Electrochemical ammonia synthesis over copper oxide derived catalysts studied by electric field dependent machine learning potential

Recently, the electrocatalytic nitrate reduction to ammonia (eNO 3 RR) has become attractive as an alternative route for the green synthesis of ammonia at ambient conditions. However, the catalytic activity and selectivity of this process at low overpotentials is still low. Cu-based catalysts exhibit the best performances for eNO 3 RR among all catalysts. Moreover, copper oxide catalysts, which can undergo reduction during eNO 3 RR, display varied catalytic performances and facet-dependent behaviors. To elucidate the structural evolution and catalytic behavior of copper oxide electrodes, we developed an electric field-dependent equivariant machine learning potential (MLP) to simulate the evolution of electrode surfaces under electroreduction conditions. Grand Canonical Monte Carlo (GCMC) simulations were conducted to simulate the reduction of different copper oxide surfaces in reaction conditions. It was found the reduced surfaces from different oxide surfaces have different proportions of 3-fold copper and 4-fold copper active sites. Following that, the reaction mechanism of these active sites was addressed. The defective 3-fold copper sites show the best performance, indicating the Cu 2 O(111) surface, which can be selectively reduced into a 3-fold-copper dominated surface at reaction conditions, should have the best catalysis performance towards eNO 3 RR.