Microfluidic cooling may prevent the demise of Moore's Law

Micro-drops of water channeled through the chip silicon looks like a promising way to keep chips cool and increase their performance.

Image: iStock/agsandrew

Existing technology's inability to keep microchips cool is fast becoming the number one reason why Moore's Law may soon meet its demise.

In the ongoing need for digital speed, scientists and engineers are working hard to squeeze more transistors and support circuitry onto an already-crowded piece of silicon. However, as complex as that seems, it pales in comparison to the problem of heat buildup.

"Right now, we're limited in the power we can put into microchips," says John Ditri, principal investigator at Lockheed Martin in this press release. "One of the biggest challenges is managing the heat. If you can manage the heat, you can use fewer chips, and that means using less material, which results in cost savings as well as reduced system size and weight. If you manage the heat and use the same number of chips, you'll get even greater performance in your system."

Resistance to the flow of electrons through silicon causes the heat, and packing so many transistors in such a small space creates enough heat to destroy components. One way to eliminate heat buildup is to reduce the flow of electrons by using photonics at the chip level. However, photonic technology is not without its set of problems.

SEE: Silicon photonics will revolutionize data centers in 2015

Microfluid cooling might be the answer

To seek out other solutions, the Defense Advanced Research Projects Agency (DARPA) has initiated a program called ICECool Applications (Intra/Interchip Enhanced Cooling). "ICECool is exploring disruptive thermal technologies that will mitigate thermal limitations on the operation of military electronic systems while significantly reducing the size, weight, and power consumption," explains the GSA website

MicroCooling 1
Image: DARPA
What is unique about this method of cooling is the push to use a combination of intra- and/or inter-chip microfluidic cooling and on-chip thermal interconnects.

The DARPA ICECool Application announcement notes, "Such miniature intra- and/or inter-chip passages (see right) may take the form of axial micro-channels, radial passages, and/or cross-flow passages, and may involve micro-pores and manifolded structures to distribute and re-direct liquid flow, including in the form of localized liquid jets, in the most favorable manner to meet the specified heat flux and heat density metrics."

Using the above technology, engineers at Lockheed Martin have experimentally demonstrated how on-chip cooling is a significant improvement. "Phase I of the ICECool program verified the effectiveness of Lockheed's embedded microfluidic cooling approach by showing a four-times reduction in thermal resistance while cooling a thermal demonstration die dissipating 1 kW/cm2 die-level heat flux with multiple local 30 kW/cm2 hot spots," mentions the Lockheed Martin press release.

In phase II of the Lockheed Martin project, the engineers focused on RF amplifiers. The press release continues, "Utilizing its ICECool technology, the team has been able to demonstrate greater than six times increase in RF output power from a given amplifier while still running cooler than its conventionally cooled counterpart."

Moving to production

Confident of the technology, Lockheed Martin is already designing and building a functional microfluidic cooled transmit antenna. Lockheed Martin is also collaborating with Qorvo to integrate its thermal solution with Qorvo's high-performance GaN process.

The authors of the research paper DARPA's Intra/Interchip Enhanced Cooling (ICECool) Program suggest ICECool Applications will produce a paradigm shift in the thermal management of electronic systems. "ICECool Apps performers will define and demonstrate intra-chip and inter-chip thermal management approaches that are tailored to specific applications and this approach will be consistent with the materials sets, fabrication processes, and operating environment of the intended application."

If this microfluidic technology is as successful as scientists and engineers suggest, it seems Moore's Law does have a fighting chance.

Also see

About Michael Kassner

Information is my field...Writing is my passion...Coupling the two is my mission.

Editor's Picks

Free Newsletters, In your Inbox