Sculpting Superior Subnanometer Catalysts Directly from Inert Gold

Bulk gold, renowned for its catalytic inertness, can be transformed into an exceptionally active catalyst by engineering undercoordinated sites on its surface. However, directly sculpting such active sites from extended terraces remains a fundamental challenge. Here, we demonstrate a programmable strategy to dynamically generate subnanometer Au clusters directly from inert Au(111) surfaces, unlocking superior activity for CO oxidation. By integrating large-scale machine learning molecular dynamics with density functional theory, we decode the atomistic pathway of this restructuring process. Our approach employs thermal-CO pressure cycles to induce the ejection of step-edge atoms, forming mobile Au-CO complexes. Subsequent cooling kinetically traps these complexes into metastable, subnanometer clusters (3-6 atoms). A controlled reduction of CO exposure then precisely exposes the catalytically crucial undercoordinated sites while maintaining cluster stability. Crucially, these sculpted clusters exhibit a CO oxidation activity that far exceeds that of pristine step edges, terraces, or conventional single-atom sites. This work establishes reaction condition engineering as a powerful paradigm for sculpting active catalysts directly from bulk materials by bypassing traditional synthetic routes.