Data-driven material discovery for photocatalysis: a short review

Abstract

In this short review, we introduce recent progress in the research field of data-driven material discovery and design for solar fuel generation. Construction of material databases under the materials genome initiative provides a great platform for material discovery and design by creating computational screening pipelines based on the materials' descriptors. In the field of solar water splitting, data-driven computational discovery approach has been effective in making material predictions. When combined with synergistic and complimentary experimental efforts, high-throughput computations based on density functional theory showed great predictive power for accelerated discovery of inorganic compounds as functional materials for solar fuel generation. As an example, we introduce the theory–experiment joint discovery of a large set of metal oxide photoanode materials that have been theoretically predicted to be efficient candidates and soon verified by synergistic experimental fabrication and characterization processes. In the field of two-dimensional materials, the application of data-driven approach has realized the prediction of many promising candidates with suitable direct band gaps and optimal band edges for the generation of chemical fuels from sunlight, greatly expanding the number of theoretically predicted 2D photoelectrocatalysts that are awaiting experimental verification. We discuss the challenges for the continued discovery and design of novel bulk and 2D compounds for photocatalysis via a data-driven approach. At the end of this review, we provide a brief outlook for future material discoveries in the field of solar fuel generation.