A D V A N C E D M A T E R I A L S & P R O C E S S E S | O C T O B E R 2 0 2 2 6 MACHINE LEARNING | AI MODEL PREDICTS PROMISING HEA ALLOYS Researchers at Texas A&M University, College Station, working with colleagues at the DOE’s Ames National Laboratory, Iowa, developed an artificial intelligence (AI) framework capable of predicting high-entropy alloys (HEAs) that can withstand extremely high temperature, oxidizing environments— such as those found in gas turbines used for power generation and aircraft propulsion. The method could significantly reduce the time and cost of finding new alloys by decreasing the number of experimental analyses required. “The next step of revolutionizing turbine technology is to change the material that is used to fabricate components, such as the blades, so that they can operate at higher temperatures without oxidizing catastrophically,” says Professor Raymundo Arroyave, FASM. When looking at different types of alloys for turbines, there is significant attention around HEAs. One unique characteristic of these alloys is that they become more stable at higher temperatures, offering the potential for use in extreme environments. The A&M team developed an AI framework capable of predicting the oxidation behavior of HEAs. The model combines computational thermo d y n am i c s , machine learning, and quantum mechanics to predict the oxidation of HEAs made of arbitrary chemical compositions— in minutes instead of years. Using the framework, the scientists predicted the oxidation behavior of multiple alloy compositions. They sent these predictions to the Ames team to test their findings and verify that the framework accurately demonstrates if an alloy would or would not resist oxidation. “This tool will help screen out alloys that will not work for our application needs while allowing us to spend more time and create a more detailed analysis of alloys that are worth investigating,” says Arroyave. tamu.edu. COLOR-CODED X-RAY DATA Along with several universities, the DOE’s Argonne National Laboratory, Lemont, Ill., developed a method for creating color-coded graphs of large volumes of data from x-ray diffraction (XRD). The team believes this could greatly accelerate future research on structural changes on the atomic scale induced by varying temperature. “What might have taken us months in the past now takes about a quarter hour, with much more fine-grained results,” says Argonne physicist Raymond Osborn. The team calls their new method “x-ray temperature clustering,” or XTEC, and it draws on unsupervised machine learning using methods developed at Cornell University. This type of machine learning does not depend on initial training with data already well studied. Instead, it learns by finding patterns and clusters in large data sets without such lessons. These patterns are then represented by color coding. As a test case, XTEC analyzed data from beamline 6-ID-D at the Advanced Photon Source, taken from two crystalline materials that are superconducting at temperatures close to absolute zero. By applying XTEC, the team extracted an extraordinary amount of information about changes in atomic structure at different temperatures. “Because of machine learning, we are able to see material behavior not visible by conventional XRD,” says Osborn. “And our method is applicable to many big data problems in not only superconductors, but also batteries, solar cells, and any temperature-sensitive device.” anl.gov. Scientists recently developed a framework capable of predicting the oxidation of high-entropy alloys that could be used in gas turbines. Courtesy of Texas A&M Engineering. Machine learning provides a colorcoded map of x-ray data based on the temperature dependence of each region. Courtesy of PNAS.org. https://doi. org/10.1073/pnas.2109665119.
RkJQdWJsaXNoZXIy MTYyMzk3NQ==