Multicomponent molecular crystals are of significant commercial and industrial importance. In the pharmaceutical industry, for example, salts, solvates, and cocrystals are widely used to modify the physical properties of APIs, such as solubility, shelf life, and mechanical properties. Despite this, crystal structure prediction (CSP) is infrequently applied to materials with more than one molecule per asymmetric unit. This is owed to the massive computational burden of multicomponent CSP. CSP requires thorough exploration of crystal potential energy landscapes, but each independent molecule contributes its own degrees of freedom, leading to an explosion of configurational possibilities. The talk will present the development of methods to improve efficiency of CSP for multicomponent systems [1]. We show that we can perform high throughput CSP of combinations of small organic molecules and evaluate convex hulls to assess how accurately we can predict the formation and structure of stoichiometric cocrystals and solvates.
[1] https://chemrxiv.org/doi/pdf/10.26434/chemrxiv-2025-l4ftw