Computational Identification of Ternary Wide-Band-Gap Oxides for High-Power Electronics

Abstract

As electricity grids become more renewable energy compliant, there will be a need for novel semiconductors that can withstand high power, high voltage, and high temperatures. Currently used or explored wide-band-gap materials for power electronics are costly (GaN), difficult to synthesize as high-quality single crystals (SiC) and at scale (diamond, BN), have low thermal conductivity (β-Ga_{2}O_{3}), or cannot be suitably doped (AlN). We conduct a computational search for novel semiconductors across 1340 known metal oxides using first-principles calculations and existing and improved transport models. We calculate the Baliga figure of merit (BFOM) and lattice thermal conductivity (κ_{L}) to identify top candidates for n-type power electronics. We find 47 mostly ternary oxides that have higher κ_{L} than β-Ga_{2}O_{3} and higher n-type BFOM than SiC and GaN. We use the branch point energy to rank the likelihood of n-type extrinsic doping, further reducing our top candidates to 14 previously unexplored compounds. Among these, several material classes emerge, including 2-2-7 stoichiometry thortveitites and pyrochlores, II-IV spinels, and calcite-type borates. Within these classes, we propose In_{2}Ge_{2}O_{7}, Mg_{2}GeO_{4}, and InBO_{3} for power electronics as they are the most favorable for n-type doping based on our preliminary evaluation and could be grown as single crystals or thin-film heterostructures.