|
Machine learning-driven high entropy alloy catalyst discovery to circumvent the scaling relation for CO2 reduction reaction
Z.-W. Chen, Z. Gariepy, L. Chen, X. Yao, A. Anand, S.-J. Liu, C. Feugmo, I. Tamblyn, C. Veer Singh ACS Catalysis, 12, 24, 14864-14871 (2022) Discovering efficient electrocatalysts for the CO2 reduction reaction (CO2RR) is critical for sustainable energy, but progress is hindered by linear scaling relations that constrain catalytic performance on conventional metal surfaces. In this work, we use machine learning models trained on 1280 adsorption site calculations to predict adsorption energies of key intermediates (COOH*, CO*, and CHO*) on high-entropy alloy (HEA) surfaces. Using density functional theory combined with these ML predictions, we design a FeCoNiCuMo HEA system that circumvents the traditional scaling relation between CO* and CHO* adsorption energies. The rotation of COOH* and CHO* adsorbates on the compositionally diverse HEA surface breaks the scaling constraint, resulting in outstanding predicted catalytic activity for CO2RR with a limiting potential of only 0.29-0.51 V, significantly lower than conventional catalysts. |
