| Abstract Scope |
Ru-based B2 precipitates have been proposed to increase the high-temperature strength of BCC refractory high entropy alloys through coherent precipitation as in Ni γ-γ’ superalloys. However, there is a large design space of possible compositions among established candidates that can be explored to optimize the properties of the end alloy, such as the morphology, high temperature stability, solutionizing temperature, and lattice mismatch of the B2 phase. To explore the space effectively, we combine CALPHAD simulation with Bayesian sampling methods and surrogate ML models to both explore and characterize the possible compositions of Ru-B2 alloys. We predict the thermodynamic and atomic properties of RuTi, RuHf, and RuAl precipitates in alloys with Nb, Mo, and V rich matrices and use this data as inputs to machine learning for cheap inference to guide further predictions. We propose new BCC-B2 Ru-RHEA candidates that have enhanced stability and strength at temperatures exceeding 1300 C. |