University of Washington researchers believe their ion pump marks the coolest research breakthrough in decades. That's because the tiny device does only one thing—it cools chips.
As chips get smaller and denser, they also run hotter. Yet while chip performance has advanced remarkably over the past several decades, cooling technologies have remained largely unchanged. When faced with the need to dissipate heat, designers continue to turn to passive heatsinks and active fans as practical and relatively inexpensive cooling solutions.
As chips get hotter and device form factors shrink, though, conventional cooling technologies simply won't be able to confine temperatures to an acceptable operating range. The ion pump, which is small enough to be mounted on top of a processor or other chip, solves this problem by using an electrical charge to create a cooling air jet directly over a chip's surface.
"It has a lot of potential for mobile devices since it can be placed in locations where a heatsink or fan just won't fit," says graduate student Nels JewellLarsen, a research team member. "We hope that it will be able to replace, or supplement, heatsinks and fans in a variety of applications."
The prototype pump uses an electrical field to expel air at speeds previously only possible with traditional blowers. The device consists of two components: an emitter and a collector. With a tip radius of about 1 µm, the emitter creates air ions (electrically charged particles) that are propelled in an electric field to the collector surface.
As the ions move from the emitter to the collector, they create an air jet that blows across the chip, removing heat. For precise temperature management, the airflow's volume can be controlled by varying the voltage between the emitter and collector. Tests showed that the prototype, drawing only 0.6 W, could significantly cool an actively heated surface.
Igor Krichtafovich, chief technology officer of Kronos Air Technologies, was impressed by the prototype's performance. "The technology was theoretical before the University of Washington became involved," he says. Kronos and Intel are both supporting the project.
Now that they've created a working prototype, the researchers are looking forward to solving the roadblocks impeding the technology's commercial release. At the top of the list is developing mathematical models for controlling multiple chip-mounted ion pumps.
"There are all sorts of microscale forces involved," Jewell-Larson says. "To provide the most efficient cooling, you have to consider electrical fields, moving charges, and electrohydrodynamic forces."
The researchers are also searching for the best possible physical materials for building durable, high-performance pumps. Not surprisingly, the team is looking closely at nanotechnology.
"Nanotubes and other nanostructures promise the most performance improvements," Jewell-Larson says, adding that the researchers are also working with Intel on microelectromechanical systems (MEMS) and CMOS processes to build pumps directly into silicon. "We're hoping it will lead to some cool new products."