Unified electronic-geometric descriptor deciphers peroxymonosulfate activation using Fe-based dual-atom catalysts

The rational design of high-efficiency Fenton-like catalysts remains hindered by insufficient understanding of electronic-geometric synergy in peroxymonosulfate (PMS) activation. We transcend classical d-band theory by proposing a machine learning-decoded binary descriptor (BD) unifying orbital electronic structure (IOES) and orbital geometric structure (IOGS) indices to predict PMS activation pathways across diverse coordination environments. This BD framework quantifies antibonding orbital occupancy (via d-p hybridization) and geometric constraints (interatomic distance/O-H elongation), enabling precise screening of Fe-based dual-atom catalysts (DACs). Among FeM DACs (M = Ti, V, Cr, Mn, Fe, Co, Ni, Cu), FeMn DACs with optimal BD values (IOES = 0.86, IOGS = 0.40) achieved 94.2% 1O2 yield and fast kinetics (kobs = 1.2 min⁻1) toward sulfadiazine degradation. Crucially, a flow-through reactor demonstrated >90% pollutant removal for 30 days at industrial flux (122.3 L m⁻2 h⁻1). This work establishes universal orbital-level design principles for sustainable water remediation, bridging atomic-scale insights to engineering-scale implementation. This study develops a unified electronic-geometric binary machine-learning descriptor to design efficient dual-atom catalysts for water purification. The optimized catalyst achieves rapid and persistent pollutant removal for 30 days.