First-Principles Predictions of Structure–Function Relationships of Graphene-Supported Platinum Nanoclusters

Abstract

Platinum-based materials play an important role as electrocatalysts in energy conversion technologies. Graphene-supported Pt nanoclusters were recently found to be promising electrocatalysts for fuel-cell applications due to their enhanced activity and tolerance to CO poisoning as well as their long-term stability toward sintering. However, structure–function relationships that underpin the improved performance of these catalysts are still not well understood. Here, we employ a combination of empirical potential simulations and density functional theory (DFT) calculations to investigate structure–function relationships of small Pt N ( N = 2–80) clusters on model carbon (graphene) supports. A bond-order empirical potential is employed within a genetic algorithm to go beyond local optimizations in obtaining minimum-energy structures of Pt N clusters on pristine as well as defective graphene supports. Point defects in graphene strongly anchor Pt clusters and also appreciably affect the morphologies of small clusters, which are characterized via various structural metrics such as the radius of gyration, average bond length, and average coordination number. A key finding from the structural analysis is that the fraction of potentially active surface sites in supported clusters is maximized for stable Pt clusters in the size range of 20–30 atoms, which provides a useful design criterion for optimal utilization of the precious metal. Through selected ab initio studies, we find a consistent trend for charge transfer from small Pt clusters to defective graphene supports resulting in the lowering of the cluster d-band center, which has implications for the overall activity and poisoning of the catalyst. The combination of a robust empirical potential-based genetic algorithm for structural optimization with ab initio calculations opens up avenues for systematic studies of supported catalyst clusters at much larger system sizes than are accessible to purely ab initio approaches.