University of Waterloo cooling system drops hotspot temp 17C in 15 seconds

University of Waterloo team reveal hybrid cooling system that greatly improves artificial hotspot temperature in simulated computing hardware. This is important as this technology helps address the thermal bottleneck problem in high density electronics. This system can offer faster speed than typical immersion cooling system. The result suggest that there is a possible solution to thermal issue where localized hotspot is holding the performance or reliability of a system.

Hybrid droplet cooling outpaces immersion methods for localized heat

This hybrid cooling system uses liquid-liquid encapsulation mechanism to address specific thermal anomaly instead of cooling the entire volume. It utilizes droplets of a specialized liquid coolant with lower boiling point than water, but higher density. The droplet is contained in a shell of silicone oil with ceramic particles to regulate the boiling action and efficiently move away the heat from heat source.

Specifications

• Cooling Speed: 17C drop in 15 seconds
• Thermal Density Tested: 14.4 W/cm2
• Cooling Mechanism: Liquid-liquid encapsulation with silicone oil shell and ceramic particles
• Coolant Properties: Boiling point lower than water, higher density than water, immiscible with water

The experiment reveals that at 14.4 W/cm^2 heat flux density, the central temperature of the simulated hotspot can be reduced by ~17 C in 15 seconds. This is several times faster than typical immersion cooling system as reported. The coolant is immiscible with water, making it work in separate phase in the cooling system without mixing with other fluids.

The researchers pointed out that these results were obtained from simulated computing system prototype and there are limited amount of datasets, so the technology is still experimental. The potential use of this cooling system is not limited to computer, but can also be applied to electric vehicle, batteries, lasers and LED lights. The technology is now focusing on proving this mechanism before scaling it to integrate into commercial hardware.