PREDICTING POLYMORPHIC CRYSTAL STRUCTURES

Olaf König, Mark C. Wahle, Gerhard E. Engel, Frank J. J. Leusen

Molecular Simulations Ltd., 230/250 The Quorum, Barnwell Road, Cambridge CB5 8RE, United Kingdom, E-mail: okoenig@msicam.co.uk

Keywords: polymorphism, crystal structure prediction, crystal engineering

Polymorphism can be a source of many problems in multiple industries, including pharmaceuticals, pigments, nonlinear optics and explosives. In crystallization processes, temperature, concentration, solvent, and impurities all play a role in the formation of multiple polymorphic forms. Different forms can possess different physicochemical properties: for example bioavailability, stability, color, density, morphology, and mechanical behavior. To fully understand these property differences, it is imperative to have detailed knowledge of all possible polymorphs of a given compound. Furthermore, rational control of solid-state properties and crystal engineering require knowledge of the crystal structure [1].

In this paper we will discuss the potential of a crystal structure prediction method implemented in the commercially available Cerius² molecular modeling package (Molecular Simulations) [2]. The approach, which has been improved significantly over the last year, is based on the generation of possible packing arrangements in all reasonable space groups to locate the lower minima in the lattice energy. In a first step a Monte Carlo simulated annealing procedure is applied to search the lattice energy hypersurface for low-energy crystal structures. In this search procedure, which typically delivers thousands of possible crystal structures, molecules are treated as rigid bodies and the search is repeated for each relevant space group. In a second step a clustering method is applied to eliminate similar crystal structures. Each unique structure is subsequently minimized with respect to all degrees of freedom (including molecular flexibility). Finally, each structure is ranked according to calculated lattice energy. The resulting low-energy structures are potential polymorphs; calculated powder patterns of these structures can be compared to experimental powder data for verification.

The crystal structure prediction of a number of compounds will be presented to illustrate this method and its limitations.

[1] Desiraju, G. R. Science, 278 (1997), 404-405
[2] Leusen, F. J. J. J. Cryst. Growth, 166 (1996), 900-903