Among various environmental remediation techniques, photocatalytic techniques, which uniquely take advantage of the sunlight, have attracted much attention. However, owing to the poor stability of catalysts, photocatalytic technology is predominantly confined to laboratory research at present, posing challenges for its large-scale application in environmental remediation and resulting in a scarcity of data on its full lifecycle management. This review proposes that machine learning (ML) may serve as a critical link in advancing photocatalytic technology from fundamental research to mature application processes. The data-explosive, scalable development approach facilitated by ML is expected to significantly overcome the limitations of the current trial-and-error method for developing photocatalysts. Integrated with a ML guided full life cycle assessment (LCA) encompassing “material design, process selection, material/energy consumption, pollution disposal, and environmental impact,” this approach can contribute to forming a comprehensive photocatalytic environmental remediation technology system.
Wang et al. (Sat,) studied this question.