Data-driven discovery of high-performance zinc-ion battery cathodes by a machine learning strategy integrating energy gradient and activation area ratio

Abstract The heteroatom doping is considered a promising strategy for enhancing the performance of the MnO2-based electrode materials for zinc-ion battery (ZIB). However, quickly discovering the high-performance doped-MnO2 remains significant challenge to simultaneously give consideration to both the various metal types, doping concentration, and the essential screening mechanism. Herein, a novel research paradigm is developed by combining machine learning predictions with systematic experiments and theoretical calculations for solving this issue. The results simulated by machine learning from the two-dimensional perspective reveal that only when Co species are introduced into δ-MnO2 can zinc ions (Zn2+) maintain the smaller binding energy gradient distribution and larger activation area ratio among the constructed various doping system database, further achieving qualitative “structure–activity” descriptor. Moreover, the density functional theory (DFT) calculations systematically unveil optimal adsorption energy/Gibbs free energy, higher negative integral crystal orbital Hamilton population (−ICOHP) (0.0125 Ha), and lower Zn2+ diffusion barrier (0.978 eV) for moderate Co-doped δ-MnO2 with oxygen vacancy (Co(M)-δ-MnO2−x, where (M) denotes moderate Co-doping concentration) compared with the other samples, which can preserve the Zn2+ adsorption/desorption equilibrium and the structure integration, and accelerate the reaction kinetics. Benefiting from these advantages, the obtained ZIB using the optimized cathode can present the large specific capacity of 655.7 mAh·g−1 at 0.5 A·g−1 and high rate capability (209.8 mAh·g−1 at 20 A·g−1), which is far higher than those of the other compound cathode materials. This study offers new insights for the design and optimization of doped-δ-MnO2 cathodes in ZIBs, and the obtained universal theoretical guidance is also suitable for constructing other high-performance layered electrode materials.