Researchers will work toward improving efficiency in electric motors, quantum computing, and large language models with IEE support.

This month, three UCSB projects received awards from the Institute for Energy Efficiency (IEE). The IEE’s Research Seed and Software Impact Seed Programs provide critical introductory funding to launch new research projects that promote energy efficiency in areas ranging from computing and communications to infrastructure.

"IEE seed funding empowers our researchers to advance innovative ideas that are still in the early stages of development but hold tremendous potential to improve energy efficiency,” said Steven DenBaars, director of the IEE and a professor of materials and electrical and computer engineering. “These projects often serve as launchpads for securing additional external funding, extending their impact from the laboratory to commercial products and broader societal benefits.”

The IEE Research Seed program began in 2019 to fund promising energy-efficiency research each year, while supporting new faculty members and novel collaborations across disciplines and departments. In 2022, the IEE added the annual Investment Group of Santa Barbara (IGSB) Software Impact Grant to bolster software projects that lead to significant advances or risk reduction, and are likely to result in commercial products that can have a positive effect on society.

Since the two IEE seed programs began, UCSB faculty members have received a total of $1.4 million in funding, said Mark Abel, executive director of IEE. The money comes entirely from private donations. “The IEE seed programs recognize innovative projects that have the potential to make a real impact on energy efficiency,” he said. “These awards have enrolled more than twenty new faculty members in IEE who have gone on to successfully apply for external grants or start new commercial ventures building on their seed projects.”

“This year’s seed projects explore critical gaps in the current energy-efficiency landscape through new collaborations and cutting-edge technology,” said emeritus professor of electrical and computer engineering John Bowers, who served as inaugural IEE director until his retirement last spring. “These researchers are continuing the tradition of excellence in scientific discovery that IEE is known for, and we are looking forward to seeing the directions and the impact of their work on the wider community.”

Sustainable Magnets for Electric Motors 
Quiet, ubiquitous electric motors power everything from smartphones and electric toothbrushes to e-bikes and public transport. While these motors have sustainability advantages over their gas-powered equivalents, their components can come from materials that have detrimental effects, said Bolin Liao, an associate professor and vice-chair of the Department of Mechanical Engineering. “One pressing challenge now is related to rare earth elements, which are super important for a lot of applications, including the magnets at the heart of electric motors,” he said. “But the mining process for rare earth is environmentally very harmful,” as it is energy-intensive and creates water pollution, radioactive waste, and health issues for miners. The demand for these magnets — in particular, neodymium-based magnets — is expected to double by 2035.

Liao will join his colleague Stephen Wilson, a professor in the Materials Department to seek out a sustainable magnetic material. “Bolin will use his expertise in machine-learning techniques and computation to try to predict interesting chemistries that will give you magnets with functional advantages,” said Wilson, who is also director of the NSF Quantum Foundry at UCSB. Wilson and his lab will then make these materials and conduct in-depth characterization of their magnetic properties.

The dream, Wilson said, is to develop materials that will not only reduce the need for rare earth materials, but that are also stronger than those used in the current technology. “If you can make a magnet stronger, then you can have less of it,” Wilson said. “The magnet — and the electric motor — can be lighter and more efficient.”

The pair plan to use their seed-grant research to pursue larger awards, which will allow them to scale up their work. “I'm very grateful for this seed grant, which will give us new resources to explore this new area,” Liao said. “This is a very rapidly evolving field. I'm hopeful that this can lead to a lot of new discoveries beyond the scope of this project.”

Cryogenic Optical Modulators for Electro-optic Transduction (COMET)
The second IEE Seed Grant tackles important work related to energy efficiency in quantum computing conducted by professor Galan Moody and project scientist Paolo Pintus, both based in the Department of Electrical and Computer Engineering. The team will conduct modeling and design work for cryogenic optical modulators, with a goal of developing a modulator that needs very little power while successfully transmitting the amount of data that a quantum computer can produce. They have also begun collaborating with Google Quantum AI and Raytheon BBN Technologies, and hope that, if they find a feasible model, they can use this grant as a springboard to additional support for hardware fabrication.

Life After GPUs: Multiplier-Free LLM Inference for Energy-Efficient AI
Large language models are everywhere, addressing questions about complex scientific research — and about what to cook for dinner. Regardless of the query, each response involves billions of multiplication operations, said electrical and computer engineering professor Kerem Çamsarı, this year’s IGSB Software Impact Grant recipient, and many AI models “think” too hard to answer the simplest of questions, adding up to enormous drains of power. To reduce the energy involved, Çamsarı and his colleagues in the Department of Electrical and Computer Engineering, postdoctoral scholar Corentin Delacour and NSF Fellow Kyle Lee, intend to swap the multiplication that AI models currently use to respond to queries for a more efficient approach based on addition, rooted in Çamsarı’s expertise in probabilistic computing. As part of their award, they are refining an algorithm that allows an AI model to adjust how much data it samples to solve a problem based on the nature of the problem it is trying to solve.