LIMITS OF THIN FILM COOLING IN MICROGAP CHANNELS
MetadataShow full item record
The forced flow of dielectric liquids, undergoing phase change while flowing in a narrow channel, is a promising candidate for the thermal management of advanced semiconductor devices. Such channels may be created by the spacing between silicon ribs in a microchannel cooler, between stacked silicon chips in a three-dimensional logic, RF, or heterogeneous microsystem, narrowly-spaced organic or ceramic substrates, or between a chip and a non-silicon polymer cover in a microgap cooler. These microgap configurations provide direct contact - and hence cooling - between a chemically-inert, dielectric liquid and the back surface of an active electronic component, thus eliminating the significant thermal resistance associated with a Thermal Interface Material (TIM) or the solid-solid contact resulting from the attachment of a microchannel cold plate to the chip. This dissertation explores the physics underpinning two-phase flow in miniature channels, through an extensive literature survey, and employs analytical, numerical, and experimental techniques to determine the thermal transport phenomena in microgap channels, with emphasis on the thermal limits of thin film heat transfer in annular flow. The applicability of several flow regime mapping methodologies has been examined. The predictions of these mapping methodologies have been compared to the visual observations of two-phase flow in microtubes and microchannels. The axial variation of two-phase heat transfer coefficients with local vapor qualities is reported, and the association of this variation with the dominant flow regime is discussed. The measured two-phase flow heat transfer coefficients are then sorted according to the dominant flow regime, and compared to the predictions of classical heat transfer correlations. Two-phase flow experiments were performed in a microgap cooler with the flow of HFE7100 and FC-87. The microgap cooler is 125 mm long, 14 mm wide, and was operated with three distinct gap sizes: 100, 200, and 500 micron. An instrumented Intel thermal test vehicle (TTV) flip-chip mounted via a BGA on an organic substrate, and equipped with 9 pre-calibrated temperature sensors, was used as the heated section of the microgap channel. Pressure drop across the channel, fluid inlet and exit temperature, and wall temperature were measured. Using commercial software, an "inverse" numerical technique was developed to identify the local heat flux and heat transfer coefficient. Local Annular heat transfer coefficients, for FC-87 flowing in the 100 micron channel, were found to display elements of the M-shaped variation with flow quality and reached a maximum value of 15 kW/m^2-K.