As AI stresses electricity grid, UB awarded $1.9 million for energy-efficient microelectronics

BUFFALO, N.Y. — Beleaguered electricity grids around the world are facing a looming challenge due to the explosive growth of power-hungry artificial intelligence systems.

To alleviate this stress, a University at Buffalo-led research team has been awarded $1.9 million grant from the National Science Foundation to develop new high-speed microelectronics that require less power than traditional silicon-based products.

“To support the growth of AI, and to do so in a way that is energy efficient and sustainable, we must develop new microelectronics,” says the grant’s principal investigator Jonathan Bird, PhD, professor and chair of the Department of Electrical Engineering at the UB School of Engineering and Applied Sciences.

The three-year grant, part of $42.4 million in NSF funding announced Sept. 16, supports the UB Center for Advanced Semiconductor Technologies, which Bird directs. The center is a key part of the NY SMART I-Corridor Tech Hub, which includes UB and earlier this summer was awarded $40 million to advance semiconductor manufacturing, research and education.

The grant also enhances New York State’s investments in semiconductor fabrication, supply chains and programs that support entrepreneurism and workforce development.

“What you’re seeing is the continued growth of an innovation ecosystem in upstate New York that’s centered on advanced research and development. This ecosystem supports the semiconductor industry and graduates highly skilled students who are ready to join the workforce,” Bird says.

Bird will lead a team that includes co-principal investigators Bibhudatta Sahoo and Fei Yao (UB), Nathaniel Cady and Robert Geer (both University at Albany), and Peter Dowben (University of Nebraska-Lincoln).

For the project, the researchers will develop new transistors, which are an essential component in modern electronics that control the flow of power on microchips.

The new transistors aim to combine the distinct advantages of two different classes of materials. The first is magnetoelectrics, whose magnetization orientation can be switched both rapidly and with low-power consumption. The second is transition-metal dichalcogenides, which are two-dimensional semiconductors consisting of atomically thin sheets that can be utilized as the conducting channel of a transistor.

These “magnetoelectric transistors” can potentially operate at higher speeds while demanding less energy than current silicon-based transistors.

They also maintain their programmed state when the device is turned off. This helps address the growing performance gap between a computer’s processor and its memory, known as the “memory wall.” This arose as computer processors sped up while memory access speeds failed to keep pace, resulting in bottlenecks and limiting the overall performance of computing systems.

Researchers will perform this work, in part, at NY CREATES’ Albany NanoTech Complex, which includes a state-of-the-art wafer-scale foundry.