(596d) Exploring the Multi-Minima Behavior of Organic Crystal Polymorphs at Finite-Temperature
AIChE Annual Meeting
2017
2017 Annual Meeting
Pharmaceutical Discovery, Development and Manufacturing Forum
Development of Processes and Products for Pharmaceuticals and Hybrid Therapeutics
Wednesday, November 1, 2017 - 4:18pm to 4:39pm
A common hypothesis put forward for why some lattice minima do not correspond with a unique experimental structure is that the thermal motions under working conditions will cause many lattice minima to rapidly restructure into a more stable form or cause multiple minima to coalesce into a single indistinguishable ensemble. Under this hypothesis, some potential energy basins explored at high temperature should be `rough' landscapes filled with multiple local energy minima rather than the classic picture of a smooth potential energy basin with a single global minimum-energy structure.
Here we demonstrate the existence of these rough basins with multiple lattice minima in a number of different polymorphic systems. The experimental crystal structures of twelve polymorphic systems were heated to a range of temperatures up to ambient conditions using molecular dynamics simulations. A set of configurations from each trajectory was then quenched to zero Kelvin using crystal energy minimization. We find that the high-temperature configurations of rigid-molecule crystals tend to all collapse to the same initial lattice minima after quenching. However, the configurations from more flexible molecules tend to minimize into a myriad of different lattice minima, suggesting that flexibility leads to multi-minima behavior. We also find that the number of minima found in the crystal basins increases with temperature, and the presense of multiple lattice minima within the free energy basin is polymorph-dependent in addition to being molecule-dependent. The existence of these multi-minima basins at ambient conditions lends credence to the theory that incorporating thermal motion can reduce the number of spuriously predicted structures in future CSP studies.