New electronic devices are faster than ever before, incorporate a higher level of integration, and as a result, need to dissipate higher heat fluxes. Active cooling is the only possible method of thermal management for these devices. A new type of microchannel heat sink has been developed and evaluated in this study. The device consists of silicon microchannels on whose bottom surfaces multi-walled carbon nanotubes are grown. The objective of the study is to investigate the effect of carbon nanotubes on the heat transfer characteristics. The heat sink size is 15 mm by 15 mm by 0.675 mm. It contains two microchannel designs. One consists of eight channels of cross section 682 micrometers by 50 micrometers; the other has six channels of cross section 942 micrometers by 50 micrometers. The heat sink is incorporated in an open loop flow facility, with water as the coolant. Six different configurations are compared. Two have no nanotubes, two have closely spaced nanotubes, while the last two designs have widely spaced nanotubes. The tests utilize an infrared camera as well as thermocouples placed in the flow for characterization. The heat transfer characteristics are compared for the different cases.

Thesis(M.S.)--Auburn University, 2006. / Abstract. Vita. Includes bibliographic references.

Flow boiling in integrated microchannel systems is a cooling technology that has received significant attention in recent years as an effective option for high heat flux microelectronic devices as it provides high heat transfer and small variations in surface temperature. However, there are still a number of issues to be addressed before this technology is used for commercial applications. Amongst the issues that require further investigation are the two-phase heat transfer enhancement mechanisms, the effect of channel geometry on heat transfer characteristics, two-phase flow instabilities, critical heat flux and interfacial liquid-vapour heat transfer in the vicinity of the wall. This work is an experimental study on two-phase flow boiling in multi- and single-rectangular microchannels. Experimental research was performed on the effect of the channel aspect ratio and hydraulic diameter, particularly for parallel multi-microchannel systems in order to provide design guidelines. Flow boiling experiments were performed using deionised water in silicon microchannel heat sinks with width-to-depth aspect ratios (a) from 0.33 to 3 and hydraulic diameters from 50 μm to 150 μm. The effect of aspect ratio on two-phase flow boiling local heat transfer coefficient and two-phase pressure drop was investigated as well as the two-phase heat transfer coefficients trends with mass flux for the constant heat fluxes of 151 kW m-2, 183 kW m- 2, 271 kW m-2 and 363 kW m-2. Wall temperature measurements were obtained from five integrated thin nickel film temperature sensors. An integrated thin aluminium heater enabled uniform heating with a small thermal resistance between the heater and the channels. The microfabricated temperature sensors were used with simultaneous high-speed imaging and pressure measurements in order to obtain a better insight related to temperature and pressure fluctuations caused by two-phase flow instabilities under uniform heating in parallel microchannels. The results demonstrated that the aspect ratio of the microchannels affects flow boiling heat transfer coefficients. However, there is not clear trend of the aspect ratio on the heat transfer coefficient. Pressure drop was found to increase with increasing aspect ratio. Wide microchannels but not very shallow, with a = 1.5 and Dh = 120 μm, have shown good heat transfer performance, by producing modest two-phase pressure drop of maximum 200 mbar for the highest heat flux and heat transfer coefficients of 200 kW m-2 during two-phase flow boiling conditions. For the high aspect ratio, values of 2 and 3 two-phase flow boiling heat transfer coefficients were measured to be lower compared to aspect ratio of 1.5. Microchannels with aspect ratios higher than 1.5 produced severe wall temperature fluctuations for high heat fluxes that periodically reached extreme wall temperature values in excess of 250 ˚C. The consequences of these severe wall temperature and pressure fluctuations at high aspect ratios of 2 and 3 resulted in non-uniform flow distribution and temporal dryout. Abrupt increase in two-phase pressure drop occurred for a > 1.5. The effect of the inlet subcooling was found to be significant on both heat transfer coefficient and pressure drop. Furthermore, the effects of bubble growth on flow instabilities and heat transfer coefficients have been investigated. Although the thin film nickel sensors provide the advantage of much faster response time and smaller thermal resistance compared to classic thermocouples, they do not allow for full two-dimensional wall temperature mapping of the heated surface. An advanced experimental method was devised in order to produce accurate two-dimensional heat transfer coefficient data as a function of time. Infrared (IR) thermography was synchronised with simultaneous high-speed imaging and pressure measurements from integrated miniature pressure sensors inside the microchannel, in order to produce two-dimensional (2D) high spatial and temporal resolution two-phase heat transfer coefficient maps across the full domain of a polydimethylsiloxane (PDMS) microchannel. The microchannel was characterised by a high aspect ratio (a = 22) and a hydraulic diameter of 192 μm. The PDMS microchannel was bonded on a transparent indium tin oxide (ITO) thin film coated glass. The transparent thin film ITO heater allowed the recording of high quality synchronised high - speed images of the liquid-vapour distribution. This work presents a better insight into the two-phase heat transfer coefficient spatial variation during flow instabilities with two-dimensional heat transfer coefficient plots as a function of time during the cycles of liquid-vapour alternations for different mass flux and heat flux conditions. High spatial and temporal resolution wall temperature measurements and pressure data were obtained for a range of mass fluxes from 7.37 to 298 kg m-2 s-1 and heat fluxes from 13.64 to 179.2 kW m-2 using FC-72 as a dielectric liquid. 3D plots of spatially averaged two-phase heat transfer coefficients at the inlet, middle and outlet of the microchannel are presented with time. The optical images were correlated, with simultaneous thermal images. The results demonstrate that bubble growth in microchannels differs from macroscale channels and the confinement effects influence the local two-phase heat transfer coefficient distribution. Bubble nucleation and axial growth as well as wetting and rewetting in the channel were found to significantly affect the local heat transfer physical mechanisms. Bubble level heat transfer coefficient measurements are important as previous researchers have experimentally investigated local temperature and high speed visualisation in bubbles during pool boiling conditions and not flow boiling. The effect of the confined bubble axial growth to the two-phase heat transfer coefficient distribution at the channel entrance was investigated at low mass fluxes and low heat fluxes. The 3D plots of the 2D two-phase heat transfer coefficient with time across the microchannel domain were correlated with liquid-vapour dynamics and liquid film thinning from the contrast of the optical images, which caused suspected dryout. The 3D plots of heat transfer coefficients with time provided fine details of local variations during bubble nucleation, confinement, elongated bubble, slug flow and annular flow patterns. The correlation between the synchronised high-resolution thermal and optical images assisted in a better understanding of the heat transfer mechanisms and critical heat flux during two-phase flow boiling in microchannels.

