About this Abstract |
Meeting |
2023 TMS Annual Meeting & Exhibition
|
Symposium
|
Computational Thermodynamics and Kinetics
|
Presentation Title |
Chemistry and Processing History Prediction from Microstructure Morphologies |
Author(s) |
Mahmood Mamivand, Amir Abbas Kazemzadeh Farizhandi |
On-Site Speaker (Planned) |
Mahmood Mamivand |
Abstract Scope |
Designing targeted microstructures has been a long-lasting problem in the materials science community. Historically experimental approaches and recently high-fidelity models have been used to address this challenge in a trial-and-error way. However, the inverse design has been out of reach. The primary motivation of this work is to develop a model that can look at a microstructure morphology and predict its chemistry and processing history. As a case study, we have focused on spinodal decomposition in FeCrCo alloy. We have developed a fused-data deep learning network that reads the Fe-based morphology and predicts how much initial Cr and Co along with what heat treatment temperature and time are needed to get that microstructure. The model is successfully validated against an experimental transmission electron microscopy (TEM) micrograph, even though the model is trained on synthetic data and the TEM image did not have the training data quality. |
Proceedings Inclusion? |
Planned: |
Keywords |
Computational Materials Science & Engineering, Machine Learning, Magnetic Materials |