Solar thermochemical water splitting is a promising pathway for H<sub>2</sub> production (STCH) with near-zero greenhouse gas emissions. To develop practical reactors, however, the efficiency of redox-active metal oxides for thermochemical cycling needs to be enhanced; specifically, their capacity for oxygen off-stoichiometry must be optimized. Oxygen vacancy formation energies (OVFEs) are the key metric determining this off-stoichiometry. Metal oxide perovskites have received intense scrutiny for clean energy applications due to their flexible structure, stability, composition, and electrical conductivity. While exciting progress continues, a simple phenomenology for trends in OVFEs remains elusive. Here, we introduce a machine-learned minimal model for the OVFEs of ABO<sub>3</sub> perovskites, where A = {Ca, Sr, Ba, and La} and B = {Ti, V, Cr, Mn, Fe, Co, and Ni}, in six lattice systems (monoclinic, orthorhombic, tetragonal, rhombohedral, hexagonal, and cubic). The model takes as inputs crystal bond dissociation energies, crystal reduction potentials, band gaps, and energies above the convex hull, which can be obtained from theoretical or experimental databases. Additionally, we demonstrate that the model can be simplified, with acceptable losses in accuracy, such that only crystal bond dissociation energies and crystal reduction potentials are needed. Finally, we present our perspectives on how to improve and extend the model, which already provides both accurate and efficient predictions for high-throughput screening and an intuitive and modular phenomenology for renewable energy applications of metal oxide perovskites and beyond.

Efficient machine-learned model for oxide perovskite performance in solar thermochemical technologies

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.