New microchannel design drives 3D chip cooling to new record
“You can pack only so much computing power into a single chip, so stacking chips on top of each other is one way of increasing performance,” said Justin Weibel, a research associate professor in Purdue’s School of Mechanical Engineering and co-investigator on the project. “This presents a cooling challenge because if you have layers of many chips, normally each one of these would have its own system attached on top of it to draw out heat. As soon as you have even two chips stacked on top of each other the bottom one has to operate with significantly less power because it can’t be cooled directly.”
The $2m project for the US Defense Advanced Research Projects Agency (DARPA) uses a new structure of micro-channels in the 3D interposer layers with an electrically non-conducting fluid. The channels are 100 to 15μm wide but 300μm deep, which is a significant challenge from a fabrication perspective, particularly for repeatable and low-cost manufacturing processes.
Earlier this year IBM showed a microfluidic cooling system for DARPA that reduced junction temperature in its Power 7+ processors by 25ᵒ C and reduced chip power usage by 7 percent compared to traditional air cooling.
The Perdue sees its system as suitable for supercomputers and high performance electronics such as radar. “I think for the first time we have shown a proof of concept for embedded cooling,” said Suresh Garimella, Goodson Distinguished Professor of Mechanical Engineering at Purdue. “This transformative approach has great promise for use in radar electronics, as well as in high-performance supercomputers. This number of 1,000 watts per square centimetre is sort of a Holy Grail of microcooling, and we’ve demonstrated this capability in a functioning system with an electrically insulated liquid.”
Next: Microchannel design for dielectric fluid
The system uses a commercial refrigerant called HFE-7100, a dielectric fluid that boils inside the microchannels. “Allowing the liquid to boil dramatically increases how much heat can be removed, compared to simply heating a liquid to below its boiling point,” he said .
“It’s been known for a long time that the smaller the channel the higher the heat-transfer performance,” said researcher Kevin Drummond. “We are going down to 15 or 10 microns in channel width, which is about 10 times smaller than what is typical for microchannel cooling technologies.”
Using ultra-small channels increases the cooling performance, but it is difficult to pump liquids through long microchannels. Instead the team designed a series of short, parallel channels with a hierarchical manifold that distributes the coolant through the channels.
“So, instead of a channel being 5,000 microns in length, we shorten it to 250 microns long,” said Garimella. “The total length of the channel is the same, but it is now fed in discrete segments, and this prevents major pressure drops.
- MICROFLUIDIC CHANNELS COOL STACKED CHIPS AND DATACENTRES
- IMMERSION COOLING BOOST FOR X86 CLUSTER
- NEW COOLING STANDARD BOOSTS OPEN ARCHITECTURE AVIONICS BOARDS
- THERMAL GELS HELP COOL BATTERIES