Random Forest Models for Accurate Identification of Coordination Environments from X-Ray Absorption Near-Edge Structure.

Abstract in brief

• The study demonstrates that random forest models trained on a large dataset of X-ray absorption near-edge structure (XANES) spectra can accurately identify the atomic coordination environment in materials, achieving an accuracy of 85.4% for main environments and an 81.8% Jaccard score for all environments across 33 cation elements.
• Unlike previous methods, this approach describes coordination environments as probability distributions over 25 different motifs, with coordination numbers from 1 to 12, providing a more detailed understanding of atomic arrangements.
• The models perform well in real experimental conditions, with minimal loss in accuracy, and analysis shows that the pre-edge and main peak regions of the spectra are crucial for determining the coordination environments.]