Session: 09-03-01: Advanced Electrochemical Energy Materials: Characterization, Modeling, and Theoretical Analysis

Paper Number: 173681

Ionic Transport Properties and Phase Stability of Solid Electrolyte Material Li7la3zr2o12: A Deep-Neural-Network Molecular Dynamics Investigation 

All-solid-state batteries (ASSBs) represent a transformative direction for next-generation energy storage, offering enhanced safety, energy density, and thermal stability compared to conventional liquid-electrolyte batteries. Among various solid electrolytes, lithium lanthanum zirconium oxide (Li₇La₃Zr₂O₁₂ or LLZO) and its doped variants, particularly aluminum-doped LLZO, have emerged as promising candidates due to their relatively high ionic conductivity and chemical stability. Over the past decade, substantial research efforts have focused on improving and diversifying electrolyte compositions since the initial laboratory synthesis of the first oxide-based all-solid-state battery electrolyte. However, challenges remain in the comprehensive understanding and controlling of the intricate nanoscale structural evolution, ion transport mechanisms, and long-term phase stability of these complex materials, hindering the path toward widespread commercial-scale ASSB deployment.

 

In this work, we present a comprehensive atomic-scale study of LLZO and Al-doped LLZO using deep neural network (DNN) potential-based molecular dynamics (MD) simulations, termed DNN-MD. Notably, we have developed and validated high-fidelity DNN interatomic potentials trained on extensive datasets from ab initio molecular dynamics, capturing a wide range of thermodynamic states and compositional variations, including lithium, aluminum, and oxygen environments. Our simulations reveal key insights into the lithium-ion diffusion mechanisms and phase transformation behaviors across a range of temperatures. Specifically, we uncover a temperature-dependent phase transition in LLZO that is significantly influenced by aluminum doping. The Al-doped LLZO exhibits enhanced structural stability and higher lithium-ion conductivity compared to undoped LLZO, in agreement with experimental findings. By adjusting the amount of aluminum atoms and changing the temperature, we were able to observe a clear and unique phase transformation, which also matches results that have been previously seen in laboratory-based experimental studies. These results provide atomic-level evidence that aluminum incorporation facilitates the stabilization of the cubic garnet phase, which is known to exhibit superior ionic transport properties.

 

The DNN-MD approach enables us to quantitatively analyze ion migration pathways, defect structures, and activation energies with high resolution, thereby offering a powerful and predictive tool for rational electrolyte design. Furthermore, our findings offer valuable guidance for optimizing the synthesis and processing conditions of LLZO-based electrolytes to maximize performance in practical ASSB configurations.

 

In conclusion, this study demonstrates the power of machine learning-driven molecular dynamics to elucidate the ion conduction mechanisms and phase behavior in advanced solid-state electrolytes. The methodologies developed here can be readily extended to other promising electrolyte systems, offering a scalable pathway to accelerate the discovery, optimization, and commercialization of next-generation materials for solid-state batteries.

Presenting Author: Chunxu Wang University of Nevada, Reno

Presenting Author Biography: Chunxu Wang is PhD student in University of Nevada, Reno.

Authors:

Chunxu Wang University of Nevada, Reno
Haoran Cui University of Nevada, Reno
Yan Wang University of Nevada, Reno
Lei Cao University of Nevada, Reno