Liquid cooling of microelectronic devices by free and forced convection

Abstract The effect of component size on convective heat transfer from small devices (with surface areas between 2·00 and 0·01 sq. cm) was investigated analytically and experimentally. An idealized two-dimensional boundary-layer analysis predicted the average convective heat transfer coefficient should increase significantly as the size of the heat source decreases. In fact, the analysis indicated that for sources as small as IC chips the average convection coefficient might be more than an order-of-magnitude greater than that obtained under the same condition with larger sources. Forced convection experiments (with Freon 113 and a silicone dielectric liquid) verified the analytical predictions. In these experiments the average convection coefficient ranged from 0·2 W/cm2°C for the largest source to above 3 W/cm2°C for the smallest source. The same devices and coolants were used in the free convection studies. Again, the average convection coefficient for the smallest source was more than an order-of-magnitude greater than that obtained with the largest source operated under the same conditions. The study showed liquid immersion is an effective means of cooling small heat sources. Liquid cooling by free convection was found to be more than three times as effective as free convective air cooling of the same device. With forced convective liquid cooling, the improvement over air cooling was found to be more than a factor of 10. This means that with liquid coolants, convective heat fluxes of several hundred watts per square centimeter can be transferred from IC chips and thin-film resistors without overheating the devices. In addition, the analysis showed far greater fluxes can be transferred from some of the minute components presently in use.