Machine learning-assisted discovery of chevrel phase tellurides

In a recent JACS article, the Velázquez lab, in collaboration with the Musgrave lab from the Department of Chemical and Biological Engineering at University of Colorado Boulder, showcased an interpretable machine-learned descriptor (Hd) capable of estimating decomposition enthalpy (ΔHd) to identify synthetically accessible molybdenum chalcogenides within the Chevrel Phase (CP) family from a set of 205,548 different CP compositions. Five new CP tellurides were identified with this descriptor and successfully synthesized using a microwave-assisted solid-state approach, doubling the number of previously identified metal-intercalated phases for the CP tellurides. The Velázquez lab was also able to experimentally confirm computational predictions regarding cavity occupancy as a function of metal intercalant for the CP tellurides synthesized. The results of this work provide a joined computational and experimental approach that can be applied for the discovery of other multinary materials that have shown promise in energy conversion and storage applications.

For more information, please see the article at https://pubs.acs.org/doi/pdf/10.1021/jacs.1c02971 .