Abstract
The limited energy density inherent in cathode materials remains a marked barrier to the widespread adoption of sodium-ion batteries. Despite considerable research efforts, the precise influence of atomic and crystalline configurations on energy density is not yet fully understood, creating a knowledge gap that hinders the rational design of advanced cathode materials. In this study, we propose a machine learning approach to systematically identify promising cathode materials with high energy densities. Our model highlights the critical roles of entropy and equivalent electronegativity, among other properties such as molecular mass, electron affinity, and average ionic radius. Based on these insights, we successfully synthesized Na(3)Mn(0.5)V(0.5)Ti(0.5)Zr(0.5)(PO(4))(3) (NMVTZP) electrodes via a sol-gel method. The resulting electrodes exhibit an impressive reversible specific capacity of 148.27 mAh g(-1) at a 0.1-C rate, outperforming several previously reported cathode materials. Additionally, the NMVTZP electrodes demonstrate an average operating voltage of 3.14 V, an energy density of 465 Wh kg(-1), and exceptional rate performance, retaining 90.20 mAh g(-1) at a 5-C rate. We anticipate that our machine learning approach will accelerate the development of high-performance materials and greatly contribute to the advancement of sodium-ion battery technology.