In silico design of copper-based alloys for ammonia synthesis from nitric oxide reduction accelerated by machine learning

Abstract

The NO electroreduction reaction (NORR) has been recognized as a promising strategy for NO removal and NH₃ synthesis, while current NORR electrocatalysts suffer from limited activity and selectivity. Here, we comprehensively investigate the NORR performance of copper alloys by virtue of first-principles calculations and machine learning (ML). It is identified that the adsorption energy of N atoms E_ads(*N) is an effective catalytic descriptor for the NORR. As a result of screening 140 copper alloys, we discover Cu@Cu₃Ni and Cu₂Ni₂@Cu₃Ni with extremely low limiting potentials and reasonably low kinetic barriers. Then, we construct a highly accurate ML model for predicting the E_ads(*N) and clarify the local elemental features as critical factors. By predicting the E_ads(*N) on ∼2 000 000 bimetallic alloy surfaces, we reveal that Ni is the optimal alloy non-noble-metal element to enhance the NORR activity. Our work not only opens a new avenue for the design of efficient alloy catalysts but also paves the way toward the ML-accelerated discovery of novel NORR catalysts.