Probing the structure–property relationships of supported copper oxide nanoclusters for methane activation

Supported metal oxide nanoclusters (MeO-NCs) have gained significant attention for their remarkable versatility in various energy and sustainability applications. Despite rapid advancements in atomic-scale synthesis and characterization techniques, the rational design of MeO-NCs with desired catalytic properties remains challenging. This challenge arises from the elusive and difficult-to-quantify structure-catalytic property relationships, particularly in the case of amorphous nanoclusters. Exploiting first-principles calculations at the density functional theory (DFT) level, we conducted a systematic investigation into the growth, geometries, and catalytic performance of a series of tetra-copper oxide nanoclusters (Cu4O-NCs) for methane activation. Focusing on the most representative geometries, we applied machine learning to extract two physically insightful descriptors involving the spin density, the p-band center of the oxygen site, and the d-band center of adjacent Cu sites. These descriptors enable us to predict free energy barriers associated with both the homolytic and heterolytic mechanisms of methane activation. This descriptor-driven approach enables rapid and intuitive prediction of the preferred reaction mechanism. Our findings lay a solid foundation for future advancements in catalysts based on amorphous nanoclusters and provide valuable insights into the mechanistic landscape of methane activation.