High-performance alloys are crucial for applications in extreme environments, such as aerospace and power generation. Traditional nickel- and cobalt-based superalloys, while offering thermal stability and good mechanical properties at high temperatures, are limited by a melting point below 1,500 °C. Refractory high-entropy alloys (HEAs) have emerged as promising alternatives owing to their high strength and ductility at elevated temperatures, but challenges such as high cost and poor oxidation resistance hinder their broader adoption. Now, writing in Science Advances, Jie Qi, Peter Liaw, Joseph Poon and collaborators describe the computational design and experimental validation of Al-enriched refractory HEAs that combine high strength and ductility with reduced cost and good oxidation resistance.

Machine learning was used to identify optimal compositional regions among quaternary HEAs that combine Al with three refractory elements. The alloys with the most promising combinations of strength and ductility were optimized using Monte Carlo simulations and experimentally investigated. The alloys with the best machine-learning-predicted performance turned out to exhibit limited plasticity due to strong long-range ordering, which hinders dislocation movement and reduces plasticity. By systematically reducing the Al content, the researchers could improve plasticity while maintaining a high strength. The resulting toughness was higher than that of most HEAs, while the presence of Al reduced cost and density and improved the resistance to oxidation and corrosion. The optimized refractory HEAs exhibit superior strength and ductility compared with typical Al-containing refractory HEAs, featuring compressive yield strengths of up to 1.7 gigapascals and fracture strains exceeding 50%.