Mechanism-aware temperature-resolved machine learning and experimental validation for optimization of Cu-based bimetallic catalysts in NO reduction by CO.
Efficient removal of NO, a major hazardous air pollutant, remains a critical challenge in industrial emission control. Machine learning enables the rational design and optimization of catalysts targeting NO reduction; however, conventional models trained across broad temperature ranges often fail to capture temperature-dependent variations in catalytic performance. Herein, a temperature-resolved machine learning (TRML) strategy is developed to enable the rational design of bimetallic catalysts for NO reduction by CO across a wide temperature range. The result indicates a distinct activity-determining factors in different temperature regimes. At low-medium temperatures, the dominance of the weighted average ionization energy (IE) suggests kinetics-controlled behaviour, whereas at medium-high temperatures, the increasing importance of the weighted average valence electron number (N) indicates a shift toward adsorption-controlled mechanisms. Based on the identified atomic-scale descriptors, a two-step screening framework was constructed to identify optimal promoter metal from a vast materials space, through which the Cu-Cr system emerged as the optimal bimetallic catalyst. Experimental results show that the Cu-Cr/Al2O3 catalyst achieves 99.7% NO conversion at 325 °C and exhibits superior performance over a wide temperature window. The superior performance of the Cu-Cr system originates from bimetallic synergy, consistent with TRML descriptors, where at low-medium temperatures, Cu-Cr synergistic interaction enhances redox capability and lowers reaction barriers, consistent with a kinetics-controlled catalytic regime. At medium-high temperatures, strengthened CO adsorption intensifies competitive adsorption, consistent with an adsorption-controlled catalytic regime. This integrated TRML framework provides a reliable strategy for the prediction and optimization of high-performance NO reduction catalysts, advancing practical hazardous gas mitigation.