Single-atom dissolution at the MN4/MXene interface and electric field-driven adsorption mechanisms: Unraveling catalytic descriptors using machine learning

Single-atom catalysts (SACs) have attracted much attention in the field of energy and environment due to their tunable activity. Here, we present a study of MXene(M₂NO) carriers to modulate the catalytic activity and reactive metal dissolution of SAC at the M−N−C(MN₄) interface. The density functional theory (DFT) results indicated that M₂NO provides axial traction for MN₄, weakening the interaction of M−3d orbitals with *O/*OH/*OOH-p orbitals in MN₄, forming higher π* and π orbitals, and lowering the ORR or OER overpotential of FeN₄/M₂NO and CoN₄/M₂NO (Ti, V, Cr, Nb and Ta). Pourbaix diagrams were constructed based on thermodynamic cycling and DFT, and the results showed that FeN₄/M₂NO (M = Ti, V, Cr, Nb and Ta) had lower metal dissolution. And with the addition of 0.2 V/Ang interfacial electric field, FeN₄/Ti₂NO exhibits excellent OER (0.28 V) overpotential. The accuracy of the constructed simple descriptors of ORR and OER catalytic performance: EA₁ × EA₂ - M₁ and M₁ - a - b was demonstrated by four machine learning methods and symbolic regression algorithms. This study reveals the law of the influence of M₂NO on the catalytic performance of the MN₄ interface and improves new ideas for the design of high-performance SACs.