John Rehr: Fast throughput calculations and the XAFS data base in the Materials Project

Modern electronic structure theory and computational methods now permit efficient calculations of ground state properties, as exemplified by the tabulation of many-thousands of structures in the Materials Project [1]. Complementary advances in the theory of excited states have led to efficient methods for calculations of x-ray and electron spectroscopies, e.g., using the real-space Green’s multiple scattering theory in the FEFF9 code [2]. Here we discuss these developments and how they have been applied to high throughput calculations of x-ray absorption spectra in the Materials Project [3]. In particular, the world’s largest x-ray database has been constructed, which currently contains nearly 200,000 computed K-edge spectra for over 40,000 materials [4]. Recently this data base has been exploited using Machine-Learning techniques to accurately predict local coordination environments [5]. The database including the FEFF input and output files, is freely available from the Materials Project [1]. Extensions to L-edge and XAFS spectra are in progress.

References:
[1] A. Jain et al., The Materials Project: A materials genome approach to accelerating materials innovation, APL Materials 1, 011002 (2013); https://www.materialsproject.org.
[2] John J. Rehr et al., Parameter-free calculations of X-ray spectra with FEFF9, Physical Chemistry Chemical Physics 12, 5503 (2010).
[3] K. Matthew, et al., High-throughput computational X-ray absorption spectroscopy, Scientific Data 5, 180151 (2018).
[4] Chen Zheng et al., Automated generation and ensemble-learned matching of X-ray absorption spectra, npj Computational Materials 4, 12 (2018).
[5] Chen Zheng et al., Random Forest Models for Accurate Identification of Coordination Environments from X-Ray Absorption Near-Edge Structure, Patterns 1, 100013 (2020).