About this Abstract |
| Meeting |
MS&T25: Materials Science & Technology
|
| Symposium
|
High-Entropy Materials: Solid Solutions, Intermetallics, Ceramics, Functional Materials and Beyond VI
|
| Presentation Title |
Active Learning Driven Materials Discovery for Low Thermal Conductivity Rare-Earth Pyrochlore-Oxide |
| Author(s) |
Amiya Chowdhury, Acacio Rincon Romero, Tanvir Hussain, Grazziela Figueredo |
| On-Site Speaker (Planned) |
Amiya Chowdhury |
| Abstract Scope |
The search for low thermal conductivity materials for power-generating gas turbines has ramped up in recent decades, and this study aims, for the first time, an active learning pipleline for designing high-entropy/multi-component ceramic (HEC/MCC) rare-earth pyrochlore with low thermal conductivity. Data collected from literature was used to build the surrogate model within a Bayesian optimization (BO) loop. Crystallographic parameters (like cationic radii) were used to describe the compositions, and the best set of parameters was then converted to a composition via a brute force search. A random forest-based surrogate model combined with expected improvement (EI) as an acquisition function was used for the active learning loop to design new HEC/MCC compositions (i.e., La0.29Nd 0.36Gd0.36)2Zr2O7 and La0.333Nd0.26Gd0.15Ho0.15Yb 0.111)2Zr2O7) that were then evaluated experimentally. TC of the new compositions matched the prediciton by the surrogate model, suggesting potential of the framework for designing new low TC pyrochlores. |