Carbon-based supercapacitors have emerged as promising energy storage components for renewable energy applications due to the unique combination of various physicochemical characteristics in porous carbon materials (PCMs) that can improve specific capacitance (SC) properties. It is essential to develop a methodical approach that exploits the synergy of these effects in PCMs to achieve superior capacitance performance. In this study, machine learning (ML) provided a clear direction for experiments in the screening of key physicochemical features; SHapley Additive exPlanations analysis on ML indicated that specific surface area and specific doping species had a significant synergistic impact on SC enhancement. Utilizing these insights, an O, N co-doped hierarchical porous carbon (ONPC-900) was synthesized using a synergistic pyrolysis strategy through K2CO3-assisted in-situ thermal exfoliation and nanopore generation. This method leverages the role of carbon nitride (graphite-phase carbon nitride) as an in-situ layer-stacked template and the oxygen (O)-rich properties of the pre-treated lignite, enabling controlled synthesis of graphene-like folded and amorphous hybrid structures engineered for the efficient N and O doping sites and high specific surface area, resulting in an electrode material with enhanced structural adaptability, rapid charge transfer, and diffusion mass transfer capacity. Density functional theory (DFT) calculations further confirmed that pyrrole nitrogen (N-5), carboxyl (-COOH) active sites, and the defect structure formed by pores synergically enhanced the adsorption of electrolyte ions (K+) and electron transfer, improving the SC performance. The optimized ONPC-900 electrode exhibited impressive SC properties of 440 F g-1 (0.5 A g-1), outperforming most coal-based PCMs. This study provides a methodology for designing and synthesizing high SC electrode materials by optimizing the key characteristic parameters of synergism from complex structure-activity relationships through the combination of ML screening, experimental synthesis, and density functional theory validation.
Read full abstract