This paper investigates the role of end-effects in superhydrophobic microchannels for frictional reduction through COMSOL based modeling. Two precursor derivations, the Kim & Hidrovo and Enright model are discussed and expanded upon through analytical and numerical simulations. The author performed numerical models on superhydrophobic microchannels with planar, stationary and finite separation distance of surface roughness element with perfect Cassie-Baxter air-layers. The simulations indicate an asymptotic limit for the flow-rate, indicating an optimum air-layer thickness. Numerical post processing reveals that this phenomenon is due to the recirculation end-effects that are relevant when the surface roughness separation distance is on order of magnitude of the channel width. These results are the first that identify end-effects as inducing a plateauing flow-rate and can serve as a benchmark for future studies. / text

A comparative study was conducted to determine whether it would be advantageous to grow carbon nanotubes on the bottom surface of anisotropically-etched silicon microchannels to facilitate greater heat removal in electronic cooling applications. The effect of the samples was evaluated based on the fluid temperature rise through the channels, the silicon surface temperature increase above ambient, and the pressure drop. The height and deposition pattern of the nanotubes were the parameters investigated in this study. The working fluid, water, was passed through the microchannels at two different volumetric flow rates (16 mL/min and 28 mL/min). Additionally, two different heat fluxes were applied to the backside of the microchannel (10 W/cm2 and 30 W/cm2). Extensive validation of the baseline channels was carried out using a numerical model, a resistor network model, and repeatability tests. Finally, the maximum enhancement when using carbon nanotubes under single-phase, laminar, internal, forced convection was investigated using basic principles in regard to the additional surface area created by the carbon nanotubes, as well as their high thermal conductivity. For the devices tested, the samples with carbon nanotubes not only had a higher pressure drop, but also had a higher surface temperature. Therefore, the baseline samples had the best performance. Furthermore, based on a basic principles investigation, the increase to thermal performance gained by increasing the surface area with CNTs is overshadowed by the decrease in mass flow rate for a fixed pressure drop. The analysis suggests that the limiting factor for heat transfer in single-phase, laminar pressure driven flows is not convection heat transfer resistance, but the bulk resistance of the fluid.

