Design and screening of a NORR electrocatalyst with co-coordinating active centers of the support and coordination atoms: a machine learning descriptor for quantifying eigen properties

Abstract

The utilization and dispersibility of single-atom catalysts make them widely employed in electrocatalysis. A series of SACs are designed to investigate their NO reduction capabilities and properties, and high throughput DFT calculations and machine learning are used for activity prediction and screening. Based on the calculated activity, selectivity, and stability, MoS₂_N₃_Zn, MoS₂_N₃_Cd, and Mo_N₃_Zr are considered as excellent performing SACs. Electronic structure analysis showed that a self-regulation mechanism of the coordination environment coordinated by the support and N atom resulted in excellent NORR activity of the three SACs, and the NORR tends to have no thermodynamic barrier. Linear descriptors for NO adsorption energy, bond length, and charge transfer are constructed through constructing quantitative formulae. Additionally, novel volcanic descriptors for adsorption energy (*NHO) and ΔG_max are also developed. The descriptors show that a good description of catalyst performance can be achieved by quantifying the intrinsic properties of the catalyst and atoms. These descriptors serve as inputs under various machine learning methods, among which GBR has proven effective in predicting and classifying SAC performance. The machine learning strategies presented in this study can significantly reduce computation time while providing valuable insights for high-throughput screening of SACs.