ADVANCED MATERIALS & PROCESSES | APRIL 2025 10 MACHINE LEARNING | AI MACHINE LEARNING IMPROVES TUNABLE METAMATERIAL Engineers at the University of California, Berkeley developed a tunable metamaterial microwave absorber that can switch between absorbing, transmitting, or reflecting microwaves on demand by mimicking the color- changing mechanism of a chameleon. “A key discovery was the ability to achieve both broadband absorption and high transmission in a single structure, offering adaptability in dynamic environments,” explains principal investigator Grace Gu. According to Gu, creating materials that can efficiently absorb electromagnetic waves has been a longstanding technological challenge. Aiming to develop a material that could dynamically change how it interacts with electromagnetic waves, researchers looked to chameleons for inspiration. These reptiles change color by adjusting the spacing between photonic crystals in their skin to modulate light reflection. Gu and her team worked to adapt a similar tuning mechanism to their metamaterial design. The result was a crisscross truss structure that can mechanically transform to control its electromagnetic properties. By collapsing or expanding its truss system, the metamaterial can vary its electromagnetic response from broadband absorption to transmission mode. Using machine learning and genetic algorithms, researchers optimized the design for specific electromagnetic responses, achieving a level of programmability. Next, they fabricated the structure using 3D printing and tested its ability to switch between absorbing and transmitting microwaves. Researchers say the new electro- magnetic material could enhance technologies in defense, wireless communications, energy, and smart infrastructure. In addition, it could be used to improve the efficiency of electromagnetic energy harvesting systems that help power sensors and batteries. “The tunable nature of the design allows it to adapt to changing needs, providing a versatile solution for electromagnetic wave management,” says Gu. berkeley.edu. AI HELPS GENERATE CRYSTAL STRUCTURES Scientists at the University of Reading, U.K., and University College London, developed a new artificial intelligence model that can predict how atoms arrange themselves in crystal structures. The technology, named CrystaLLM, works like an AI chatbot—learning the “language” of crystals by studying Generated structures of various inorganic compounds: (a) Ba2MnCr; (b) CsCuTePt; (c) YbMn6Sn6; (d) AuO2; and (e) Sm2BS4. Courtesy of Nature Communications, 2024, doi.org/10.1038/s41467-024-54639-7. millions of existing structures. The new system soon will be distributed to the scientific community to aid the discovery of new materials for everything from advanced batteries and solar panels to faster computer chips, say researchers. The current process for deter- mining how atoms will arrange themselves into crystals relies on computer simulations of physical interactions between atoms. In contrast, CrystaLLM learns by reading millions of crystal structure descriptions contained in Crystallographic Information Files, the standard format for representing crystal structures. CrystaLLM treats these descriptions like text. As it reads each one, it predicts what comes next, gradually learning patterns about how crystals are structured. The system was never taught any physics or chemistry rules, but instead figured them out on its own. It learned things such as how atoms arrange themselves and how size affects crystal shape, just from reading these descriptions. When tested, CrystaLLM could successfully generate realistic crystal structures, even for materials it had never seen before. www.reading.ac.uk. A chameleon’s color-changing mechanism (top) and the bioinspired tunable metamaterial microwave absorber (bottom).
RkJQdWJsaXNoZXIy MTYyMzk3NQ==