Computational study of amorphous In₂O₃ photo-electrodes

Indium oxide (In₂O₃) is a wide-band-gap semiconductor used in thin-film transistors and optoelectronic devices which has also been proposed as a candidate photoelectrode for two-electron water-oxidation catalysis. Despite its many favorable properties, crystalline In₂O₃ exhibits an optical gap of ~ 3.55 eV, due to a forbidden transition between the valence band maximum and conduction band minimum, thus limiting its ability to capture a large portion of the solar spectrum upon light absorption. Possible strategies to lower the optical gap include straining the crystal or using amorphous samples. Here we consider the latter and use first-principles molecular dynamics and the Qbox code [1], to generate an ensemble of a-In₂O₃ structures with varied densities. We then train a machine-learning (ML) interatomic potential (using MACE [2]) to investigate larger samples and to generate structural models of a-In₂O₃ surfaces that will be interfaced with water, for a study analogous to that carried out for crystalline In₂O₃ [3]. After discussing the validation of the ML potential and the optical properties of the amorphous solid, we will describe the properties of a-In₂O₃/water interfaces.



[1] qboxcode.org

[2] Batatia, I.; Kovacs, D. P.; Simm, G. N. C.; Ortner, C.; Csanyi, G.; Advances in Neural Information Processing Systems (NeurIPS), 2022.

[3] Bousquet M.; Zhan J.; Luo C.; Martinson A.; Gygi F.; Galli G.; J. Phys. Chem. C 129 (17), 8395-8403 (2025).