Investigating Stability and Lithium Diffusion in Solid Electrolytes Using Neural Networks
Author Information
Author(s): Sawada Ryohto, Nakago Kosuke, Shinagawa Chikashi, Takamoto So
Primary Institution: Preferred Networks, Inc.
Hypothesis
Can neural network potentials optimize the dopant configuration and lithium diffusion pathways in solid electrolytes?
Conclusion
The study demonstrates that neural network potentials can effectively optimize dopant configurations and lithium diffusion pathways in solid electrolytes.
Supporting Evidence
- The proposed method allows for the optimization of dopant configurations without prior knowledge.
- Results showed that the activation energy of lithium diffusion increases with certain dopant concentrations.
- NNP calculations were found to be in good agreement with experimental results.
Takeaway
This study shows how scientists can use smart computer programs to find the best ways to make batteries work better by looking at how materials behave at a tiny level.
Methodology
The study used molecular dynamics simulations with neural network potentials to analyze the stability and lithium diffusion in doped solid electrolytes.
Limitations
The method may require significant computational resources for large supercells and may not predict specific transition points accurately.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website