Bio-derived N/P co-doped activated carbons for supercapacitor: AI-assisted experimental study

The demand for energy sustainability has driven interest in biomass-derived carbon electrode materials for electrochemical energy storage. Heteroatoms-rich porous activated carbon (AC) electrode materials can be inherently synthesized via a bio-self doping procedure to promote pseudo-capacitances in supercapacitors. In this study, soybean hull waste (SBW) materials were pyrolyzed and activated through an integrated fast pyrolysis and KOH chemical activation procedure. The ideal SBW-derived AC (SBW-AC-3) exhibited a sponge-like hierarchical porous structure with micropores ratio of 80% of total porosity. The SBW-AC-3, prepared using the highest KOH activation ratio, showed specific capacitance of 234.52F/g at 2 A/g and rate capability of 70% at 20 A/g. Dendrogram cluster correlation and kernel density estimator revealed relationships among physicochemical properties and specific capacitance. Multivariate Gaussian sampling was used to generate an experimental-based synthetic database for electrochemistry-informed artificial intelligence (AI). The physics-informed neural networks (PINNs) model integrated physicochemical descriptors with the conventional capacitance equation, achieving a high predictive accuracy with R 2 of 0.989. This AI-assisted framework supports the smart selection of biomass precursors for designing functionalized porous ACs as sustainable electrode materials in supercapacitor.