| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
Advanced Materials for Energy Conversion and Storage 2025
|
| Presentation Title |
Discovery of Thermodynamically Stable Disorder in High-Entropy Li-Oxides from Ab-Initio Simulation and Crystal Graph Neural Network Prediction |
| Author(s) |
R. Seaton Ullberg, John D. Langhout, Megan M. Butala, Simon R. Phillpot |
| On-Site Speaker (Planned) |
R. Seaton Ullberg |
| Abstract Scope |
As improvements in the energy density of Li-ion batteries have become more difficult to achieve, research into novel classes of materials is emerging as a solution to enhance energy capacity. Recent work has shown that Li-based cation-disordered rocksalt structures are a promising cathode material due to their high capacity and vast compositional space which can be exploited to tune performance. However, short-range ordering of Li can impede Li-ion transport which reduces performance. In this work, a high-fidelity crystal graph neural network, CHGNet, is employed to predict the favorability of disorder for compositions in the Li1.2(Mn,X,Y)0.8O2 system with X and Y representing a subset of transition metals. Compositions predicted to favor disorder are validated using density functional theory (DFT) calculations to identify candidates for future experimental investigation. Of nine promising candidates identified by CHGNet, one is successfully confirmed with DFT to exhibit thermodynamically stable disorder. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, Energy Conversion and Storage, Machine Learning |