Received 4 September 2013
Solvent-guest control of two extremely similar tetrahydrofuran inclusion structures
Racemic 2,4,6,8-tetracarbomethoxybicyclo[3.3.0]octa-2,6-diene-3,7-diol, C16H18O10 (1), was known previously to yield two solvent-free polymorphs and also a clathrate inclusion crystal form. Crystallization of (1) yields two inclusion compounds containing tetrahydrofuran (THF): (1)4·THF is obtained from a mixture of THF and methanol, whereas (1)2·THF is obtained from pure THF. The X-ray crystal structures reveal that the two compounds are extremely similar and that their host arrangements are essentially identical. They differ, however, in the proportion, orientation and host-guest interaction of the included THF molecules. The disordered guest molecules in (1)4·THF are oriented along the guest channel direction, whereas in (1)2·THF they lie across the channel. This unusual solvent-guest control of inclusion structures has implications relating to the formation of polymorphic structures and other competing crystal forms.
The question of which substances can crystallize in multiple crystal forms, and what their structures are, is a matter of considerable importance for both academic and commercial reasons (Bernstein, 2002, 2011). The racemic form of 2,4,6,8-tetracarbomethoxybicyclo[3.3.0]octa-2,6-diene-3,7-diol (1) is obtained easily by condensation of one equivalent of glyoxal and two of dimethyl 3-ketoglutarate (Bertz et al., 1990). Crystallization of racemic (1) from acetone yielded solvent-free crystals in space group P21/n (Vega et al., 2008). In these crystals the dish-shaped tetraester has assembled into a centrosymmetric dimer by utilizing two identical C-HO-C weak hydrogen bonds. Repetition of these dimeric units, linked by C-HO (Desiraju, 1996) and orthogonal C=OC=O (Allen et al., 1998) interactions, produces the complete crystal structure.
We considered that tetraester (1) was a prime candidate for adopting more than one crystal form, based on the following reasoning: (i) the only intermolecular interactions of significance in the P21/n structure were weaker than traditional hydrogen bonds (Desiraju & Steiner, 1999), so alternative arrangements might therefore be possible; (ii) the concave face-to-concave face dimeric association of dishes produced a repeat unit that was an inefficient shape for crystal packing. Such steric characteristics are known to facilitate inclusion of guest molecules in many instances (Kitaigorodsky, 1973; Dastidar & Goldberg, 1996).
Consequently, we conducted a detailed crystallization screening of the tetraester (1) using a wide variety of solvents (Gao et al., 2012). This demonstrated that, at laboratory temperature and pressure, around 40% of solvents gave the known solvent-free structure [subsequently termed polymorph (I)]. Crystallization from methanol at 273 K, however, produced a new solvent-free structure in space group [polymorph (II)]. In addition, the remaining 60% of solvents tested under laboratory temperature and pressure formed an essentially isostructural family [inclusion form (I)] of clathrate inclusion compounds in space group P21/c (Gao et al., 2012). Encouraged by these findings we have carried out more detailed studies on the crystallization behaviour of (1), and we report here on its unusual relationship with the guest molecule tetrahydrofuran (THF).
Suitable single crystals were selected under a polarizing microscope (Leica M165Z) and were picked up on a MicroMount (MiTeGen, USA) consisting of a thin polymer tip with a wicking aperture. The crystals were coated with paraffin oil and then quickly transferred to the cold stream from an Oxford Cryosystems Cryostream attachment. X-ray diffraction measurements were carried out on a Bruker Kappa APEX-II CCD diffractometer at 150 K using graphite-monochromated Mo K radiation. Experimental details are given in Table 1.
