Bridging Oxide Thermodynamics and Site-Blocking: A Computational Study of ORR Activity on Platinum Nanoparticles

The Oxygen Reduction Reaction (ORR) is a key reaction in fuel cells and metal-air batteries involving multiple intermediates and parallel pathways. Unfortunately, the overpotential associated with the kinetic of the process is a significant barrier to the efficiency of these devices. Despite the substantial body of work dedicated to this problem, a unified picture that seamlessly integrates thermodynamic and other kinetic aspects of the ORR remains elusive. In this work, we perform grand-canonical minimization Monte Carlo simulations using the recent MACE-MP-0 forcefield on an experimentally reconstructed Pt nanoparticle. From the Monte Carlo simulations, we extract evenly spaced oxidized configurations to perform an in-depth study with varying oxygen content. Our results show that the thermodynamic of the system is highly dependent on the oxygen ratio. By using the computational hydrogen electrode scheme we are able to link various quantities to the potential and to predict oxidation state vs. potential. We computationally identify the place-exchange mechanism occurring at 1.06 V vs. SHE, close to the onset potential of the ORR. By using well-established scaling relations we are able to predict the rate determining step on the oxide-covered nanoparticle to be the initial reduction of O2 to OOH*. We explore the possibility of distinguishing between blocking and reactive sites by performing a deletion energy analysis. We find a fraction of oxide sites to be readily available for reduction, which decreases as the potential increases. Finally, we show the possibility of building a simplified kinetic model with parameters directly derived from our simulations. At all levels, our results are compared with the available literature and showed to be in reasonable agreement with both theoretical and experimental studies. Despite partial disagreement with DFT, the MACE-MP-0 forcefield is able to predict the general trends of the system, showing the potential of such foundation models in the field of computational electrochemistry without fine-tuning. Our work is a step towards the understanding of the various aspects causing the overpotential of the ORR through atomistic modelling.