Decoding intrinsic features of carbons from their capacitive performance

Abstract

Carbon-based materials, prized for their diverse applications, are at the forefront of scientific exploration. Performance of these materials relies on their basic properties, yet traditional characterization methods are often laborious. In our study, we introduce a machine learning (ML) based approach that leverages electrochemical performance data to elucidate the intrinsic properties of carbon materials. We highlight a technique, Cap2SSA, that expeditiously and precisely determines the specific surface area (SSA) of carbon materials through capacitive responses gathered during electrochemical tests. The efficacy of Cap2SSA is substantiated using the commercial carbon material across diverse current densities in both aqueous and organic electrolytes. Our findings reveal that the SSA values derived from capacitance measurements align closely with actual measurements, with discrepancies under 14 %. Significantly, this evaluation method extends beyond the SSA estimation, offering a versatile tool for inferring additional carbon material features, such as nitrogen and oxygen doping levels. Overall, combining electrochemical measurements with ML enables rapid assessment of material composition and structure, driving forward material science analysis.