Design of Catalysts for Electrochemical Nitric Oxide Reduction to Ammonia Based on Stacked Ensemble Learning

• • Stacked-SVR model achieved R² of 0.9223 and RMSE of 0.0608 for predicting ammonia yield (YNH3), enabling high-throughput catalyst screening with reduced experimental burden. • • Stacked-RF model achieved R² of 0.9042 and RMSE of 0.0900 for predicting Faradaic efficiency (FNH3), providing reliable performance estimates for process optimization. • • SHAP analysis identified Cu content as the most influential descriptor, guiding catalyst composition design toward Cu-based systems for enhanced NORR activity. • • Optimal system configuration: wet chemical reduction synthesis, carbon fiber (CF) substrate, HCl electrolyte, moderately lower working potential, low-to-medium electrolyte volume, reduced catalyst loading, and increased electrolyte concentration—collectively improve both YNH3 and FNH3.