Machine learning-guided design of energy-related catalysts from nanoparticles to single-atom sites

Machine learning (ML) has been widely applied to accelerate the design and discovery of new molecules and materials. Recently, there has been growing interest in integrating ML into the design of heterogeneous catalysts, including nanoparticle catalysts (NC) and, in particular, single-atom catalysts (SACs). With the rapid advancements in related research, there is an urgent need for a timely and comprehensive review in the catalysis and ML communities to provide valuable insights and guidance for future research. Here we summarize the recent advances in using ML to develop NCs and SACs for energy applications, with a focus on the characteristics of ML models, particularly in selecting features and descriptors from atom-scale structural information. First, we introduce the main ML algorithms currently in use and present examples of their application in the design of traditional NCs, which can offer valuable insights and methodological references for SAC research. We then provide a detailed overview of ML-assisted SAC design, including the establishment of structure–performance relationships, high-throughput screening, and stability prediction. Finally, we identify the challenges in this emerging field and discuss the opportunities for developing next-generation SACs using reliable datasets and advanced ML models to create better catalysts for energy applications. There has been growing interest in the application of machine learning (ML) to the design of heterogeneous catalysts, including nanoparticle catalysts (NC) and single-atom catalysts (SACs). In this Review, the authors summarize recent advances in the ML-guided design of NCs and SACs for energy applications, focusing on the selection of features and descriptors for ML models from atom-scale structural information and identifying challenges and opportunities for the development of next-generation SACs through reliable datasets and advanced ML models.