Unraveling Structure and Spectroscopic Signatures of Pt/γ-Al2O3 via an Artificial Intelligence-Enabled Simulation Framework

Supported subnanometer clusters have attracted significant attention in catalysis. Yet, it is challenging to resolve their structural evolution, particularly in the presence of adsorbates. Here, we present an artificial intelligence (AI)-enabled simulation framework that integrates machine learning force field (MLFF) construction, molecular dynamics (MD) simulations, and ML-assisted infrared (IR) and X-ray absorption near-edge structure (XANES) spectra analysis to investigate the restructuring of supported subnanometer clusters. Using Pt10/γ-Al2O3(110) as a model system, we explored the CO-induced restructuring of the cluster under different conditions. The simultaneous machine learning (ML)-spectra elucidate a strong interplay among adsorbate coverage, cluster restructuring, and cluster-support interaction. This work sheds light on the complex dynamics of supported subnanometer clusters that is extendible to other systems.