Although clearly located in different orientations in the two structures, the guest solvent THF molecules exhibit a high degree of disorder. In (1)4·THF, where the crystals were obtained by slow evaporation from a mixture of THF and methanol, the THF guest is disordered with the carbon C3T (Fig. 1) statistically distributed across the centre of symmetry. Thus, the two positions of THF, with occupancy of 0.25 each, are located at the centre of the columns created by the host framework. The difference Fourier peaks indicated extensive orientational disorder of the THF molecule, keeping its centre of mass almost in the same position (see the supporting information1 for diagrams). Orientations with lower occupancy were not included in the final refinement. The THF molecule in (1)2·THF, crystallized from pure THF, shows a different location in the channel, that is quite offset from the centre of inversion. The site occupancy of this position is assigned as 0.5, and again here there are some peaks in the difference Fourier map suggesting other orientational disorder with lower occupancy at this position (see the supporting information ). In this structure, the envelope-head O atom is modelled over two sites of equal occupancy (0.25 each) with only one of the positions making an O-HO contact with the host molecule. In both structures, the geometry and anisotropic displacement parameters of the THF molecule were restrained.
| || Figure 1 |
Molecular structures of (a) (1)4·THF and (b) (1)2·THF, showing displacement ellipsoids at 50% probability for non-H atoms. In (1)4·THF, all atoms of the THF molecule have site occupancy 0.25. In (1)2·THF, all C atoms of the THF molecule have site occupancy 0.50, while atoms O1T and O1T' have occupancy 0.25.
The resulting structures show clearly the difference between the orientations of the THF molecules (and thereby the different identities of the two compounds), but the crystallographic R factors remain relatively high (Table 1). An alternative strategy to refine these structures may be to employ the SQUEEZE solvent-area model (van de Sluis & Spek, 1990) in the PLATON program (Spek, 2009). The application of SQUEEZE to the current structures with the THF guest molecules eliminated indicates a void space of 406.7 Å3 per unit cell for (1)4·THF, occupied by 43 electrons, and 483.7 Å3 per unit cell for (1)2·THF, occupied by 75 electrons. Thus, the indicative electron counts support the stated THF content of one molecule per unit cell for (1)4·THF and two molecules per unit cell for (1)2·THF (one THF molecule = 40 electrons). Refinement against the data produced by SQUEEZE gives R1 = 0.047, wR2 = 0.160 for (1)4·THF, and R1 = 0.060, wR2 = 0.211 for (1)2·THF. This confirms that the high R factors for the conventional refinements are associated with difficulties modelling the THF guest region.
Crystallization of the tetraester (1) was carried out from a mixture (1:1 by volume) of tetrahydrofuran and methanol, and also from pure tetrahydrofuran, at room temperature and pressure. Different materials were obtained from these two experiments and their structures solved using single-crystal X-ray methods. Details of the solution and refinement are shown in Table 1, and labelled ellipsoid plots are shown in Fig. 1.
Crystallization of the racemic tetraester (1) from tetrahydrofuran and methanol (volume 1:1) yielded crystals of (1)4·THF in space group P21/c. This is a further example of the inclusion form (I) family that we reported earlier (Gao et al., 2012). In this structure, homochiral molecules of (1) assemble around a 21 screw axis running along the b direction. These helices pack parallel to each other and produce host layers of the same chirality in the ab plane. Layers alternate in their handedness along c, and parallel channels are situated between neighbouring helices along b. These guest-containing channels have an approximately rectangular cross-section (ca 5.0 × 6.1 Å) and the size of potential guest molecules is important. The p-xylene guest is completely ordered in the structure reported previously, but the smaller THF guests are disordered over two equivalent sites surrounding an inversion centre (Gao et al., 2012). The THF molecules in (1)4·THF are oriented along the guest channels (Fig. 2). Crystallization of (1) from methanol had been found in our earlier work to yield the solvent-free polymorph (I) compound, and there was also no evidence of methanol inclusion in this new material.
| || Figure 2 |
Part of the structure of (1)4·THF from tetrahydrofuran and methanol, showing the THF guest molecules disordered over two equivalent sites surrounding an inversion centre: (a) projection on the ac plane; (b) projection with the b direction vertical. Opposite host enantiomers are shown in light or dark green, and all H atoms have been omitted for clarity.
