A FAIR approach towards constructing a computational perovskite database

Halide and chalcogenide perovskites are promising materials for electronics, solar absorption, and quantum technologies, but their vast compositional landscape necessitates the use of density functional theory (DFT) simulations accelerated by machine learning (ML) to guide experiments. To advance this research, it is imperative to establish standardized and comprehensive workflows for the systematic study of perovskites. This article outlines the methodologies used to construct a “DFT + ML” perovskite database and implement FAIR (Findable, Accessible, Interoperable, and Reusable) data/model-sharing principles. We discuss our computational workflow, dataset, and different types of ML models, which are each released publicly via the nanoHUB platform.