Machine-Learning and Atomic-Scale Mechanistic Insights for Designing Gradient Porous MOF-Derived Carbon Electrodes.

MOF-derived heteroatom-doped porous carbons hold strong potential for supercapacitor electrodes, yet their optimization is hindered by complex coupling among pore geometry, composition, and electrochemical behavior. To clarify these structure-performance relationships, we establish a predictive design strategy by integrating a data-driven machine-learning (ML) framework with density functional theory. A hierarchical ensemble model combining Gradient Boosting and Gaussian Process Regression achieves high predictive accuracy (test R2 = 0.99) and strong noise tolerance. ML analysis reveals that capacitance enhancement originates from a coordinated micro/mesopore architecture. Micropores of ∼1.2 nm coupled with mesopores of ∼2.8 nm create an optimal regime that balances charge storage and ion-transport kinetics. Guided by this insight, an MOF-derived O and Co codoped gradient pore model is constructed to probe atomic-scale mechanisms in a neutral KCl electrolyte. Charge-density analysis shows that C-O-Co hybridization and oxygen-induced electron-rich sites enhance interfacial polarization and ion-electrode interactions. Molecular dynamics simulations demonstrate that electrostatic confinement in micropores, together with chemisorption-assisted charge transfer in mesopores, generates a continuous adsorption-energy gradient, accelerating ion migration and improving both storage density and transport kinetics. Overall, this work provides an interpretable framework for designing next-generation energy-storage electrodes with optimized pore architectures.