The electrochemical reduction of nitrogen to ammonia represents a promising route for sustainable ammonia synthesis, crucial for fertilizer production and energy storage. In this study, we employ a combined computational approach integrating density functional theory (DFT) calculations and machine learning (ML) techniques to screen the electrochemical catalytic activity of doped transition metal dichalcogenides (TMDs) for the nitrogen reduction reaction (NRR). Using DFT calculations, we generate a dataset containing the binding energies of nitrogen and intermediates of NRR on each catalytic material as target features. Subsequently, we employ various ML algorithms, including Random Forest (RF), Support Vector Machine for Regression (SVR), eXtreme Gradient Boost (XGBoost) and Recurrent Neural Network (RNN), to construct predictive models. These models enable us to anticipate the catalytic activity of new 2D TMD-based materials within seconds. TMDs have garnered significant interest due to their tunable electronic properties and high surface-to-volume ratio. Through systematic computational simulations, we investigate the effects of various dopants on the catalytic activity, aiming to enhance the NRR efficiency. Our results unveil the crucial role of dopants in modulating the binding strength of N2 and intermediates, thus influencing the reaction kinetics. Additionally, we explore the electronic structure and charge transfer mechanisms to elucidate the underlying principles governing the NRR performance of doped TMDs. This computational screening provides valuable insights for the design and optimization of efficient electrochemical catalysts for sustainable ammonia production.
