Dynamic Evolution Redefines the Active Site of Cu/ZnO for Methanol Synthesis Through Iterative Molecular Dynamics Sampling.

The atomic-scale identification of active sites in Cu/ZnO catalysts for CO2 hydrogenation to methanol remains a longstanding challenging, owing to their dynamic transformation into poorly defined ensembles under working conditions. The interactions between adsorbates and active sites are the key driving force behind this interfacial evolution. Herein, we develop an iterative molecular dynamics sampling approach by integrating density functional theory, ab initio molecular dynamics, and genetic algorithm optimization to effectively capture the Cu/ZnO interface ensembles under reaction conditions, as induced by adsorbates such as CO2 and H species. We find that specific interfacial active-site configurations lower the activation barrier for formate formation through distinct electronic and geometric features, which we define as the peripheral state effect. This effect substantially improves the agreement between the calculated and experimental turnover frequencies, reducing the several-orders-of-magnitude discrepancy typically observed with conventional ground-state models. Furthermore, we propose a dynamic evolvability principle, showing that low-coordination Cu clusters better exploit the peripheral state effect owing to their greater structural flexibility, whereas high-coordination clusters are limited by reduced adaptability. This work provides a dynamic framework for understanding Cu/ZnO active sites and designing catalysts beyond static models.