Research progress of machine learning in the field of photocatalysis applications

Photocatalysis is a technology that uses solar energy to drive chemical reactions, and it promotes oxidation–reduction reactions under the condition of light irradiation by photocatalysts, achieving the purification of pollutants, synthesis and transformation of substances, and so on. The essence of photocatalytic technology centers on photocatalysts. Nevertheless, the traditional trial-and-error development process of photocatalytic materials cannot meet the development needs of modern society due to unfavorable factors such as high cost, low efficiency, and lengthy research and development periods. In recent times, the fast advancement of machine learning (ML) technology has opened up new avenues for the design of photocatalysts. With the continuous deep integration of big data and artificial intelligence (AI), machine learning, which is data-driven, has made tremendous progress in the design, screening, and performance prediction of new materials, greatly promoting the research and application of novel materials. This review first introduces the basic process of ML and its commonly used algorithms in materials science, and then focuses on the latest research progress in the use of ML in photocatalytic water splitting for hydrogen production, photocatalytic pollutant degradation, and photocatalytic harmful gas conversion in recent years. Furthermore, it provides a prospect on the existing problems and development prospects of ML in the screening and design of photocatalysts, the prediction of photocatalytic performance, the parameter optimization of photocatalytic processes, and so on.