About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Materials and Chemistry for Molten Salt Systems
|
| Presentation Title |
Uncovering Key Features of Molten Salt Corrosion in High-Entropy Alloys Through Machine Learning and High-Throughput Experiments |
| Author(s) |
Kevin Gautier, Laure Martinelli, Adrien Couet |
| On-Site Speaker (Planned) |
Kevin Gautier |
| Abstract Scope |
Molten salts are of growing interest for low-carbon energy systems due to their high heat capacity and low viscosity, making them suitable as nuclear fuels and coolants, or in concentrated solar power. However, at operating temperatures (500-800°C), materials in contact with molten chlorides or fluorides experience severe corrosion.
Goh et al. evaluated the corrosion behavior of ~100 additively manufactured CrFeMnNi high-entropy alloys in molten LiCl–KCl-EuCl3. With a random forest machine learning (RF-ML) model, they identified the Ni tracer diffusion coefficient as a key feature governing corrosion.
Maintaining this high-throughput approach, the robustness of the droplet corrosion test was improved, and new corrosion data were generated with different salt chemistry. The RF-ML model was refined by incorporating segmented SEM images of corroded surfaces, enabling links between thermodynamic and morphological features. Additionally, surface tension values from the sessile drop method were measured to investigate correlations between interfacial properties and corrosion resistance. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, Other, Other |