Accelerated Development of Perovskite-Inspired Materials via High-Throughput Synthesis and Machine-Learning Diagnosis

Accelerating the experimental cycle for new materials development is vital for addressing the grand energy challenges of the 21st century. We fabricate and characterize 75 unique perovskite-inspired compositions within a 2-month period, with 87% exhibiting band gaps between 1.2 and 2.4 eV, which are of interest for energy-harvesting applications. We utilize a fully connected deep neural network to classify compounds based on experimental X-ray diffraction data into 0D, 2D, and 3D structures, more than 10 times faster than human analysis and with 90% accuracy. We validate our methods using lead-halide perovskites and extend the application to lead-free compositions. The wider synthesis window and faster cycle of learning enables the realization of a multi-site lead-free alloy series, Cs3(Bi1-xSbx)2(I1-xBrx)9. We reveal the non-linear band-gap behavior and transition in dimensionality upon simultaneous alloying on the B-site and X-site of Cs3Bi2I9 with Sb and Br. [Display omitted] •Investigation of 75 perovskite-inspired compositions in thin-film form (41 Pb free)•A deep neural network classifies perovskites into 0D, 2D, and 3D structures•Non-linear band-gap behavior discovered in Cs3(Bi1-xSbx)2(I1-xBrx)9 dual-site alloys To meet increasing global energy demand, it is critical yet challenging to accelerate the development of novel energy materials. High-throughput experimentation (HTE) and machine-learning techniques have become increasingly accessible to scientific researchers. We herein demonstrate a case study on perovskite-inspired materials, where a combination of fast synthesis and machine-learning-assisted data diagnostics of 75 compositions achieves an acceleration of over an order of magnitude per experimental learning cycle over our laboratory baseline. The increased throughput and streamlined workflow enable the realization of new candidate photovoltaic materials, which sheds light on the search for lead-free perovskites in this multi-parameter chemical space. Our study demonstrates that combining an accelerated experimental cycle of learning and machine-learning-based diagnosis represents an important step toward realizing fully automated laboratories for materials discovery and development. Fast experimental cycles enable exploration of wide chemical space and data-driven analysis.