Understanding ORR activity of nanoclusters electrocatalysts: combined DFT and machine learning approach for multiscale modelling

Abstract

Unravelling the nature of ORR interaction with catalyst surfaces under reaction conditions is essential for designing next-generation electrocatalysts. In this work, we explore the coverage-dependent adsorption behavior of these intermediates on graphene-supported platinum subnano clusters (Ptn,n=7−13) using spin-polarized DFT and ab initio thermodynamic analysis. Our study uncovers a delicate balance of lateral interactions both attractive and repulsive that influence adsorption strength as surface coverage increases. By calculating differential average adsorption energies across various coverage scenarios, we reveal non-monotonic trends in stability that highlight the complex thermodynamic landscape of subnanometer clusters, demonstrating the pivotal role of multi-site interactions and structural fluxionality in catalytic performance. When combined with machine learning models built on geometric and electronic descriptors, our approach provides an effective strategy to predict active sites and break free from traditional scaling limitations.