Data- and AI-Driven Catalysis Science: From Descriptors to Digital Catalysis for Next-Generation Catalysts

Data science and artificial intelligence (AI) are profoundly changing the paradigm of mechanistic research and design of catalysts. This work emphasizes the central role of data science and AI in catalyst design, progressing from low-dimensional to high-dimensional approaches, covering reactivity-related properties using density functional theory (DFT) descriptors, high-dimensional catalysis science supported by various databases, advanced applications of machine learning (ML) and ML interatomic potentials (MLIPs) in mechanistic analysis and catalyst design, and a perspective of Digital Catalysis as an emerging direction. Specifically, we first discuss the fundamental role of DFT descriptors in constructing activity volcano models which bridge microscopic mechanisms with macroscopic performance, and then explore the critical value of theoretical database combined with ML and MLIPs in accelerating thermodynamic and reaction mechanistic calculations as well as decoupling structure–activity relationships. Finally, taking the emerging Digital Catalysis Platform ( DigCat ) as a key example, we discuss the current status and development prospects of digital catalysis platforms. This work highlights the transformative role of data science and AI in bridging theoretical insights, computational efficiency, and experimental validation, ultimately accelerating the design of next-generation catalysts for a sustainable future.