(780d) A Kinetic Approach to Polymorph Prediction: Polymorph Specific Alpha and Beta Glycine Nucleation Rates and Solubility Curves from Molecular Dynamics

Authors: 
Parks, C., Purdue
Koswara, A., Purdue University
Tung, H. H., AbbVie Inc.
Nere, N., AbbVie Inc.
Bordawekar, S., AbbVie Inc.
Nagy, Z. K., Purdue University
Ramkrishna, D., Purdue University
Predicting polymorphism, the ability of a molecule to self-assemble into multiple solid-state crystalline forms, continues to elude both experimentalists and molecular modelers alike. Landscape energy minimization techniques, although elegant in their own right, fail to include standard design crystallization variables such as solvent, temperature, or pressure. Furthermore, the effect of crystal size, a critical variable for molecules that display size dependent polymorphism, remains unaddressed in these optimization algorithms. Molecular dynamics (MD), which encapsulates the aforementioned kinetic design variables naturally, has suffered from time scale challenges associated with activated rate processes, such as nucleation. However, with the advent of modern supercomputers, and HPC hardware such as xeon phi coprocessors, these types of simulations are just now becoming within reach of computational scientists. To date, only a handful of attempts to determine the effects of the nucleation pathway on polymorphism of organic molecules have been performed using MD . 

In this work, a first principles molecular dynamics approach to the study of homogeneous nucleation rates and solubility of polymorphs is presented for both alpha and beta glycine in water as a model system. To this end, a rare-event molecular dynamics simulation methodology is used in combination with a newly derived stochastic model for one-dimensional nucleation barrier crossings. This works seeks to assess the degree to which polymorph specific nucleation kinetics and solubility, both at the nanoscale and bulk scale (>1000nm), can be used for polymorph prediction. Bulk solubility curves over a 45 K temperature range are predicted directly from molecular dynamics. Both the trend-line estimated and MD predicted alpha glycine solubility values are within 5% to 11% agreement with experimental values. Beta glycine solubility estimation is achieved purely by MD prediction, as the solubility of this polymorph is not measurable in water due to rapid recrystallization kinetics. From the solubility calculation, alpha glycine is shown to be less soluble than beta glycine over the entire 45K temperature window at bulk length sales. This is in agreement with experimental evidence that shows the presence of alpha glycine at equilibrium in aqueous solvent for micron-sized particles. However, it is also shown that beta glycine is less soluble than alpha glycine at the nanoscale. Furthermore, consistent with Ostwaldâ??s rule of stages, the nucleation kinetics show a strong preference for the formation of beta glycine at ~2nm length scales, due to beta glycine critical clusters being ~40 kcal/mol more stable than alpha glycine. This yields the prediction of beta formation at these length scales as well as the size dependent polymorphism of glycine.