| About this Abstract |
| Meeting |
2023 TMS Annual Meeting & Exhibition
|
| Symposium
|
Accelerated Discovery and Insertion of Next Generation Structural Materials
|
| Presentation Title |
Machine Learning-assisted Discovery of Novel High Temperature Ni-rich NiTiHfZr Multi-component Shape Memory Alloys |
| Author(s) |
John Broucek, Daniel Salas, William Trehern, Ibrahim Karaman |
| On-Site Speaker (Planned) |
John Broucek |
| Abstract Scope |
Machine learning (ML) and artificial intelligence (AI) are emerging in material science and engineering as methods to discover novel materials for various applications. AI/ML methods can survey small and unexplored regions of a material design space with desired shape memory characteristics, increasing the rate of material discovery by limiting the number of experiments needed to demonstrate the properties of interest. In this work, we utilized ML to predict the martensitic transformation characteristics of Ni-rich NiTiHfZr multi-component shape memory alloys intended for high-temperature aerospace applications. These alloys are characterized by having multiple principal elements and typically demonstrate martensitic transformation between B2 austenite and B19’ martensite occurring at temperatures as high as 600°C. Experimental examination of thermal properties such as transformation temperatures, and enthalpy of transformation was conducted to validate predictions through direct comparison. Active learning was used during data collection to further enhance the accuracy of predictions. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
High-Entropy Alloys, High-Temperature Materials, Machine Learning |