Optimized selectivity in CO2 electrochemical reduction using amorphous CuNi catalysts: insights from density functional theory and machine learning simulations

Amorphous materials represent a promising platform for advancing CO2 electrochemical reduction due to their inherently diverse coordination environments. In this study, we demonstrate computationally the superior performance of amorphous CuNi alloys for CO2 electrochemical reduction. By integrating machine learning forcefields for efficient structure generation and density functional theory for subsequent structural refinement and property calculations, we reveal the potential of these disordered systems to outperform their crystalline counterparts. Machine learning forcefields can generate a bulk structure containing a mixture of Cu and Ni atoms, resulting in enhanced catalytic performance. Effective screening of the amorphous surfaces is used to identify undercoordinated Cu and Ni sites in the amorphous structure to synergistically promote selective CO production and favor ethanol formation over ethylene via the stabilization of the *COCHO intermediate, resulting in significantly lower Gibbs free energy changes compared to the crystalline counterpart. The varying atomic coordination environments on amorphous surfaces promote both C–C bond formation and subsequent proton-electron transfer, leading to ethanol formation. These findings demonstrate the superior catalytic performance of amorphous CuNi, highlighting its potential for efficient and selective electroreduction of CO2.