Physics-Driven 3D Structural Prediction and Transport Kinetics of Porous Carbons via Random Field and Phase-Field Evolution.
The complex pore topology of hierarchical porous carbons constrains energy transport, yet conventional research relies on homogenized scalar descriptions, treating pore evolution as a black box. To overcome 3D characterization limits, this study establishes an integrated R&D framework coupling bidirectional performance prediction with physics-driven 3D structural prediction. A high-precision bidirectional mapping model (R2 = 0.8595) was constructed using CatBoost and differential evolution (DE) to enable target-oriented inverse optimization. In the structural dimension, we developed a synergistic algorithm combining Gaussian random fields (GRF) and Cahn-Hilliard (C─H) phase-field dynamics to dynamically predict authentic 3D topologies by simulating interfacial energy-driven pore evolution. Findings reveal that global connectivity is achieved at a total porosity of 0.46, with specific critical thresholds of 0.15, 0.25, and 0.35 for micro-, meso-, and macropores, respectively. By integrating particle tracking, the study identifies transport hotspots contributing 80% of the total flux and calibrates a non-Darcy kinetic scaling law (exponent n = 1.6357), highlighting mesopores' role in alleviating kinetic bottlenecks. Experimental validation confirms that the inverse-optimized conditions accurately meet performance targets (error 3.2%-7.4%), while the 3D structural prediction model achieves high-fidelity restoration of experimental morphologies. This work provides a robust physics-driven paradigm for transitioning from empirical trial-and-error to intelligent, target-oriented 3D structural prediction and design.