular bond distances and angles in order to control the molecular geometry. The caffeine and anthranilic acid molecules were also restrained to be planar. A common isotropic displacement parameter was refined for all non-H atoms in each structure, and the displacement parameter of the H atoms was constrained to be 1.2 times that value. For the toluene hemisolvate and tbn:ana co-crystal, the refined displacement parameter became negative, so it was finally constrained to Beq = 4.0 (Ueq = 0.05). Preferred orientation was not included for any of the structures. The refined structures were optimized using dispersion-corrected density functional theory (DFT-D) calculations, as outlined in the supporting information , and each structure was confirmed to be a valid energy minimum.
Samples were prepared by placing material in a standard 40 µL Al pan (2-7 mg) onto which an Al lid was crimped using a press. A small hole was made in the lid if solvent was known to be present in the sample. Analysis was performed using a Mettler Toledo DSC822 instrument with an 80 cm3 min-1 N2 purge flow rate.
Samples were prepared by placing 7-10 mg of material in a standard 100 µL pan. Analysis was performed using a Mettler Toledo TGA/SDTA851e/SF/1100 instrument with an air purge flow rate of 50-70 cm3 min-1. The samples were heated from ambient temperature to 573 K (ramp 10 K min-1).
The caf:ana co-crystal form (I) was initially obtained by liquid-assisted grinding of caffeine and anthranilic acid with 50 µL of nitromethane (or chloroform, diethylether, hexane, tetrahydrofuran, ethanol). All attempts to obtain single crystals were, however, unsuccessful and the structure of the co-crystal was therefore solved using synchrotron PXRD data (Table 2). The melting point of the co-crystal was determined by DSC to be 373 K (supporting information ), a value significantly different from those of caf and ana. In addition, no weight loss was observed on heating to the melting of the co-crystal during thermogravimetric analysis, confirming that the co-crystal did not contain solvent.
Form (I) crystallizes in space group with two molecules of caffeine and two molecules of anthranilic acid in the asymmetric unit (Z' = 2). The caffeine and anthranilic acid molecules interact through O-H(carboxyl)N(imidazole) and N-H(amine)O=C(carbonyl) hydrogen bonds to form zigzag-type molecular chains (Fig. 1). The amine group of anthranilic acid is involved in an intramolecular N-HO=C hydrogen bond with the carboxyl group.
| || Figure 1 |
(a) Hydrogen-bonding arrangement and (b) crystal packing in caf:ana form (I).
Form (II) was initially obtained by liquid-assisted grinding of caffeine and anthranilic acid with 50 µL of acetonitrile. To date, this is the only liquid that by LAG has given rise to form (II). Again, due to the microcrystalline nature of the sample, and the inability to grow suitable single crystals, the crystal structure was solved from synchrotron powder X-ray data (Table 2). The melting point of the co-crystal was 376 K (supporting information ), higher than that of form (I). In addition, no weight loss was observed on heating to the melting point of the co-crystal during thermogravimetric analysis, confirming that the co-crystal did not contain solvent.
In form (II) the caffeine and anthranilic acid molecules interact via O-H(carboxyl)N(imidazole) and N-H(amine)O=C(carbonyl) hydrogen bonds (Fig. 2). Furthermore, the amine group of ana is involved in an intramolecular N-HO=C hydrogen bond with the carboxyl group. Twisted zigzag-type caffeine anthranilic acid chains with V-shape cross sections are observed. These chains stack to form layers and the hydrogen-bonding directions in adjacent layers are perpendicular.
| || Figure 2 |
(a) Hydrogen-bonding arrangement and (b) crystal packing in caf:ana form (II).
The 13C solid-state NMR spectra of caf:ana forms (I) and (II) (supporting information ) were recorded to assist in the structure solution from PXRD data, in particular to investigate the number of molecules in the crystallographic asymmetric unit. The spectra show lines for caffeine and anthranilic acid which have moved upfield or downfield (depending upon the particular resonance) compared with the starting materials due to the changes in the short-range aggregation and shielding/deshielding after co-crystal formation. The spectra of the two polymorphs differ, and in particular the spectrum of form (I) shows a doubling of lines compared with form (II), suggesting that form (I) is a Z' = 2 phase. This can be rationalized by considering that the parallel nature of the zigzag chains of form (I) means that they can only be related by translation, whereas in form (II) the chains can be related by a herringbone motif as they are approximately orthogonal.
