Energy partitioning of pharmaceutical co-crystal structures

Abstract

Active pharmaceutical ingredients (APIs) and selected coformers were studied in crystalline self-environment, in selected co-crystal structures, and gas phase. To obtain comparable geometries from experimental crystal structures, we rely on fast quantum mechanical GFN2-XTB molecule-in-cluster optimizations. Optimized crystal structures are first analysed in terms of molecule-pair interaction energies, by subdividing a cluster generated from asymmetric unit content through space-group symmetry into pairs of molecules. Accurate energies are then obtained by single-point molecular orbital (MO/MO) two layer “ONIOM” energy partitioning, comparing the same molecules in different crystal environments. Clusters approximate the local environment of a crystal structure for ONIOM. The pre-processor program BAERLAUCH (Acta. Cryst. A 2012) was used to set up all cluster computations. Solid-state computations using dispersion-corrected density functional theory provide reference results. ONIOM partitioned energies shed light on the driving force of co-crystal formation, with the energy gain from the complementarity of molecule-pair wavefunctions in such structures being the main driving force. It follows that improving prediction of co-crystal structure formation beyond current approaches, e.g. COSMO-RS excess enthalpies, statistics from structural databases or full crystal structure prediction is possible by finding complementary molecule-pairs through conformational sampling, as ultimately exemplified.