Discover how scientists have leveraged machine learning to accelerate the discovery of refractory high-entropy alloys with exceptional mechanical properties, paving the way for groundbreaking advancements in high-temperature applications.
High-entropy alloys with unconventional refractory elements
Refractory High-entropy Alloys (RHEAs) have been viewed as the holy grail of high-temperature materials owing to their unprecedented combination of primitive strength, ductility, and thermal stability. Unfortunately, the enormous compositional space of these alloys has been a long-standing challenge for conventional design methodologies.
Now, an outstanding study by a research team from the University of Science and Technology Beijing, Guangdong Ocean University, US, and AiMaterials Research LLC presents a novel approach for using machine learning (ML) to predict RHEA compositions with exceptional mechanical properties[7]. The combination of ML algorithms, genetic search (an optimization process inspired by biological evolution), cluster analysis to interpret the underlying trends and experimental design enabled them to quickly narrow down over a billion possible compositions to discover four new alloy systems that exhibit impressive high-temperature yield strength coupled with room-temperature ductility.
ZrNbMoHfTa Alloy with Record High-Temperature Properties
Of the alloys identified, the ZrNbMoHfTa system — and in particular the composition Zr0. 13Nb0. 27Mo0. 26Hf0. 13Ta0. 21. This alloy has a yield strength that approaches 940 MPa at an incredible temperature of 1200 °C, outperforming the behavior of existing RHEAs and also common nickel-based superalloys typically limited to much lower temperatures.
The advances made here create new opportunities for the use of our materials in high-temperature structural applications, such as gas turbines in aerospace propulsion and nuclear reactors. The ZrNbMoHfTa alloy demonstrates the kind of high performance that represents a major evolution for materials science, moving into space far beyond what researchers had been able to imagine for conventional high-temperature alloys.
Conclusion
Combining machine learning with classical alloy design methods has allowed researchers to find entirely new compositions of RHEAs and to determine their respective degrees of lattice distortion rather quickly. The result goes beyond just filling a hole in the array of high-temperature materials available today: It sets a new standard for the field. This opens the door to a myriad of engineering applications for advanced high-temperature alloys in which researchers may produce further advancements as they refine their strategy and study new compositions.
