Polymorphs of an organic crystal have different energies and thermal stabilities. Theoretical studies have been focused on thermodynamics and kinetics of crystal growth regarding polymorph formation, resulting in development of several widely-adopted phenomenological and thermodynamic rules (Ostwald rule, phase rule, density rule, etc.). The role of a solvent has been thought as a kinetic factor that may trap a metastable form of a crystal due to its higher solubility in the solvent. However, why a unique crystal structure is formed in a specific solvent remains unanswered, in particular, with regard to the nucleation in which solvet-solute interactions may dictate the packing and conformation of solute molecules.
Current polymorph prediction efforts rely on a brute-force manner to search all possible packing motifs of molecules in the energy space, totally ignoring or unable to take into account the role of solvent-crystal interactions. Due to the limitation of molecular models such as the force filed to calculate molecular interactions, limited success has been achieved. The energy difference between the most stable form and a metastable form can be too small to be accurately calculated. Using QM (quantum mechanics) is out of question because computation of an energy space is merely overwhelming. It is even more challenging for organic crystals, where weak intermolecular interactions are dominant, susceptive to polymorphism. In lieu of searching endless combinations of molecules in a periodic pattern, we believe electronic structures of a solvent and crystal surfaces of the polymorph developed in the solvent should match, and therefore finding such matching patterns will produce new insights and inspire new prediction methods.
Together with protein folding, polymorphism of organic crystals is claimed to be another grand challenge of structure prediction. Given the fact that current energy-ranking methods fail to produce convincing results, new approaches must be developed.