Leakage losses and overall increased power dissipation in the microprocessor are causing significant thermal, mechanical, and reliability problems. Aside from the issue of cooling chip hot spots in order to reduce stress-inducing thermal gradients, the traditional challenge of quelling overall operating temperatures remains. Conventional cooling methods are reaching their practical limits and new methods of lowering the operating temperature of microprocessors are being explored. Microfluidics-based cooling schemes are one approach being considered. Implementation of microchannels for forced convection at the chip level shows much promise, as the effective heat transfer surface area and attainable heat transfer coefficient are very favorable. A major design limitation to such an implementation is the pressure developed with such micro-flows, and the stresses that could result. In this study, multiple discrete microchannel configurations are analyzed computationally and compared in a cooling capability optimization sense, while total pressure drop across the flows is carefully considered. A single cooling channel over an energy source is split into two smaller channels, and so on, while total pressure drop is maintained constant, and specified such that all flows remain in the laminar regime. It is shown that for the configurations analyzed, there exists a definitive optimum cooling scheme. In addition, the effects of variation of channel height for the initially-determined optimum scheme are studied. It is shown that a slimmer design may be implemented with very little effect on cooling capability.

This content is only available via PDF.
You do not currently have access to this content.