About this Abstract |
Meeting |
2023 TMS Annual Meeting & Exhibition
|
Symposium
|
Transmutation Effects in Fusion Reactor Materials: Critical Challenges & Path Forward
|
Presentation Title |
Accurate Fe-He Machine Learning Potential for Studying Helium Effects in Ferritic Steels |
Author(s) |
Krishna Pitike, Wahyu Setyawan |
On-Site Speaker (Planned) |
Krishna Pitike |
Abstract Scope |
Nanostructured Ferritic Alloys (NFAs) are being actively investigated for their potential application as structural materials in advanced fusion reactors. Due to the lack of fusion test reactors, a predictive mesoscale model is required to understand radiation damage in NFAs, including helium bubble accumulation effects. Machine learning interatomic potentials (MLPs) have shown promising results due to their high accuracy for a fractional computational cost, and high adaptability towards complex chemical environments, such as Fe-He-H-(YTO) in NFAs. Here, we develop a Fe-He potential, based on ≈10,000 atomic configurations sampled using DFT. The developed MLP predicts the bulk properties of BCC-Fe, and thermodynamics of helium bubbles accurately in the iron matrix. E.g., the mean average error (MAE) of He binding energies with HenV (Hen) bubbles (clusters) estimated using MLP is ≈4 times smaller than that of classical potentials. The current Fe-He MLP can be further developed to include all chemical interactions in NFAs. |
Proceedings Inclusion? |
Planned: |
Keywords |
Computational Materials Science & Engineering, Machine Learning, Nuclear Materials |