Abstract
Solution-processed inorganic solar cells with less toxic and earth-abundant elements are emerging as viable alternatives to high-performance lead-halide perovskite solar cells. However, the wide range of elements and process parameters impede the rapid exploration of vast chemical spaces. Here, we developed an automated robot-embedded measurement system that performs photoabsorption spectroscopy, optical microscopy, and white-light flash time-resolved microwave conductivity (TRMC). We tested 576 films of quaternary element-blended wide-bandgap Cs-Bi-Sb-I semiconductors with various compositions, organic salt additives (MACl, FACl, MAI, and FAI, where MA and FA represent methylammonium and formamidinium, respectively), and thermal annealing temperatures. Among them, we found that the maximum power conversion efficiency (PCE) was 2.36%, which is significantly higher than the PCE of 0.68% for a reference film without an additive. Machine learning (ML) and statistical analyses revealed significant features and their relationships with TRMC transients, thereby demonstrating the advantages of combining ML and automated experiments for the high-throughput exploration of photovoltaic materials.