AI Discovers Six 3D-Printable Alloys for Extreme Heat Applications

A team at the University of Toronto Engineering has developed a new set of metal alloys designed to withstand extreme conditions, leveraging an AI-driven materials design process. These alloys are particularly well-suited for additive manufacturing, also known as 3D metal printing, opening possibilities for custom-made components in sectors such as aerospace and power generation.

The demand for materials that can endure significant temperature and pressure fluctuations, like those found in jet engines or nuclear power plant steam generators, is substantial. Traditional manufacturing methods often struggle to produce complex, lightweight, yet strong components. This new approach allows for the creation of materials with varying compositions, offering a hard exterior and a lighter interior, a feat difficult to achieve with conventional techniques.

Traditional high-performance alloys often rely on a single base element like nickel or cobalt, with limited additions. However, the vast number of potential alloy combinations, especially those with three or more principal elements, presents a significant challenge for exploration. To overcome this, Zou's team employed an AI technique called active learning, combining computer modeling, machine learning, and robotic manufacturing in a self-driving lab.

This AI system, supported by the University of Toronto's Acceleration Consortium, addresses the data limitations common in AI material discovery. By using data-lean models and strategically selecting samples for manufacturing and testing, the system iteratively refines its search. In a few weeks, the lab identified six promising new alloys within the nickel-cobalt-chromium system. One alloy demonstrated superior hardness retention at 600°C compared to industry standards like Inconel 625, while another exhibited exceptional oxidation resistance at 1,000°C, outperforming Inconel 625 by 85% and targeting even higher temperatures up to 1,200°C.