Crystallization of the tetraester (1) from pure tetrahydrofuran yielded different crystals but also belonging to the inclusion form (I) family. This substance had the same space group P21/c, but the different composition (1)2·THF. The guest molecules are again disordered over two positions surrounding an inversion centre. However, they have a different orientation rotated by about 45° from the first and they now lie across the channel as illustrated in Fig. 3. Further, the guest sites are close to each other in (1)4·THF but are much more separated in the structure of (1)2·THF.
| || Figure 3 |
Part of the structure of (1)2·THF obtained from pure tetrahydrofuran and showing the THF guests disordered over two sites surrounding an inversion centre. The THF guest molecules now lie across the channel. Comparison should be made with Fig. 2.
The two THF inclusion compounds obtained have the same space group and both belong to the inclusion form (I) family. Their crystal structures are extremely similar and, indeed, the host arrangements are almost identical. Our previous work on (1) had revealed a significantly different conformation involving one ester group in the solvent-free polymorph (I) compared with the inclusion form (I) crystal structure (Gao et al., 2012). Fig. 4 compares the tetraester conformations present here, and these are almost identical in the two new P21/c materials.
| || Figure 4 |
Comparison of the host conformations in the two P21/c inclusion structures. Colour code: (1)4·THF from THF and methanol (purple), and (1)2·THF from pure THF (orange).
The differences in these crystal structures relate to the THF guests. In both cases, they are arranged around an inversion centre, but the occupancy of the sites differs (see §2). This results in the differing compositions of (1)4·THF from the THF/methanol solvent mixture, and (1)2·THF from pure THF. The additional solvent inclusion in the latter case results in small increases in the a, b and c dimensions (Table 1). Consequently, the unit-cell volume increases by 4.3%. Despite this, the calculated density remains almost unchanged (1.288 against 1.291 g cm-3) due the concomitant increase in guest inclusion.
The THF molecules also are rotationally disordered around their centre of mass and this complicates analysis of their host-guest interactions. The best apparent hydrogen-bonding geometry is shown for each case in Fig. 5, with the geometrical parameters listed in Table 2. Compound (1)4·THF employs a host enol -O-HO guest interaction and a host H-C guest interaction (Fig. 5a). In contrast, (1)2·THF employs a more complex arrangement comprising a host enol -O-HO guest interaction and three host OH-C guest interactions (Fig. 5b).
| || Figure 5 |
Host-guest interactions (blue lines) present for the most favoured rotational disorder component of the THF guest: (a) (1)4·THF and (b) (1)2·THF. The intramolecular host O-HO interactions are also indicated in blue.
The formation of these alternative crystals occurs under identical temperature and pressure conditions and results in extremely similar, but subtly different, structures. It is hard to provide a definitive explanation for this unusual solvent-guest behaviour. However, different intermolecular forces are present in the two experiments, namely methanol CH3-O-HO THF stronger hydrogen bonds in the mixed solvent case, and THF C-HO THF weaker hydrogen bonds in pure tetrahydrofuran (Fig. 6). These different solvation environments can clearly result in different kinetic crystallization pathways being employed.
| || Figure 6 |
Different intermolecular interactions present in the two solvent systems: (a) methanol-THF stronger hydrogen bond; (b) THF-THF weaker hydrogen bond.
Comparison of the present structures with the earlier p-xylene inclusion compound (Gao et al., 2012) is informative. The guest molecules were completely ordered in the p-xylene material and were orientated in a very similar manner (Fig. 7) to those in (1)4·THF. Its composition, however, was (1)2·(p-xylene) since the larger guest was able to bridge the guest channel completely. It is therefore clear that several different guest stoichiometries and orientations are possible across the inclusion form (I) series of compounds.
| || Figure 7 |
The guest arrangement present in the ordered inclusion compound (1)2·(p-xylene). These diagrams should be compared with Figs. 2 and 3.
Alternative inclusion structures produced from the same host and guest molecules are encountered occasionally, but their crystal structures usually are markedly different (Ung et al., 1993; Skobridis et al., 2011). The remarkable similarity of the two THF structures reported here is unprecedented. These new results reveal a weakness in the screening protocol often used to identify different crystal forms. A common procedure is to measure the unit-cell parameters, identify the space group and then see if the single crystal is the same or different from earlier samples. Full structure determination is only completed for new polymorphs or examples of inclusion. In this instance, two different crystals were obtained in the same space group P21/c. Although their cell parameters differed slightly, it would have been easy to miss this and conclude that both materials were identical. It is improbable that this unexpected behaviour is an isolated case and examples of more closely related structures could be even more difficult to identify.
