W0079

Progress and Problems in the Computer Prediction of Molecular Crystal Structures and Polymorphism. S.L. Price, Centre for Theoretical and Computational Chemistry, University College London, 20 Gordon Street, London WC1H 0AJ, UK

Considerable progress has been made towards methods of predicting the crystal structures of rigid organic molecules, using only the molecular structure as input. These methods are based on searching for the global minima in the static lattice energy, with any competitive low energy structures being possible polymorphs.

We have used the MOLPAK [1] program to generate hypothetical close packed structures for molecules such as uracil, 6-azauracil, allopurinol [2], alloxan [3], indigo and pyridone. These have been optimized by minimizing their lattice energies, using an accurate model for the dominant electrostatic forces (an atomic multipolar expansion of an ab initio charge distribution). The experimental structures were successfully found, although there were alternative structures within the energy range of possible polymorphs in many cases. Thus, state-of-the-art estimates of lattice energies confirm the common observation that many molecules have more low energy minima in the lattice energy than known polymorphs. The problems in improving our ability to predict polymorphism will be highlighted for discussion.

1. J.R. Holden, Z.Y. Du, H.L. Ammon, J. Comput. Chem., 14, 422 (1993).

2. S.L. Price, K.S. Wibley, J. Phys. Chem. A, 101, 2198 (1997).

3. D.S. Coombes, G.K. Nagi, S.L. Price, Chem. Phys. Letts., 265, 532 (1997).