Unraveling electronic structure modulation mechanism in cobalt spinel Fenton-like catalysis by integrating density...

Heterogeneous Fenton-like catalysis holds significant promise for environmental remediation, yet conventional trial-and-error methodologies fail to capture the intrinsic structure-activity relationships in catalyst design. Herein, we present an integrated density functional theory (DFT) and machine learning (ML) framework for the rational design of Co spinel catalysts, emphasizing the dominant role of electronic structure modulation. After rigorous comparison of eight ML algorithms, XGBoost achieved the highest predictive accuracy (R² = 0.874, RMSE = 0.190). SHAP analysis and feature grouping revealed B-site d-band center and B-site covalency as key descriptors. Crystal field theory coupled with partial dependence plots elucidated the modulation mechanism of covalency competition and d-band center. The resulting DFT-ML strategy predicted peracetic acid activation efficiencies across 12 normal Co spinels, which provided fundamental insights into bimetallic synergy and elemental doping effects. Experimental validation identified two unreported advanced catalysts including ZnCo₂O₄ and LiCoO₂, confirming the prediction reliability. Moreover, the novel electronic synergy factor was established to quantitatively bridge atomic-scale descriptors with macroscopic kinetics. This study delivers a generalizable and interpretable paradigm for data-driven catalyst development, accelerating catalyst discovery cycles and reducing design costs.