Exploring the Defect Landscape and Dopability of Chalcogenide Perovskite BaZrS3

Abstract

BaZrS₃ is a chalcogenide perovskite that has shown promise as a photovoltaic absorber, but its performance is limited because of defects and impurities, which have a direct influence on carrier concentrations. Functional dopants that show lower donor-type or acceptor-type formation energies than naturally occurring defects can help tune the optoelectronic properties of BaZrS₃. In this work, we applied first-principles computations to comprehensively investigate the defect landscape of BaZrS₃, including all intrinsic defects and a set of selected impurities and dopants. BaZrS₃ intrinsically exhibits n-type equilibrium conductivity under both S-poor and S-rich conditions, which remains largely unchanged in the presence of O and H impurities. La and Nb dopants created stable donor-type defects which make BaZrS₃ even more n-type, whereas As and P dopants formed amphoteric defects with relatively high formation energies. This work highlights the difficulty of creating p-type BaZrS₃ owing to the low formation energies of donor defects, both intrinsic and extrinsic. Defect formation energies were also used to compute expected defect concentrations and make comparisons with experimentally reported values. Our dataset of defects in BaZrS₃ paves the path for training machine learning models to subsequently perform larger-scale prediction and screening of defects and dopants across many chalcogenide perovskites, including cation-site or anion-site alloys.