About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Mechanical Behavior at the Nanoscale VIII
|
| Presentation Title |
Towards Accurate Nanocrystalline Aluminum Simulations: A Comparative Analysis of MEAM and Machine Learning Potentials |
| Author(s) |
Ankit Yadav, Lucia Bajtošová, Miroslav Cieslar, Jan Fikar |
| On-Site Speaker (Planned) |
Ankit Yadav |
| Abstract Scope |
Accurate simulation of nanocrystalline(NC) metal behavior remains a challenge due to limitations in classical interatomic potentials. In this study, we investigate the mechanical response of NC aluminum(Al) thin films using both the traditional MEAM potential and a Machine Learning(ML) potential trained on high-fidelity DFT data. Earlier simulations using MEAM incorporated realistic grain boundaries and porosity, helping reduce the overestimation of mechanical properties but still deviating from experimental observations. We construct NC Al samples with improved structural features using MEAM and evaluate their mechanical behavior. These configurations are then verified using the ML potential to assess its effectiveness in enabling more realistic modeling of NC materials. This comparative approach allows us to evaluate how well classical potentials capture essential deformation mechanisms and to what extent ML-based models improve predictive accuracy. The study highlights the promise of ML potentials in bridging the gap between atomistic simulations and experimental reality for NC metals. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Aluminum, Thin Films and Interfaces, Computational Materials Science & Engineering |