Our work on the tetraester (1) has shown that, at least occasionally, it is possible to identify candidate molecules that are prone to producing multiple crystal forms (Gao et al., 2012). Although the advance towards the prediction of crystal structures using computation is making great progress (Neumann et al., 2008; Price, 2009; Cruz-Cabeza et al., 2010), the complication of alternative isolable crystal forms remains problematical. Indeed, it has been remarked that crystal landscapes often predict many more polymorphs of favourable energy than are obtained experimentally (Price, 2013). This is likely also to be the case for solvates, hydrates and mutually hydrogen-bonded co-crystals. While there are many factors contributing to this disparity, one of the major parameters is the precise role of the solvent during the crystallization process. Whether this is ultimately excluded to give a solvent-free crystal, or retained to yield an inclusion material, it plays an intimate role that generally remains cryptic. Perhaps, in the future, we should pay greater attention to the Polish saying that `a guest sees more in an hour than the host in a year'. The present report demonstrates that the solvent system can be critical in controlling crystal outcomes. Greater attention should be paid to employing mixtures of solvents, and also solvents containing small amounts of selected additives.
Allen, F. H., Baalham, C. A., Lommerse, J. P. M. & Raithby, P. R. (1998). Acta Cryst. B54, 320-329.
Bernstein, J. (2002). Polymorphism in Molecular Crystals. Oxford University Press.
Bernstein, J. (2011). Cryst. Growth Des. 11, 632-650.
Bertz, S. H., Cook, J. M., Gawish, A. & Weiss, U. (1990). Org. Synth. Collect. VII, 50-56.
Bruker (2001). SADABS. Bruker AXS Inc., Madison, Wisconsin, USA.
Bruker (2007). APEX2 and SAINT. Bruker AXS Inc., Madison, Wisconsin, USA.
Cruz-Cabeza, A. J., Karki, S., Fábián, L., Friscic, T., Day, G. M. & Jones, W. (2010). Chem. Commun. 46, 2224-2226.
Dastidar, P. & Goldberg, I. (1996). Comprehensive Supramolecular Chemistry, edited by D. D. MacNicol, F. Toda & R. Bishop, Vol. 6, Ch. 10, pp. 305-350. Oxford: Pergamon Press.
Desiraju, G. R. (1996). Acc. Chem. Res. 29, 441-449.
Desiraju, G. R. & Steiner, T. (1999). The Weak Hydrogen Bond in Structural Chemistry and Biology. Oxford University Press.
Gao, J., Bhadbhade, M. M. & Bishop, R. (2012). CrystEngComm, 14, 138-146.
Kitaigorodsky, A. I. (1973). Molecular Crystals and Molecules. New York: Academic Press.
Macrae, C. F., Bruno, I. J., Chisholm, J. A., Edgington, P. R., McCabe, P., Pidcock, E., Rodriguez-Monge, L., Taylor, R., van de Streek, J. & Wood, P. A. (2008). J. Appl. Cryst. 41, 466-470.
Neumann, M. A., Leusen, F. J. J. & Kendrick, J. (2008). Angew. Chem. Int. Ed. 47, 2427-2430.
Price, S. L. (2009). Acc. Chem. Res. 42, 117-126.
Price, S. L. (2013). Acta Cryst. B69, 313-328.
Sheldrick, G. M. (2008). Acta Cryst. A64, 112-122.
Skobridis, K., Paraskevopoulos, G., Theodorou, V., Seichter, W. & Weber, E. (2011). Cryst. Growth Des. 11, 5275-5288.
Sluis, P. van der & Spek, A. L. (1990). Acta Cryst. A46, 194-201.
Spek, A. L. (2009). Acta Cryst. D65, 148-155.
Ung, A. T., Bishop, R., Craig, D. C., Dance, I. G. & Scudder, M. L. (1993). Tetrahedron, 49, 639-648.
Vega, A., Donoso-Tauda, O., Ibañez, A. & Escobar, C. A. (2008). Acta Cryst. C64, o199-o204.