A slurrying experiment where equal amounts of forms (I) and (II) were stirred in heptane resulted in complete conversion to form (I), suggesting that form (I) is the thermodynamically stable form at room temperature. The synchrotron PXRD patterns of forms (I) and (II) are shown in Fig. 3.
| || Figure 3 |
Synchrotron PXRD patterns of caf:ana (a) form (I) and (b) form (II) ( = 1.0338 Å).
During LAG experiments performed to screen for polymorphs and solvates of the caf:ana co-crystal, solvents such as chlorobenzene, bromobenzene, toluene, o-xylene, p-xylene and water yielded solvates. Attempts were made to grow single crystals by slow evaporation of solutions of caf and ana in the appropriate solvents. Each of the solvates was characterized by PXRD, DSC and TGA, and crystal structures were solved using either single-crystal X-ray or powder X-ray diffraction data.
The toluene, chlorobenzene and bromobenzene hemisolvates form an isostructural group, crystallizing in space group with two molecules of caffeine, two molecules of anthranilic acid and one solvent molecule in the asymmetric unit. Caffeine and anthranilic acid molecules interact through O-H(carboxyl)N(imidazole) and N-H(amine)O=C(carbonyl) hydrogen bonds. The amine group of anthranilic acid is involved in an intramolecular N-HO=C hydrogen bond with the carboxyl group. Zigzag-type caffeine-anthranilic acid chains pack to give a structure containing cavities. The solvent molecules are located in the cavities as shown in Figs. 4 and 5.
| || Figure 4 |
Crystal packing of the caf:ana chlorobenzene hemisolvate.
| || Figure 5 |
Crystal packing of the caf:ana toluene hemisolvate
The crystal structure of the chlorobenzene hemisolvate was solved using single-crystal X-ray data, and is of good quality (Table 1). For the toluene and bromobenzene hemisolvates, it was not possible to obtain crystals suitable for single-crystal analysis and their crystal structures were solved from laboratory powder X-ray data (Table 2). These structures (in particular the toluene solvate, for which only relatively poor data could be obtained) are of lower precision, but the isostructurality with the chlorobenzene solvate is clearly established. To date, equivalent hemi-solvates containing o-xylene or p-xylene have not been obtained.
Toluene and o-xylene monosolvates were initially obtained by LAG of caffeine and anthranilic acid with 100 µL of liquid. Subsequently, crystals were obtained for single-crystal X-ray structure determination. The toluene and o-xylene monosolvates crystallize in space groups and P21/n, respectively, but their structures are very similar, indicating that these two co-crystal solvates are essentially isostructural. The toluene monosolvate contains two molecules of caffeine, two molecules of anthranilic acid and two molecules of solvent in the asymmetric unit (Z' = 2), while the o-xylene monosolvate contains one of each molecule type in the asymmetric unit (Z' = 1). Caffeine and anthranilic acid molecules again interact through O-H(carboxyl)N(imidazole) and N-H(amine)O=C(carbonyl) hydrogen bonds. The amine group of anthranilic acid is involved in an intramolecular N-HO=C hydrogen bond with the carboxyl group as depicted in Fig. 6. The caffeine and anthranilic acid molecules assemble to form zigzag-type molecular chains which interact in an anti-parallel and offset manner via van der Waals forces to form channels within which solvent molecules are located. The reduction of the symmetry from P21/n for the o-xylene solvate to for the toluene solvate results from the orientation of the solvent molecules within the channels. In the former, the o-xylene molecules appear to be related by a 21 screw axis running parallel to the channel, while this symmetry is not present in the toluene solvate. The high R factors for both refinements (Table 1) probably reflect disorder and/or twinning, which has not been resolved here. It was not possible to prepare a chlorobenzene monosolvate, although it has yet to be determined whether this form is unfavourable or simply slower to nucleate than the obtained hemisolvate.
| || Figure 6 |
Crystal packing of (a) the caf:ana toluene monosolvate and (b) the caf:ana o-xylene monosolvate.
