In photovoltaic (PV) materials development, the complex relationship between device performance and underlying materials parameters obfuscates experimental feedback from current-voltage (J-V) characteristics alone. Here, we address this complexity by adding temperature and injection dependence and applying a Bayesian inference approach to extract multiple device-relevant materials parameters simultaneously. Our approach is an order of magnitude faster than the cumulative time of multiple individual spectroscopy techniques, with added advantages of using device-relevant materials stacks and interface conditions. We posit that this approach could be broadly applied to other semiconductor- and energy-device problems of similar complexity, accelerating the pace of experimental research. Photovoltaic (PV) research has historically taken decades to bring new materials to market, as the pace of development is often limited by our ability to identify the causes of underperformance. There are many ways to make an underperforming PV cell, and their signatures on device efficiency alone are not unique. This uncertainty plagues other energy-storage or -conversion devices with similarly complex combinations of materials but may be used to our advantage. By probing or biasing a PV cell in different operating conditions, the signature of underlying material properties becomes more unique. This connection between the operating conditions, the material properties, and the PV cell output may be solved through Bayesian inference algorithms. While computationally expensive, high-performance computing can enable such inference as a tool for experimentalists, a tool that could become increasingly valuable for accelerating the pace of materials research in PV and related fields. High-performance computing can greatly improve the workflow of experimentalists in energy materials, through the use of Bayesian inference. This allows us to solve the inverse problem of extracting underlying materials properties through the measurement of the electrical behavior of completed devices. Cheaper, faster measurements can be substituted for longer direct measurements of individual properties, without sacrificing accuracy or precision. We provide a general framework to apply this to other materials systems and devices.

Tin monosulfide (SnS) is an emerging thin-film absorber material for photovoltaics. An outstanding challenge is to improve carrier lifetimes to >1 ns, which should enable >10% device efficiencies. However, reported results to date have only demonstrated lifetimes at or below 100 ps. In this study, we employ defect modeling to identify the sulfur vacancy and defects from Fe, Co, and Mo as most recombination-active. We attempt to minimize these defects in crystalline samples through high-purity, sulfur-rich growth and experimentally improve lifetimes to >3 ns, thus achieving our 1 ns goal. This framework may prove effective for unlocking the lifetime potential in other emerging thin-film materials by rapidly identifying and mitigating lifetime-limiting point defects.