Bandgap engineering in the configurational space of solid solutions via machine learning: (Mg,Zn)O case study

Midgley, S. D., Hamad, S., Butler, K. T. and Grau-Crespo, R. ORCID: https://orcid.org/0000-0001-8845-1719 (2021) Bandgap engineering in the configurational space of solid solutions via machine learning: (Mg,Zn)O case study. Journal of Physical Chemistry Letters, 12 (21). pp. 5163-5168. ISSN 1948-7185 doi: 10.1021/acs.jpclett.1c01031

Abstract/Summary

Computer simulations of alloys’ properties often require calculations in a large space of configurations in a supercell of the crystal structure. A common approach is to map density functional theory results into a simplified interaction model using so-called cluster expansions, which are linear on the cluster correlation functions. Alternative descriptors have not been suf-ficiently explored so far. We show here that a simple descriptor based on the Coulomb matrix eigenspectrum clearly outper-forms the cluster expansion both for total energy and bandgap energy predictions in the configurational space of a MgO-ZnO solid solution, a prototypical oxide alloy for bandgap engineering. Bandgap predictions can be further improved by introducing non-linearity via gradient-boosted decision trees or neural networks based on the Coulomb matrix descriptor.