The CrystalPredictor&lt;span style="font-size:10.8333px"&gt; &lt;/span&gt;and CrystalOptimizer codes have been used to explore the space of crystal structures successfully in several crystal structure prediction (CSP) investigations in recent years, including in the series of blind tests organised by the Cambridge Crystallographic Data Centre and in the prediction of the crystal structures of pharmaceutically-relevant and industrially relevant (Organic semi conductors and switch sensor) molecules. Being able to predict the polymorphs a given molecule can form, a priori, has several applications: To derisk the catastrophic appearance of a late-appearing polymorph To target experimental crystallsations by screening candidate molecules for desirable physical properties To aid in the resolution of structures in the absence of quality single crystal experimental data To computationally screen coformers for cocrystal crystallisation However, CSP landscapes can often overpredict the number of viable structures a given molecule can form, and fail to correctly rank experimental polymorphs. This can be an issue with the accuracy of the necessarily efficient underlying energy model, and this presentation will describe a recent development that uses a surrogate model of molecule interactions in crystals to achieve periodic DFT accuracy at forcefield cost. Periodic DFT itself is, occasionally, insufficiently accurate to correctly rank the experimental polymorphs for a given molecule, with the molecule ROY a notorious example, due to overbinding. Using a conformational energy correction approach, we will present the results of a recent collaboration that was able to correctly rank the ROY polymorphs for the first time, as well as suggest the only route to any more polymorphs would be through very high pressure crystallisation techniques. Even given energy models that can describe crystal structures completely accurately, what of instances where crystallisation is difficult or impossible? This presentation will also consider the metadynamics refinement of polymorphic landscapes, and its application to the “Sulflower” class of potential organic semi conductors to answer this question.

Getting the most out of crystal structure prediction landscapes

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.