About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Thermodynamics and Kinetics of Alloys IV
|
| Presentation Title |
Completely bypassing DFT calculations to predict vacancy formation energies via graph neural networks in FCC high entropy alloys |
| Author(s) |
Nathan Linton, Dilpuneet S. Aidhy |
| On-Site Speaker (Planned) |
Dilpuneet S. Aidhy |
| Abstract Scope |
Vacancy formation energies (VFE) in structural alloys are critical to understand high temperature deformation processes such as diffusional creep. Due to chemical and lattice symmetries, while pure metals have single values of VFE, high entropy alloys (HEAs) have a range of VFEs due to diverse nearest neighbor (NN) environments. Using density functional theory (DFT) calculations, we show that atomic volume and electronegativity emerge as key parameters in HEAs that control VFEs. Further, because DFT is expensive, especially for HEAs, we employ a graph neural network (GNN) ML model that enables VFE predictions in HEAs from simple binary alloys. In addition, we show that DFT calculations can be completely bypassed by carefully fine-tuning CHGNeT to predict atomic volume, charge transfer and atomic displacement in HEAs. While this model is developed for FCC alloys in Ni-Cu-Au-Pd, it enables accurate predictions in Ni-Co-Cr system with minor tuning indicating extrapolative nature of the model. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, High-Entropy Alloys, Machine Learning |