2025 AIChE Annual Meeting

(301f) Predicting Nucleation Pathways of Molecular Crystals through Low-Dimensional Solvent and Solute Order Parameters

Authors

Erik Santiso, NC State University
Nucleation and crystallization are extremely prevalent in nature, and manufacturing. Despite its apparent simplicity and technological importance, our understanding of crystallization, particularly for organic molecules of relevance to drug manufacturing, remains incomplete. Organic molecules exhibit a wide variety of geometries and functional groups, and their interactions with solvents, co-solutes, and impurities are diverse and difficult to predict. The stochastic nature of the process provides challenges to its observation in both experimental and computational settings. We have developed simulation methods to study the nucleation mechanism of different organic crystals from the melt and from solution, with the aim of gaining molecular-level understanding of the early stages of crystallization.

Our approach is based on the string method in collective variables (Maragliano et al. J. Chem. Phys. 125, 024106, 2006), in combination with order parameters obtained from the structure of the target crystal (Santiso and Trout, J. Chem. Phys. 134, 064109 (2011)). In the past we have applied this method to the crystallization of benzene, dmim-based ionic liquids, and sulfonamide drugs. A full description of the nucleation pathway is a challenging task, it is typically a complex function of many variables, which limits our choice of biased sampling method, and makes order parameter selection challenging. Recently, we have worked to extend our approach to incorporate parameters sensitive to the local structure of the solvent, and modified the order parameters to reduce the dimensionality of the collective variable space, as well as the degeneracy of the representation of the solid basin These modifications allowed us to model the nucleation with greater precision, and the reduction of dimensionality also made other rare event simulation methods computationally feasible. With these changes came new challenges in sampling the free energy surface. We will present the results obtained by incorporating these changes on the study of nucleation of a sulfonamide drug, sulfadiazine, from different pharmaceutically-relevant solvents, and will illustrate the importance of solvent collective variables in the description of crystallization.