Machine learning-assisted discovery of outside-in structure Ni-rich cathode with high performance

Ni-rich oxides have emerged as leading cathode candidates for lithium-ion batteries because of high specific energy, lower cost, and improved sustainability compared to cobalt-based materials. However, Ni-rich cathodes suffer from voltage and capacity degradation driven by anisotropic lattice strain and interfacial reconstruction. Here, we report a high-performance Ni-rich cathode featuring a robust outside-in architecture, achieved via machine learning-assisted identification of Al³⁺ and Sn⁴⁺ dopants. Through a competitive doping mechanism, these dopants form a Sn-rich rock-salt surface layer and a uniformly Al-doped bulk. This high-quality outside-in structure enhances interfacial stability and structural reversibility by mitigating cathode/electrolyte interfacial degradation and alleviating anisotropic lattice strain associated with H2/H3 phase coexistence. Moreover, nonmagnetic Al³⁺ and Sn⁴⁺ weaken superexchange interactions and suppress Li─Ni disorder. As a result, the cathode retains 96.9% of its capacity after 200 cycles with minimal voltage fade. These findings provide insights into the development of high-performance Ni-rich cathodes.