Abstract
The photocatalytic construction of carbon-carbon (C-C) bonds provides an environmentally friendly pathway for synthesizing complex organic molecules. Yet, the systematic design of highly efficient photocatalysts still poses a substantial challenge. This research introduces a innovative method that utilizes interpretable deep learning to streamline the design of single-crystal perovskite catalysts, aiming to enhance photocatalytic C-C coupling reactions. By combining first-principles calculations with advanced machine learning frameworks, we develop a predictive model that uncovers the structure-activity relationships influencing catalytic performance. Our approach allows for the identification of critical structural and electronic characteristics in perovskite materials that facilitate effective C-C bond formation under light exposure. This work highlights the capacity of artificial intelligence-driven strategies to drastically accelerate the discovery and optimization of next-generation photocatalysts for organic synthesis applications.
