| About this Abstract |
| Meeting |
2020 TMS Annual Meeting & Exhibition
|
| Symposium
|
Deformation and Transitions at Grain Boundaries VII
|
| Presentation Title |
Simulating Grain Boundary Structures with DFT Accuracy Through Active Learning of Interatomic Potentials |
| Author(s) |
Tolga Akiner, Srikanth Patala |
| On-Site Speaker (Planned) |
Srikanth Patala |
| Abstract Scope |
Grain boundaries (GBs) influence a wide array of physical properties in polycrystalline materials and play an important role in governing microstructural evolution under extreme environments. Therefore, to develop robust predictive models of structural alloys, the structures and energetics of GBs, with DFT accuracy, are desired. Traditionally, ab initio techniques can only be used to simulate highly-symmetric GB structures and are limited to system sizes of the order of hundreds of atoms. In this talk, we present a machine learning approach based on Moment Tensor Potentials and active learning (AL) to simulate complex GB structures. Using the AL strategy, we develop an interatomic potential that is specifically suited for the atomic configurations of interest – the interface structures. We validate this method for the general GB structures of Aluminum and discuss the potential applications of this technique for simulating GBs in multi-component alloy systems with DFT accuracy. |
| Proceedings Inclusion? |
Planned: Supplemental Proceedings volume |