ADVANCED MATERIALS & PROCESSES | APRIL 2024 10 MACHINE LEARNING | AI NEW CERAMICS HANDLE THE HEAT A multi-university team led by Duke University, Durham, N.C., developed a method to rapidly discover a new class of materials with heat and electronic tolerances tough enough to withstand extremely high temperatures— greater than several thousands of degrees Fahrenheit. These new materials are both harder than steel and stable in chemically corrosive environments. Potential applications include new wear and corrosion-resistant coatings, thermoelectrics, batteries, catalysts, and devices resistant to radiation. The recipes for these materials— ceramics made using carbonitrides or borides—were discovered by using a new computational method called disordered enthalpy-entropy descriptor (DEED). In its first demonstration, the program predicted the synthesizability of 900 new formulations of high-perfor- mance materials, 17 of which were tested and produced in laboratories. Professor Stefano Curtarolo’s group maintains the Duke Automatic-FLOW for Materials Database (AFLOW), a huge data reservoir of materials properties connected to various online tools for materials optimization. The information enables algorithms to accurately predict properties of unexplored mixtures without having to make them in a lab. For the past few years, Curtarolo’s group has been working to develop predictive powers for high-entropy materials that achieve enhanced stability from a chaotic mixture of atoms. DEED is specifically tailored to hot-pressed sintering, which involves taking powdered forms of the constituent compounds and heating them in a vacuum to as high as 4000°F while applying pressure for a few hours. The finished ceramics have a dark metallic appearance, feel like metal alloys such as stainless steel, and have a similar density— although they are hard and brittle like conventional ceramics. The group expects other researchers to begin using DEED to synthesize and test the properties of new ceramic materials for a wide range of applications. duke.edu. MACHINE LEARNING DETECTS ELUSIVE PHASE OF MATTER Researchers at Cornell University, Ithaca, N.Y., detected an elusive phase of matter called the “Bragg glass phase” using large volumes of x-ray data and a new machine learning data analysis tool. The discovery settles an enduring question of whether or not this particular state of matter can exist in real materials. Collaborators include scientists at Argonne National Laboratory and Stanford University. The team presented the first evidence of a Bragg glass phase detected by x-ray scattering in a systematically disordered charge density wave (CDW) Molecular structure of new ceramic materials for extreme temperature applications. Courtesy of Duke University. material, PdxErTe3. They used comprehensive x-ray data and a machine learning data analysis tool called x-ray temperature clustering (X-TEC). Theoretically, there is a sharp distinction between three phases: long-range order, Bragg glass, and the disordered state. In the disordered state, the CDW correlation decays within a finite distance. In the long-range ordered state, the CDW correlation continues indefinitely. In the Bragg glass phase, the CDW correlation decays so slowly that it will only completely vanish at infinite distances. X-TEC made it possible to analyze the massive volume of data with a scalable and automated approach. Beyond exploring the Bragg mystery, the study presents a new mode of research in the age of big data. The team reports that this detection of Bragg glass order and the resulting phase diagram significantly advances our understanding of the complex interplay between disorder and fluctuations. Moreover, using X-TEC to target fluctuations through a high- throughput measure of peak spread could revolutionize how fluctuations are studied in scattering experiments. cornell.edu. Crystal structure of pure ErTe3.
RkJQdWJsaXNoZXIy MTYyMzk3NQ==