A team led by UC Santa Barbara materials and chemistry professor Ram Seshadri has been awarded a multi-UC-campus project aimed at using AI to accelerate materials research toward the goal of developing new energy-efficient platforms for manipulating information and reducing the cost of computing.

Seshadri, who is the Fred and Linda R. Wudl Chair and Distinguished Professor of Materials, as well as associate dean for research in the College of Engineering (COE) and director of the UCSB Materials Research Laboratory, will lead a group focused on leveraging AI to, he says, “narrow the field of materials that will be tested in the lab, thus accelerating discovery of new chip materials that will switch faster and use less energy,” both concerns as the use of AI and the amount of energy it requires grow exponentially. 

“At the heart of any computer chip is a little device called a CMOS [complementary metal-oxide semiconductor], which has become extremely cheap to manufacture at scale,” Seshadri explains. “In every operation, a CMOS switches between two states. When you're doing it a gazillion times on every chip, and you have a gazillion chips all over the world, it becomes very expensive in terms of energy use. So, the question that was posed was: How can we come up with new ways of carrying out these switching operations?”

One place to look is the area of topological materials, characterized by the fact that their electrons obey certain rules of topology. Seshadri refers to topology as “rubber-sheet geometry, the idea that two objects can be related if you can deform one of them into the shape of the other. They're not related to each other if, for example, you must poke holes in one to change it into the shape of the other. The classic example of this is a coffee cup and a donut. Because they both have a single hole, they belong to the same topological family. But a donut and a bun, which has no hole, would be in different topological families, and then a pretzel belongs to yet another class, because it has more than one hole, and to turn those multiple holes into the one hole of a donut, you'd have to tear the pretzel, which would destroy its characteristics.

“Topological materials have grown in stature over the past several years, with several UC Santa Barbara researchers contributing to an increased understanding of them,” Seshadri continues. “This project [called Low-Energy, AI-Informed Phase Transitions (LEAP)], is intended to address the question of whether we can use these materials in new switching paradigms. It's fundamental work, with a lot of AI incorporated into it to accelerate the discovery of new topological materials.”

Seshadri explains that researchers from UCSB, UC Berkeley, UC Irvine, and UC Merced, as well as from the three UC-operated national laboratories — Lawrence Berkeley, Livermore, and Los Alamos — will combine theory, real-world data, and simulations to train large language models to advance effective topological materials for switching. While the goals of the project are aimed at conventional computing, Seshadri notes an added benefit: that topological materials have also been proposed as possessing the materials qualities required of quantum computing platforms.

The University of California has an eighty-year history of managing and operating the three national laboratories as a public service to the nation. Since 2017, UC has directed more than $92 million of the lab-management funding fees it receives toward areas of scientific and national-security importance through the Collaborative Research and Training Awards. The Artificial Intelligence Science at Scale initiative provides three grants of $6 million each over three years to develop new AI approaches in three areas of strategic importance: genomics, geothermal energy, and quantum-materials discovery.

“This special initiative brings together the unmatched academic expertise of UC faculty, world-class scientific talent and capabilities at our national labs, and the strategic leadership within our system to accelerate the scale, reach, and impact of AI-powered scientific discovery research that benefits the nation,” said June Yu, Vice President of UC National Laboratories.

“The rapid growth of artificial intelligence is accelerating both the opportunities for, and threats to, the United States’ long-standing economic leadership,” said Theresa Maldonado, UC Vice President of Research & Innovation. “To help keep America in the lead, the University of California is scaling up its commitments in critical emerging areas of scientific research, such as AI, and moving forward with unprecedented speed to fund targeted research that fosters innovation.”