Bridging the complexity gap in computational heterogeneous catalysis with machine learning

Heterogeneous catalysis underpins a wide variety of industrial processes including energy conversion, chemical manufacturing and environmental remediation. Significant advances in computational modelling towards understanding the nature of active sites and elementary reaction steps have occurred over the past few decades. The complexity gap between theory and experiment, however, remains overwhelming largely due to the limiting length and timescales of ab initio simulations, which severely impede the discovery of high-performance catalytic materials. This Review summarizes recent developments and applications of machine learning to narrow and, optimistically, bridge the gap created by the dynamic, mechanistic and chemostructural complexities inherent to the reactive interfaces of practical relevance. We foresee the prospects and challenges of machine learning for the automated design of sustainable catalytic technologies within a data-centric ecosystem that coevolves with computational and data sciences. Computational chemistry has the potential to aid in the design of heterogeneous catalysts; however, there is currently a large gap between the complexity of real systems and what can be readily computed at scale. This Review discusses the ways in which machine learning can assist in closing this gap to facilitate rapid advances in catalyst discovery.