Machine learning-optimized bimetallic MOF/MXene composite with improved supercapacitor performance

Abstract

A wide range of bimetallic MOF-based composites integrated with different conductive materials, such as graphene, carbon, and MXenes, is explored to overcome the intrinsically poor conductivity and sluggish charge transport of pristine MOFs. However, their electrochemical performance is highly dependent on the mixing ratios and interfacial compatibility, requiring precise optimization to maximize the active sites and charge flow. To address this challenge, herein, we employed machine learning (ML) to optimize the ratios of Co, Zn, and MXene, as the key synthesis parameters, and identified an optimal CoZn-ZIF@MXene composition. The resulting ML-optimized CoZn-ZIF@MXene@NF delivered outstanding supercapacitive performance with a specific capacitance of 622.95 F g⁻¹ at a current density of 1 A g⁻¹, significantly outperforming the monometallic ZIF-67@MXene (235.62 F g⁻¹) and ZIF-8@MXene (196.46 F g⁻¹). Furthermore, it exhibited a high energy density of 55.37 Wh kg⁻¹ at a power density of 354.8 W kg⁻¹, which can be attributed to its hierarchical micro-/meso-porous architecture that facilitates rapid ion diffusion and efficient charge transport. This study highlights the potential of data-driven ML approaches to enable the rational design of next-generation electrode materials for high-performance supercapacitors.