Machine-learning-assisted discovery of a stable Li₃As₂ intermediate phase in the Li-As binary system and its...

Area of Science:

• Materials Science
• Computational Chemistry
• Electrochemistry

Background:

• Alloy-type anodes offer high capacity but suffer from volume expansion and instability during deep lithiation.
• Identifying stable intermediate phases with moderate potentials is crucial for mitigating these issues.

Purpose of the Study:

• To explore phase evolution in the Li-rich Li-As system and identify stable intermediate phases for improved anode performance.
• To investigate the electrochemical properties and structural stability of potential intermediate phases.

Main Methods:

• A hierarchical computational workflow combining global structure search and machine-learning interatomic potentials.
• Density Functional Theory (DFT) for potential training and validation.
• Phonon, electronic-structure, climbing-image nudged elastic band (CI-NEB), and molecular dynamics (MD) simulations.

Main Results:

• A stable intermediate phase, *C*2/*c*-Li₃As₂, was identified on the convex hull between LiAs and Li₃As.
• This phase is dynamically stable, metallic, and exhibits an equilibrium potential of ~0.95 V (vs. Li/Li⁺).
• Theoretical capacity of 536.6 mAh g^-1 with 68.6% volume expansion, and a low Li⁺ migration barrier were predicted.

Conclusions:

• *C*2/*c*-Li₃As₂ is a promising intermediate phase for shallow-lithiation strategies in anodes.
• This phase has potential for developing fast-charging battery anodes by mitigating deep lithiation issues.