(55f) Coarse-Grained Modeling of Crystal Growth Using Forcefields from Iterative Boltzmann Inversion and Particle-Swarm Optimization

Because of the infeasibility of directly simulating growth of drug crystals using molecular dynamics (MD) with atomistic force fields, we develop course-grained (CG) force fields (FFs) at two levels, with each CG bead representing either 5-7, or 2-3 “heavy” (i.e., non-hydrogen) atoms within ring-containing drug molecules. The first, highly coarse-grained, FF is produced using Iterative Boltzmann Inversion (IBI), while the second is obtained using Particle Swarm Optimization (PSO). Both are optimized to produce stable crystal unit cells for a representative drug molecule, carbamazepine, and a co-crystal of it with succinic acid. Both forcefields allow growth of crystal facets into a melt of the same molecules, but the more highly coarse-grained forcefield can give unrealistic structure of the melt, and in some cases unphysical growth rates of crystal facets, while growth rates of crystal facets derived for the model with less coarse graining are more realistic physically. In addition, the more resolved CG forcefield is better able to capture the effects of hydrogen bonding as well as conformational changes in the molecule, and are much better suited to simulation of co-crystals. Similar studies of a second crystallizing drug molecule, phenytoin, will also be presented. Various challenges encountered in this work, and the prospects for further work on coarse-graining of drug molecules for MD are discussed.