Abstract: Molecular simulations are a powerful tool in the study of crystallization and polymorphic transitions yielding detailed information of transformation mechanisms with high spatiotemporal resolution. However, characterizing various crystalline and amorphous phases as well as sampling nucleation events and structural transitions remain extremely challenging tasks. The integration of machine learning with molecular simulations has the potential of unprecedented advancement in the area of crystal nucleation and growth. In this article, we discuss recent progress in the analysis and sampling of structural transformations aided by machine learning and the resulting potential future directions opening in this area. Graphical Abstract: [Figure not available: see fulltext.].
Bibliographical noteFunding Information:
J.R. acknowledges financial support from the Deutsche Forschungsgemeinschaft (DFG) through the Heisenberg Program Project No. 428315600. This material is based upon work supported by the US Department of Energy, Office of Science, Office of Basic Energy Sciences, under Award No. DE-SC0015448. S.S. acknowledges support from the National Science Foundation CAREER Grant Award No. 1653352 and Ruhr University Cluster of Excellence RESOLV for support to travel to and stay in Germany. S.S. acknowledges startup funds from the Department of Chemistry, University of Minnesota.
© 2022, The Author(s), under exclusive License to the Materials Research Society.
- Machine learning
- Molecular simulations