An emerging machine learning strategy for the assisted-design of high-performance supercapacitor materials by mining the relationship between capacitance and structural features of porous carbon

How to design high-performance materials by mining the relationship between properties and structure features of materials is a major challenge today. We developed a new strategy for the assisted-design of high-performance supercapacitor materials by mining the relationship between capacitance and structural features of porous carbon materials (PCMs) using machine learning (ML) on the basis of hundreds of experimental data in the literature. Six ML models were selected to predict capacitance with the closely related structural features of PCMs. XGBoost demonstrates best predictive performance of supercapacitor (R = 0.892) among all ML models. The accurate predicted ability of the developed models could significantly reduce experiment workload for the assisted-design of high-performance supercapacitor materials. Smicro/SSA, SSA, and PS provided more contribution to the capacitive performance among all porous structural features. The overall results of this study will provide a new idea for design high-performance materials by mining the relationship between properties and structure features of materials using an emerging ML strategy.

Automated summary

The research team used machine learning to explore the relationship between the structural features and capacitance of porous carbon materials, providing a new approach for the assisted design of high-performance supercapacitor materials. The XGBoost model demonstrated the best predictive performance among all models, and the developed models' accurate predictive ability could significantly reduce experimental workload, with Smicro/SSA, SSA, and PS contributing the most to capacitive performance.