Investigating polymorphic behaviour of organic molecules using theoretical and experimental techniques
This thesis describes complementary theoretical and experimental investigations into polymorphism of several structurally similar small organic molecules. Five biologically important molecules (DNA/RNA bases) and five barbiturate type compounds were subjected to a wide range of crystallisation conditions which resulted in the characterisation of new products, including new polymorphs of barbituric acid and 6- methyluracil. These results were correlated with computational polymorph predictions and in some cases rationalised to investigate why some structures have unused hydrogen bond acceptors. The crystal structure prediction process was evaluated on hydantoin as part of the formal third crystal structure prediction blind test organised by the Cambridge Crystallographic Data Centre. This involved performing the calculations from just the known molecular structure. The compounds 3-oxauracil and 5-hydroxyuracil were also studied in an informal blind test scenario in collaboration with GlaxoSmithKline. In all cases the experimental crystal structure was found within the lowest energy structures. This work demonstrates that structurally similar organic molecules can have different patterns of the relative energies of the hypothetical low energy crystal structures, along with differences in the experimental polymorphic behaviour. Molecular flexibility and the model applied for the intermolecular forces can reorder the relative stabilities of the low energy structures. These variations are often comparable to the differences in the energies of the different crystal structures which considerably decrease the confidence in the computational predictions. The kinetics of crystallisation and the limitations in the range of crystallisation techniques available are important in determining which polymorphs are seen experimentally. The range of experimental and computational polymorphic behaviour exhibited by the molecules described in this thesis highlights the challenges involved in polymorph prediction.