Work

Machine learning-driven regulation of the four-electron LiOH pathway in lithium–oxygen batteries with prolonged cycling

Year: 2026
Type: article
Abstract: The practical deployment of high-energy-density lithium–oxygen batteries is fundamentally restricted by sluggish oxygen redox kinetics and the instability of discharge intermediates. Here, we report an artificial intelligence-integrated framework leveraging interpretable machine learning and density functional theory to navigate a vast chemical space of thousands of active sites. This approach identifies Ni-Co dual-atom catalysts as a better configuration among the selected elements for orchestrating multi-electron transfer. Theoretical analysis reveals that synergistic electronic coupling between the dual-sites fundamentally reconfigures the d -band center, thereby lowering the thermodynamic barriers for oxygen redox. Moreover, the catalysts enable precise regulation of the reaction pathway, inducing a controllable transition from the conventional two-electron pathway to a four-electron pathway. This shift yields a discharge capacity of ~60,000 mAh g −1 . Beyond its role as a positive electrode catalyst, this architecture functions as a multifunctional interfacial protective layer that stabilizes the lithium metal negative electrode by suppressing dendrite growth. Consequently, the lithium–oxygen system achieves stable lifespan of 2075 cycles under the four-electron pathway with reduced voltage decay. This work establishes a data-guided strategy for the rational design of atomic-scale architecture via artificial intelligence, unlocking the potential of multi-electron oxygen chemistry for lithium–oxygen batteries.
Language: English
Open Access status: gold
Estimated APC paid: $6,990
OpenAlex ID: W7236337632
FWCI:
Cites: 35
Cited by:
Related to: 10