所属：Osaka University, Graduate School of Engineering

概要：Platinum (Pt) supported on graphene is an established catalyst for CO oxidation, oxygen reduction, and other fuel cell reactions, but the cost and rarity of platinum necessitates further research into reducing Pt loading. The single-atom catalyst form stabilized on a support, namely, graphene, maximizes catalyst surface area and minimizes Pt usage, but faces issues related to low reactivity and instability due to sintering or poisoning.
To improve single-atom platinum supported on graphene (SAC-Pt-G), this study applies machine learning to determine novel structures stabilized on the edge of nanoflake, armchair nanoribbon (AGNR), and zigzag nanoribbon (ZGNR) graphene scaffolds. Using the Global Optimization with First-principles Energy Expression (GOFEE) method, a thorough yet rapid structure search was accomplished on each of the scaffolds to determine the most stable structures at several different combinations of carbon terminations and hydrogen concentrations. The behavior of nitrogen (N) dopants was also examined by the addition of 1 to 2 N atoms to the structure search.
The structure search confirms the preferential edge adsorption of Pt encapsulated in the graphene rings. The most stable structures were evaluated using density functional theory calculations to assess the stability and adsorption behavior of various reaction intermediates. The structures identified outperformed pure Pt(111) in OH adsorption while minimizing CO adsorption, while the ZGNR systems produced the most stable structures compared to those of AGNR and flake systems.