Two-phase convection devices yield by far the highest cooling rates per unit volume in electronic systems, as evidenced by the effectiveness and large commercial impact of heat pipes in portable systems.  Other two-phase devices that provide even higher cooling rates include microchannels, micro jets (Fig. 1), and vapor-compression refrigerators.  However, these devices are rarely found in mass-produced electronic systems because they require high-pressure pumping (> 5 atm) of the liquid phase.  Until now, the IC industry has lacked liquid pumps that are compact, reliable, and capable of delivering sufficient pressure.

This project solves this fundamental problem by developing the first two-phase cooling devices based on high-pressure electrokinetic (EK) liquid pumps.  EK pumps use electric fields rather than moving parts and are therefore cheaper to manufacture and more compact and reliable than liquid pumps based on competing technology.  Research by one of us (Paul et al., 1998a) developed EK pumps with pressures as large as 500 atm, more than two orders of magnitude larger than pressures delivered by MEMS-based reciprocating, electrohydrodynamic, and peristaltic pumps (Shoji & Esahi, 1994).  Currently, DARPA is investing in EK-pumped microfluidic systems for actuators and bioanalytic devices.  However, no one is exploring their enormous potential for heat-transfer augmentation.

This project uses EK pumps to develop a new class of voltage-controlled coolers for electronic systems, which are compact, reliable (no solid moving parts), and self contained.  Task 1 (low-risk demo during year 1) yields EK-Pumped Phase-Change Packaging, which uses the high-pressures in EK pumps to remove heat from chips using two-phase microchannels and liquid microjets (see Fig. 1).  Heat is rejected through condensation over a large surface area condenser.  We will remove 200 W with less than 20 K chip temperature rise, a dramatic improvement over heat pipe technology; all with less than 1 W of EK pump power.  Task 2 (high-risk / high-yield, demo during year 3) yields a Package-Integrated EK Refrigerator.  Feasibility studies will choose the best of two strategies: a Joule-Thompson  cycle based on an EK vapor compressor (Fig. 1), and an EK-pumped adsorption cycle refrigerator.

These devices are prototyped and commercialized with Intel Corporation.  Intel has commited 1/2 time of a scientist and SRC customization funds.  Intel will also hire Stanford students and post-docs during the project to aid with tech transfer.

The heat exchanger technology developed for Task 1 includes both two-phase microchannels and liquid micro jet arrays.  The heat removal capabilities of both liquid jets and microchannels have been the subject of much research (e.g., Goodson, Kurabayashi, & Pease, 1997, or the Session on “MicroChannel Cooling,” MEMS Proc.. at the 1996 ASME IMECE, DSC-Vol. 59, and Lienhard, 1996).  However, these devices are having only a limited impact on commercial electronic systems due to the fact that there are no reliable, miniature pumps capable of generating the high pressure head required to propell sufficient mass flowrates through microstructures.

The EK pumps proposed here are far better suited for heat transfer applications than the compact (< 100 mm3) pumps reported in the literature.  Competing micromachined pumps reported in the literature include diaphragm pumps based on diffusers and check valves, electro-hydrodynamic pumps, and peristaltic and reciprocating pumps (see, for example, the review by Shoji & Esashi, 1994).  These devices provide high flowrates, in many cases more than 10 mL/min, but yield pressure differences below 1 atm.  The diaphragm pump of Woias et al. (1998), for example, provides pressure and flowrate of 0.9 atm and 1 mL/min, respectively, in an unmounted footprint of 7 x 7 x 1.1 mm3 and operates on an external switching power supply swinging from –40 to +120 V.  In contrast, EK pumps with an unmounted footprint of 1 x 1 x 3 mm3 generate similar flowrates with pressures in excess of 10 atm using less than 150 V.  Recent results by a PI (Paul) demonstrated the ability to build EK pumps propelling 5 ml/min through 10 atm.  These results show that EK pumps can usher in a new class of highly-effective coolers for electronic systems.
 

Figure 1: EK-pumped phase-change packaging (left), which generates the pressures needed for novel liquid microcoolers including liquid micro jets (right).