Artificial intelligence for catalyst design and synthesis

Accurate and efficient synthesis of catalysts is a critical issue in catalysis research. Recent advancements in artificial intelligence (AI) provide a transformative opportunity for catalyst synthesis, shifting from traditional experience-driven approaches to data-driven, automated, and intelligent design and synthesis methodology. This perspective highlights representative progress in AI-assisted catalyst synthesis, emphasizing the machine learning (ML) methods that predict catalyst structure and performance, optimize synthesis conditions, and drive automated high-throughput experimentation and characterization. In the vision of promoting deeper integration of AI into catalyst synthesis and achieving a closed-loop workflow, feasible routes of applying ML for promoting descriptor identification, large-scale calculation, and catalytic activity and stability prediction are specifically reviewed. The application of cutting-edge ML methods, such as the active learning and generative model, in catalyst design and synthesis is also discussed. The new paradigm of AI-driven catalyst synthesis is promising to benefit discovery efficiency in catalysis, material, and energy fields.