A Perspective on Employing Computational Approaches to Interrogate Catalyst Restructuring

Catalyst restructuring under reaction conditions influences their reactivity, selectivity, and long-term stability. Because restructuring occurs across a wide range of length and time scales, computational modeling has emerged as a valuable complement to understanding these processes. These models have identified the driving forces behind restructuring, proposed molecular-level mechanisms, and codified theories for restructuring. A range of computational approaches has been employed, including empirical models, first-principles simulations, and, more recently, machine learning-based methods. These approaches are often combined with molecular dynamics, Monte Carlo simulations, and genetic algorithms to capture the structural evolution of catalysts accurately and efficiently. In this perspective, we present insights from computational models on four major types of catalyst restructuring: surface segregation, phase transition, sintering, and strong metal–support interaction. Building on these findings, we outline emerging frontiers that computational research can address in the future. These frontiers include incorporating more realistic reaction environments within computational models, developing descriptors for the restructuring of catalysts, and understanding the restructuring under dynamic catalytic conditions. Taken together, we present a unified perspective on elucidating, predicting, and ultimately controlling catalyst restructuring under realistic conditions.