Can Chip Hot Spots Be Cooled Efficiently?

Researchers at two University of California campuses are cookingup one cool technology. Led by Ari Shakouri, an electrical engineering professor at UC Santa Cruz, a team of scientists is refining what are essentially microscopic refrigerators used for cooling semiconductors and optoelectronic devices such as lasers. The result ultimately could mean substantial savings for chip makers and telcos, as well as more dependable data integrity for those using future products cooled by the tiny devices.

As chips get smaller and more powerful, and as telecom networks increasingly depend on wavelength division multiplex technology to distribute data, temperature control is a growing concern. Smaller, faster chips are prone to overheating. Shakouri says spot-cooling chip hot spots can greatly improve performance. Meanwhile, the WDM networks popular among telcos deliver data on up to 80 colors of laser through a medium that previously carried just a single laser. Consequently, a small temperature change can alter the wavelength enough to degrade the colors, resulting in lost data.

Until now, chipmakers have depended largely on fans. But by refining a new thermoelectric material made from silicon, germanium, and carbon, Shakouri and his team believe they've found a way to yield improved cooling that can be put directly onto the silicon substrate during chip making.

The researchers thus far have achieved a net cooling of about 7 degrees for chips operating at 100 degrees Celsius, but they say they'll have to do better than that. "The performance we have so far isn't good enough for commercial uses," Shakouri admits. He and one of his key collaborators, John Bowers, an electrical and computer engineering professor at UC Santa Barbara, are confident that further refinements in materials will result in more cooling.

But Frank Dzubeck, an analyst with Communications Network Architects, says there are numerous uses for the technology, but the potential customer base will gravitate toward future technologies that don't depend upon "plumbing," as he calls it. "It's a niche," Dzubeck says. "It exists until materials and technology sort of replaces it." Bowers, however, believes the market is substantial enough to warrant continued research. "It's real clear that this is a big need."

We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.
Comment  | 
Email This  | 
Print  | 
More Insights
Copyright © 2020 UBM Electronics, A UBM company, All rights reserved. Privacy Policy | Terms of Service