The computational prediction of pharmaceutical crystal structures and polymorphism.

A computational method of predicting all the polymorphs of an organic molecule would be a valuable complement to polymorph screening in the developmental phase. Such a computational method is in its early stages of development, and the current methodologies, which are based on searches for the most stable lattice structure, are critically reviewed. This crude thermodynamic approach generally overestimates the propensity for polymorphism, at least for most of the molecules studied so far, showing the need to model kinetic effects as well as to refine the thermodynamic models. Although the ultimate goal of these studies is still far off, computational predictions of crystal structures have proved useful in aiding the characterisation of polymorphs from powder X-ray data, and in providing insights into the range of types of packing that may be adopted by a given molecule. Thus, computational studies already have the potential to be a valuable tool in pharmaceutical solid state science.