Hierarchical porous N/S-doped carbon with machine learning to predict advanced potassium-ion batteries

Abstract

Potassium ion batteries (PIBs) have promising prospects for next-generation energy storage. However, the advanced anode materials needed for these systems are challenging because normal graphite cannot store the large-radius potassium ions. Here, we report a strategy to construct hierarchical porous sponge-like carbon. The resulting NS–C-1100 benefits from its unique pore structure and N/S-doping and offers a superior reversible capacity of 313.5 mA h g⁻¹ at 1 A g⁻¹ and long cycle stability with a capacity of 250 mA h g⁻¹ after 7000 cycles at 1 A g⁻¹. Kinetic and mechanistic studies of potassium storage attribute the excellent electrochemical performance to the high nitrogen/sulfur content, enlarged interlayer spacing, and abundant defects. Machine learning (ML) was then used applied to predict the relationship between different parameters and performance. The results showed further evidence of excellent performance. Density functional theory (DFT) calculations demonstrated that vacancy defects and N/S-doping can efficiently promote the adsorption of K⁺ and promote K storage. As expected, the as-assembled NS–C-1100//PB pouch cell showed excellent capacity retention after 200 cycles at 1 A g⁻¹. This rationally designed porous electrode has excellent performance and offers a new approach to energy storage.