Yicheng Zhao, Bowen Liu, Ziyi Liu
University of Electronic Science and Technology of China, Chengdu, China 

Abstract: With the swift progress of wet chemistry, a series of functional materials can be synthesized into precursor solutions utilizing liquid solvents, including quantum-dot colloid solutions, perovskite solutions, and organic solutions etc.. Various electronic devices derived from these solution-processed materials are inherently influenced by modulating their ratios, combinations, and processing parameters. To expedite the development of complex semiconductor materials and devices, our lab, grounded in a high-precision multi-channel liquid distribution system, pioneered the first high-throughput intelligent experimental platform for electronic thin films and devices. The platform covers autonomous experimental design, high-throughput thin film preparation, high-throughput device fabrication, high-throughput optoelectronic characterization, and intelligent data analysis. The daily throughput can reach 104, 103, and 104 for solutions, devices, and data analysis respectively. Employing such a new research paradigm, we successfully optimized complex perovskite semiconductor thin films (AB X3,'A'site: potassium (K+), rubidium (Rb+), cesium (Cs+), methylammonium (MA+), and formamidinium (FA+); 'B'site: lead (Pb+) and tin (Sn+);'X'site: iodine (I-), bromine (Br-), and chlorine (Cl-)). Based on the stable perovskite thin films, we further fabricated ultra-stable perovskite solar cell devices, demonstrating the superiority of this data-driven research paradigm.
Keywords: High-throughput experimentation; Machine learning; Perovskite; Solar cells; Data-driven.