Transforming Catalysis with Machine Learning: Emerging Tools and Next-Gen Strategies.

Abstract

Catalysis plays a central role in the modern chemical industry, yet the discovery of high-performance catalysts remains constrained by traditional experimental and ab initio calculation approaches. Recently, the rapid development of machine learning methods has caused a major revolution in the field of catalytic chemistry, which promises to accelerate the catalyst development with unprecedented efficiency. This review systematically introduces the fundamental concepts and workflows of ML in catalysis, followed by a comprehensive overview of both traditional machine learning─typically based on small data sets and shallow models─and deep learning (DL), which leverages large-scale data and complex architectures. We highlight key modeling strategies, algorithmic frameworks, and representative applications in catalyst design, reaction prediction, and surface adsorption systems. Finally, we discuss current challenges, including fragmented and inconsistent data, limited physical interpretability, and difficulties in integrating ML with experimental workflows, and propose future directions to address these issues and further advance the field.