硕士生王金丹、陈坚强论文"Coupling High-Throughput DFT Calculation and Machine Learning for Dopant Classification in Nickel-Based Layered Cathodes"被Energy Storage Materials接收发表
发布时间：2025-11-11
点击次数：
Nickel-based layered cathode materials offer high specific capacity and low cost, making them potential candidates for lithium-ion batteries. However, their practical application is limited by rapid capacity fading due to microcrack formation, cation mixing, and gas release during cycling, especially at high voltages. To address this, we employed high-throughput DFT calculations combined with machine learning clustering to explore doping strategies for LiNiO2. Among 62 screened elements, 56 thermodynamically stable dopants were evaluated using substitution doping formation energy and binding energy. These dopants were clustered into seven groups using the ROCK algorithm with five structural descriptors, revealing diverse effects on the LiNiO2 matrix. To validate the computational results, Sr- and Mo-doped LiNiO2 samples were synthesized. Both single-doped and co-doped samples showed improved structural stability and electrochemical performance. Particularly, the Sr, Mo co-doped LNO material exhibited a discharge specific capacity of 192.2 mAh g-1 with a capacity retention of 90.8% after 100 cycles at 1C under 25 °C, effectively suppressing the H2→H3 phase transition, Li/Ni disorder, and oxygen release. This study demonstrates a reliable high-throughput approach for dopant screening and provides guidance for designing multi-doped Ni-rich cathodes with high performance.