A p-xylene solvate of the caf:ana co-crystal was obtained by slow evaporation of a solution of caffeine and anthranilic acid in p-xylene. The solid crystallizes in space group with one molecule of caffeine, one molecule of anthranilic acid and one and a half molecules of p-xylene in the asymmetric unit (giving a caf:ana:p-xylene ratio of 1:1:1.5). Caffeine and anthranilic acid molecules again interact via O-H(carboxyl)N(imidazole) and N-H(amine)O=C(carbonyl) hydrogen bonds. The amine group of anthranilic acid is involved in an intramolecular N-HO=C hydrogen bond with the carboxyl group. Zigzag-type caffeine anthranilic acid chains stack, leading to the formation of nano-size channels, as in the toluene and o-xylene monosolvates, within which p-xylene molecules are located (Fig. 7). A second p-xylene solvate was obtained by liquid-assisted grinding of caffeine and anthranilic acid with 50 µL of p-xylene. The PXRD pattern of this crystal form did not match that of the p-xylene sesquisolvate, nor those of the isostructural sets of hemi- and monosolvates described above. To date, this structure has not been established.
| || Figure 7 |
Crystal packing of the caf:ana p-xylene sesquisolvate.
A caf:ana monohydrate was initially obtained by exposing form (II) to 98% relative humidity. Subsequently, single crystals were obtained by slow evaporation of a solution of caffeine and anthranilic acid in dichloromethane (undried). The compound crystallizes in space group P21/c with two molecules of caffeine, anthranilic acid and water in the asymmetric unit (Z' = 2). There are two types of water molecules, which are cross-linked by caffeine molecules and anthranilic acid molecules via O-H(water)O(amide in caf) and N-H(amine in ana)O(water) hydrogen bonds. Furthermore, the water molecules are linked via O-HO hydrogen bonding. The packing diagram for the monohydrate is shown in Fig. 8.
| || Figure 8 |
Crystal packing of the caf:ana monohydrate.
A second hydrate, denoted caf:ana hydrate (II), was initially obtained by liquid-assisted grinding of caffeine and anthranilic acid with 50 µL of water. Attempts to obtain single crystals for structure determination were unsuccessful, however, and the formation of the new co-crystal phase was shown only by the comparison of PXRD traces. The PXRD patterns of the caf:ana solvates are given in Fig. 9.
| || Figure 9 |
PXRD patterns (Cu K radiation) of caf:ana: (a) chlorobenzene hemisolvate, (b) toluene hemisolvate, (c) toluene monosolvate, (d) o-xylene monosolvate, (e) p-xylene sesquisolvate, (f) p-xylene solvate, (g) monohydrate, (h) hydrate-(II).
For comparative purposes, theobromine (tbn), a compound with a similar molecular structure to caffeine, was co-crystallized with anthranilic acid. A set of LAG experiments were conducted with the same set of solvents as used in the caf:ana work described above. Interestingly, all LAG experiments yielded the same tbn:ana acid co-crystal form. In the crystal structure, the theobromine and anthranilic acid molecules form hydrogen-bonded chains via O-H(carboxyl)N(imidazole) and N-H(amine)O=C(carbonyl) interactions (Table 2). Each pair of chains is further interconnected through hydrogen bonds between theobromine molecules to give a one-dimensional tape motif. The tapes stack to give sheets, as shown in Fig. 10.
| || Figure 10 |
(a) Hydrogen-bonding arrangement and (b) crystal packing in the tbn:ana co-crystal.
Eleven new multi-component crystal forms of a caffeine:anthranilic acid co-crystal were prepared by liquid-assisted grinding, and the crystal structures of nine of these species were solved from either single-crystal or powder X-ray data. The crystal structures contain O-H(carboxyl)N(imidazole) and N-H(amine)O=C (carbonyl) hydrogen bonds. These robust interactions give zigzag-type molecular chains in all of the isolated crystal forms, with the exception of caf:ana monohydrate.
