Berlin 2018 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

MM: Fachverband Metall- und Materialphysik

MM 75: Topical Session (Symposium MM): Big Data in Materials Science - Managing and exploiting the raw material of the 21st century

MM 75.5: Talk

Friday, March 16, 2018, 12:30–12:45, H 0107

Machine learning the structure-energy-property landscapes of molecular crystals — •Felix Musil¹, Sandip De¹, Jack Yang², Joshua Campbell², Graeme Day², and Michele Ceriotti¹ — ¹COSMO, EPFL, Lausanne, Switzerland — ²University of Southampton, Southampton, UK

Molecular crystals play an important role in several field of science and technology. They often crystallize in many different polymorphs with substantially different physical properties. To help prioritize the synthesis of candidate materials, atomic-scale modelling can be used to enumerate the stable polymorphs and to predict their properties, as well as to propose heuristic rules to rationalize the correlations between crystal structure and materials properties.

Here we show how a recently-developed machine-learning (ML) framework [1] can be used to achieve inexpensive and accurate predictions of the stability and properties of polymorphs, and a data-driven classification that is less biased and more flexible than typical heuristic rules. We discuss, as examples, the lattice energy landscapes and the automatic structural classification of pentacene and two azapentacene isomers that are of interest as organic semiconductor materials. We show that we can estimate lattice energies with sub-kJ/mol accuracy, using only a few hundred reference configurations, and provide a more detailed picture of molecular packing than that provided by conventional heuristics.

[1] De, S., Bartok, A. P., Csanyi, G., & Ceriotti, M. (2016). Phys. Chem. Chem. Phys., 18(20), 13754.