Rational Computational Design of Next-Generation Semiconductors

We developed a rational virtual materials design strategy powered by high-throughput density functional theory (DFT) and machine learning (ML) to drive the discovery of novel semiconductors for solar absorption and photocatalysis. This strategy involved compiling massive DFT datasets of relevant properties using different semi-local and non-local hybrid functionals, performing multi-fidelity active learning to rationally generate new high-fidelity data, and training crystal graph-based neural network (GNN) models to accurately obtain any property of interest directly from the semiconductor structure. GNN models were rigorously trained on a dataset of both bulk and defect-containing structures, and combined with optimization algorithms such as Bayesian optimization and simulated annealing to quickly and accurately perform geometry optimization of completely new structures, drastically cutting down the time for full DFT. Using this approach, we successfully designed dozens of novel halide and chalcogenide compounds with suitable thermodynamic stability, electronic band gaps and edges, optical absorption behavior, intrinsic defect tolerance, ability to be doped, and carrier mobilities, concentrations, and cross-sections, with utility for a variety of optoelectronic applications.