The structural differences in the two polymorphs of the co-crystal [forms (I) and (II)] mainly lie in the orientation of the caf:ana zigzag molecular chains. Form (I) contains a planar type of chain while form (II) contains twisted, non-planar molecular chains. It has been suggested that co-crystal polymorphism is a less common phenomenon than polymorphism in single-component phases. However, there are now several reported examples of co-crystal polymorphs. Ueto et al. (2012) have reported four polymorphs of a 1:1 furosemide:nicotinamide co-crystal. Aitipamula et al. (2009) have described three polymorphs of the ethenzamide:gentisic acid co-crystal, and three polymorphs were also reported for the 1:1 phenazine:mesaconic acid co-crystal and the 1:1 L-malic acid:L-tartaric acid co-crystal (Eddleston et al., 2013; Eddleston, Arhangelskis et al., 2012). In addition, several reported co-crystal systems exhibit two co-crystal polymorphs (Skovsgaard & Bond, 2009; Aitipamula et al., 2010; Porter et al., 2008; Seefeldt et al., 2007; Childs et al., 2009; Mukherjee & Desiraju, 2011).
Solvates and hydrates also provide a means of increasing the number of solid forms for a given co-crystal system. The caf:ana crystal system described in this study is unique for co-crystals isolated to date in that it shows polymorphic behaviour in addition to a strong propensity for solvate formation, including two types of hydrates. The key feature of this co-crystal that might explain the large number of crystal forms may be the zigzag-type chain arrangement of caffeine and anthranilic acid molecules. The chains can interact in an anti-parallel and offset manner, which allows the formation of either cage- or channel-type skeletons, allowing solvent molecules to be incorporated into the structure. The tbn:ana co-crystal is clearly less prone to forming multiple crystal forms. This is thought to be due to the absence of zigzag-type chains in this structure and that all hydrogen-bonding sites are occupied.
The existence of isostructurality has been observed in the caf:ana solvates. The toluene monosolvate is essentially isostructural with the o-xylene monosolvate and the toluene hemisolvate is isostructural with the chlorobenzene and bromobenzene hemisolvates. We note that several examples of isostructural co-crystal systems involving host lattice and guest molecules of different size and shape have been reported. Cincic et al. (2008) have reported seven isostructural halogen bonded co-crystals involving six different molecules. Caira (2007) has observed isostructurality in the crystals of sulfametrole: tetroxoprim salts, hydrates and solvates. In addition, Galcera et al. (2012) have recently reported four isostructural salts of lamotrigine.
In conclusion, the co-crystal system of caffeine and anthranilic acid is crystallographically rich, showing polymorphs, solvates, hydrates and isostructural solvates. This is in contrast to the case of theobromine and anthranilic acid, for which only a single co-crystal form and no co-crystal solvates or hydrates have been identified to date.
Dr John E. Davies is acknowledged for the single-crystal X-ray structures and we are thankful to Adam Brewer for synchrotron PXRD data collection. We thank Dr Jacco van de Streek (Department of Pharmacy, University of Copenhagen) for help with Rietveld refinements. Dr David G. Reid is acknowledged for recording 13C solid-state NMR spectra, and we thank Dr Ranjit Thakuria for assistance with preparing figures. We are grateful to the Cambridge Commonwealth Trust, the EU INTERREG IVA 2 Mers Seas Zeeën Cross-border Cooperation Programme and the EPSRC for financial support.
Aitipamula, S., Chow, P. S. & Tan, R. B. H. (2009). CrystEngComm, 11, 1823-1827.
Aitipamula, S., Chow, P. S. & Tan, R. B. H. (2010). CrystEngComm, 12, 3691-3697.
Aitipamula, S., Chow, P. S. & Tan, R. B. H. (2012). CrystEngComm, 14, 2381-2385.
Allen, F. H. (2002). Acta Cryst. B58, 380-388.
Almarson, O. & Zaworotko, M. J. (2004). Chem. Commun. pp. 1889-1896.
Altomare, A., Cascarano, G., Giacovazzo, C., Guagliardi, A., Burla, M. C., Polidori, G. & Camalli, M. (1994). J. Appl. Cryst. 27, 435.
Blessing, R. H. (1995). Acta Cryst. A51, 33-38.
