Scientists from Arizona State University and UNSW Sydney have launched an international collaboration to redesign high-temperature alloys for hypersonic aircraft, nuclear-powered submarines, and many of the most advanced defence systems.
An alloy is a material obtained by combining two or more metallic elements to achieve properties that no single metal can offer alone, such as increased strength or enhanced corrosion resistance. Refractory alloys are based on elements such as tungsten, niobium, and molybdenum, which possess some of the highest melting points among all metals and do not melt or weaken easily, even under extreme temperatures.
Teaching computers to design new metals
Redesigning refractory alloys using traditional trial-and-error methods would take decades. The alternative approach by researchers at Arizona State University and UNSW Sydney utilises reinforcement learning, a form of artificial intelligence best known for training computers to master games such as chess. Designing a new alloy is somewhat like mixing ingredients for a recipe, but at an atomic level. Instead of planning moves on a chessboard, the AI system explores thousands of possible alloy combinations, such as different variations of chemical elements. Even minute modifications to the ingredients can completely alter the behaviour of the final material. The AI virtually evaluates every candidate based on multiple criteria, including resistance to temperatures exceeding 1,000 degrees Celsius and resistance to damage caused by reacting with oxygen at high temperatures, as well as weight, cost, and, crucially, the reliability of 3D printing it. Alloys predicted to exhibit high performance are rewarded, while those that fail to meet the criteria are discarded. Through repeated cycles, the system learns which chemical combinations function best. The most promising alloys designed by the AI can then be produced and tested in the laboratory. Their real-world performance feeds back into the model, constantly improving its predictions.
Strategic advantages beyond the laboratory
For defence agencies, faster materials development translates to more rapid deployment for next-generation engines, hypersonic vehicles, and heat protection systems. AI-designed alloys can be optimised for strength, heat resistance, and manufacturability. For instance, NASA’s GRX-810 alloy, designed through computational methods and 3D printed, is 1,000 times more resistant to high temperatures than traditional alloys. Traditional production of refractory metals wastes up to 95% of the raw material through mechanical machining—the removal of unwanted material to create a precise shape—but 3D printing can reduce this percentage to nearly zero. This work is an international collaboration; at Arizona State University, the focus is on AI-driven computational design. The facilities at UNSW Sydney allow for high-temperature testing by analysing the metal’s microstructure and conducting additive manufacturing under realistic conditions.
This approach is not without its obstacles. One of the greatest is data scarcity: AI models learn from existing experimental results, and for refractory alloys, such data is limited. Far fewer alloys in this class have been systematically tested compared to more common materials like steel or aluminium. Practical constraints also exist. Refractory metal powders suitable for 3D printing are expensive and difficult to source, and the transition from small laboratory samples to full-scale components is complex. An alloy that performs well as a small-scale test sample may behave very differently when printed as a large, complex component. Finally, AI predictions must always be experimentally validated, and these experiments are costly and time-consuming. The system does not eliminate the need for rigorous physical testing.
A new model for defence-oriented research
The research is in its early stages. Currently, scientists at Arizona State University and UNSW Sydney are developing the AI model and gathering the experimental databases from which it will learn. By the end of the year, the first candidate alloy compositions will be selected for 3D printing and laboratory testing. The results will be reintroduced into the model. The team is also collaborating with defence research agencies to ensure the work aligns with real-world requirements and to lay the groundwork for large-scale programs. In an era where technological advantage increasingly depends on speed and adaptability, rethinking the way the metals underlying defence systems are designed can improve the systems themselves.
