Building Catalyst Exploration Highways by Integrating High‐Throughput and Machine Learning Technologies

ABSTRACT There is growing interest in highly active catalysts to meet the increasing requirements of high‐efficiency energy conversion and storage. However, catalyst investigation through traditional trial‐and‐error experiments is impeded by the long duration and high cost involved. Recently, high‐throughput (HT) technologies that enable rapid catalyst screening in a vast parameter space within days (or even hours) have revolutionized catalyst research. Machine learning (ML) is a powerful tool for extracting knowledge from large datasets generated by HT experiments and computational calculations. The integration of HT with ML technologies demonstrates great promise for facilitating the discovery of advanced catalysts. From this perspective, this study highlights the use of frameworks integrating HT experimental/computational techniques with ML methodologies for catalyst research, with emphasis on the progress in such frameworks in identifying optimal candidates, elucidating reaction mechanisms, and determining critical descriptors. Furthermore, future directions for accelerating high‐performance catalyst discovery using automated HT synthesis platforms, advanced in situ characterization, and intelligent ML algorithms are outlined. This work is expected to establish a closed‐loop workflow toward the design of highly active catalysts and provide a new avenue for fundamental theory and practical applications in catalyst research.