Boultif, A. & Louër, D. (2004). J. Appl. Cryst. 37, 724-731.
Bucar, D.-K., Henry, R. F., Duerst, R. W., Lou, X., MacGillivray, L. R. & Zhang, G. G. Z. (2010). J. Chem. Cryst. 40, 933-939.
Bucar, D.-K., Henry, R. F., Lou, X., Duerst, R. W., MacGillivray, L. R. & Zhang, G. G. Z. (2009). Cryst. Growth Des. 9, 1932-1943.
Caira, M. R. (2007). Mol. Pharm. 4, 310-316.
Childs, S. L., Wood, P. A., Rodríguez-Hornedo, N., Reddy, L. S. & Hardcastle, K. I. (2009). Cryst. Growth Des. 9, 1869-1888.
Cincic, D., Friscic, T. & Jones, W. (2008). Chem. Eur. J. 14, 747-753.
Coelho, A. A. (2007). TOPAS-Academic, Version 4.1. Coelho Software, Brisbane, Australia.
David, W. I. F., Shankland, K., van de Streek, J., Pidcock, E., Motherwell, W. D. S. & Cole, J. C. (2006). J. Appl. Cryst. 39, 910-915.
Eddleston, M. D., Arhangelskis, M., Friscic, T. & Jones, W. (2012). Chem. Commun. 48, 11340-11342.
Eddleston, M. D., Lloyd, G. O. & Jones, W. (2012). Chem. Commun. 48, 8075-8077.
Eddleston, M. D., Sivachelvam, S. & Jones, W. (2013). CrystEngComm, 15, 175-181.
Enright, G. D., Terskikh, V. V., Brouwer, D. H. & Ripmeester, J. A. (2007). Cryst. Growth Des. 7, 1406-1410.
Friscic, T., Trask, A. V., Motherwell, W. D. S. & Jones, W. (2008). Cryst. Growth Des. 8, 1605-1609.
Galcera, J., Friscic, T., Hejczyk, K. E., Fábián, L., Clarke, S. M., Day, G. M., Molins, E. & Jones, W. (2012). CrystEngComm, 14, 7898-7906.
Karki, S., Friscic, T., Jones, W. & Motherwell, W. D. S. (2007). Mol. Pharm. 4, 347-354.
Mukherjee, A. & Desiraju, G. R. (2011). Chem. Commun. 47, 4090-4092.
Nonius (1998). COLLECT. Nonius BV, Delft, The Netherlands.
Otwinowski, Z. & Minor, W. (1997). Methods in Enzymology, Vol. 276, Macromolecular Crystallography, Part A, edited by C. W. Carter Jr & R. M. Sweet, pp. 307-326. New York: Academic Press.
Panalytical (2010). X'Pert Data Collector. Panalytical BV, Almelo, The Netherlands.
Pirttimäki, J. & Laine, E. (1994). Eur. J. Phar. Sci. 1, 203-208.
Porter, W. W., Elie, S. C. & Matzger, A. J. (2008). Cryst. Growth Des. 8, 14-16.
Schultheiss, N., Roe, M. & Boerrigter, S. X. M. (2011). CrystEngComm, 13, 611-619.
Seefeldt, K., Miller, J., Alvarez-Núñez, F. & Rodríguez-Hornedo, N. (2007). J. Pharm. Sci. 96, 1147-1158.
Shan, N., Toda, F. & Jones, W. (2002). Chem. Commun. pp. 2372-2373.
Sheldrick, G. M. (2008). Acta Cryst. A64, 112-122.
Skovsgaard, S. & Bond, A. D. (2009). CrystEngComm, 11, 444-453.
Trask, A. V., Motherwell, W. D. S. & Jones, W. (2004). Chem. Commun. pp. 890-891.
Trask, A. V., Motherwell, W. D. S. & Jones, W. (2005). Cryst. Growth Des. 5, 1013-1021.
Ueto, T., Takata, N., Muroyama, N., Nedu, A., Sasaki, A., Tanida, S. & Terada, K. (2012). Cryst. Growth Des. 12, 485-494.
Waldvogel, S. R. (2003). Angew. Chem. Int. Ed. 42, 604-605.