2025/04/03 by Jay Johal, Graeme M. Day · 1 voice
Materials Science · Computer Science · Chemistry · #Machine Learning in Materials Science #Computational Drug Discovery Methods #Various Chemistry Research Topics
paper · pdf · doi:10.26434/chemrxiv-2025-v8692
openalex publication_date 2025/04/03 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/31
Organic molecular crystals offer a broad spectrum of potential applications. The vast number of possible molecules is both an opportunity and a challenge, because of the prohibitive expense of exhaustively searching chemical space to find novel molecules with promising solid-state properties. Computational methods can be applied to direct experimental discovery programs using high-throughput or guided searches of chemical space. However, to date, such approaches have largely focused on molecular properties, ignoring the often significant effects of the arrangement of molecules in their crystal structure on the molecule's effectiveness for the chosen application. Here, we present CSP-EA, an evolutionary algorithm for searching chemical space that incorporates crystal structure prediction (CSP) into the evaluation of candidate molecules, allowing their fitness to be evaluated based on the predicted materials' properties. As a demonstration, CSP-EA is applied here to a search space of organic molecular semiconductors, demonstrating that the inclusion of CSP in the fitness assessment outperforms searches based on molecular properties alone in identifying molecules with high electron mobilities.