Neural Network Potentials for Accelerated Metadynamics of Oxygen Reduction Kinetics at Au-Water Interfaces

The application of ab-initio molecular dynamics (AIMD) for the explicit modeling of reactions at the solid- liquid interfaces can provide new understandings towards the reaction mechanisms. However, prohibitive computational cost severely restricts the time- and length-scale of AIMD. Equiv- ariant graph neural network (GNN) based accurate surrogate potentials can accelerate the speed of performing molecular dynamics after learning on representative structures in a data efficient manner. In this study, we combined uncertainty- aware GNN potentials and enhanced sampling to investigate the reactive process of the oxygen reduction reaction (ORR) at Au(100)-water interface. By using a well-established active learning framework based on CUR matrix decomposition, we can evenly sample equilibrium structures from MD simulations and non-equilibrium reaction intermediates that are rarely vis- ited during the reaction. The trained GNNs have shown excep- tional performance in terms of force prediction accuracy, the ability to reproduce structural properties, and the low uncer- tainties when performing MD and metadynamics simulations. Furthermore, the collective variables employed in this work en- ables an automatic search of reaction pathways and provide a detailed understanding towards the ORR reaction mechanism on Au(100). Our simulations identify an associative reaction mechanism where adsorbed O2 reacts with water to form hy- droxyls through an *OOH transition state. The reaction pro- ceeds without formation of *O with a low reaction barrier of 0.3 eV. The low barrier agrees with the fast reaction kinetics observed experimentally. The methodology employed in this study can pave the way for modeling complex chemical reac- tions at electrochemical interfaces with an explicit solvent at ambient conditions.