With the growing power dissipation and more densely packed circuits, the issue of efficient thermal management has become crucial. The safe and reliable operations of microchips have a requirement on a junction temperature below 85??C. In order to meet the heat dissipation requirement at the level of 1 MW m???? of the next generation microchips, a new cooling approach has been proposed by combining the merits from forced convection in the microchannel and the synthetic jet impingements. A parametric study has been carried out on the operating conditions on the synthetic jet actuator, these parameters including: the frequency of the diaphragm in the actuator, the jet outlet velocity both in magnitude and the wave shape as well as the pressure difference between the channel two ends. It was found that these parameters have combined effect on the flow structure as well as the heat transfer rate in the microchannel. When the average jet velocity was at 2.36 ms??¹(Rej= 130), with a fixed pressure difference at 750 Pa, the maximum temperature in the silicon wafer has been reduced to about 343 K at both 560 and 1120 Hz, which was 2 K lower than when 280 Hz was used. However when the average jet velocity was increased by 50 %, the optimal heat transfer then occurred at 1120 Hz, the maximum temperature was reduced to 337 K, with 4 K and 5K difference of 280 and 560 Hz, respectively. Furthermore when the average jet velocity was doubled from Rej= 130, the frequency at 280 Hz achieved the lowest maximum temperature in the wafer at 336 K that was 5 K and 3 K lower than 560 Hz and 1120 Hz. The flow temperature in the actuator is an important factor which affects the heat transfer in the microchannel. In order to lower the cavity temperature and avoid the ingestion of the already mixed flow, the time portion of the ingestion and ejection phases has been altered, by reducing the ejection time and increasing the ingestion time. However this approach did not show any significant effect in the heat transfer process or decreasing the flow temperature in the cavity. However in a later study by increasing the pressure difference across the channel, the flow temperature in the cavity has been substantially reduced and the heat transfer in the channel changed significantly according to the flow structure. It was found that the high pressure in the channel could deliver the vortical structure to the hotter part of the wafer thus decreasing the maximum temperature in the silicon effectively, especially when high jet velocity was used. When high jet velocity has been used, irregular variation of the flow was found The unrepeatable feature of the flow is related to the frequency, jet velocity as well as the channel pressure difference.

This project will investigate mixing in microchannels. Specifically, the advection and diffusion of a passive scalar, using a split step Monte Carlo method. Numerically the implementation of this method is well understood. The current experimental geometry is a rectangular pipe with grooves on one wall. Mixing results with straight walls agree closely with experiment. The velocity field over grooves is also studied.

A cooling system for high heat flux applications is examined using microchannel evaporators with water as the working fluid and boiling as the heat transfer mechanism. Experimental studies are performed using single channel microevaporators allowing for better control of the flow mechanics unlike other investigations where multiple, parallel, flow channels can result in a non-uniform distribution of the working fluid. High-speed flow visualizations are performed in conjunction with heat transfer and pressure drop measurements to support the quantitative experimental data. Flow patterns associated with a range of boundary conditions are characterized and then presented in the form of novel flow regime maps that intrinsically reflect the physical mechanisms controlling two-phase pressure distributions and heat transfer behavior. Given the complexity associated with modeling of boiling heat transfer and the lack of a universal model that provides accurate predictions across a broad spectrum of flow conditions, flow regime maps serve as a valuable modeling aid to assist in targeted modeling over specific flow regimes. This work represents a novel and original contribution to the understanding of boiling mechanisms for water in microchannels. The flow patterns in this work are found to be closely coupled with mass flux, heat flux, and channel size; where re-wetting and pressure oscillations play a crucial role, and are likely responsible for its development and evolution. Reversed flow, typically attributed to a non-uniform fluid distribution in multiple channel microevaporators by other researchers, is shown to be a result of the upstream expansion of confined bubbles. During flow boiling, the pressure drop in the microchannel increases with the heat flux for a constant flow rate due to the significant acceleration effects associated with smaller channels, unlike in single-phase flow where the pressure drop is constant. Water flow boiling in rectangular microchannels, although not extensively explored in the published literature, provides an extremely high cooling capacity, with dissipation rates approaching 440 W/cm², making this an ideal candidate for cooling of next generation electronic systems. Single-phase flow studies revealed that pressure and heat transfer coefficient macroscale models are transferable to microchannels with hydraulic diameters down to 200 µm, when the entrance effects and minor losses are properly considered. These studies include laminar developing flow conditions not commonly considered in the literature and fully developed flow. Since the applicability of macroscale theories to microchannels is often questioned, this investigation helps clarify this issue for microchannels within the range of experimental conditions explored in this work. Finally, new correlations for the hydrodynamic entrance length are proposed for rectangular microchannels and good agreement is found when compared with published experimental data over a wide range of Reynolds number. These correlations are more accurate, and original in the sense that they incorporate the effects of channel aspect ratio, and include creeping flow conditions which are currently unavailable for rectangular microchannels. This work represents a major advance in the development of new cooling systems for high heat flux applications requiring dissipation rates in excess of 100 W/cm².

An experimental and analytical study of a microchannel ammonia-water desorber was conducted in this study. The desorber consists of 5 passes of 16 tube rows each with 27, 1.575 mm outside diameter x 140 mm long tubes per row for a total of 2160 tubes. The desorber is an extremely compact 178 mm x 178 mm x 0.508 m tall component, and is capable of transferring the required heat load (~17.5 kW) for a representative residential heat pump system. Experimental results indicate that the heat duty ranged from 5.37 kW to 17.46 kW and the overall heat transfer coefficient ranges from 388 to 617 W/m2-K. The analytical model predicts temperature, concentration and mass flow rate profiles through the desorber, as well as the effective wetted area of the heat transfer surface. Heat and mass transfer correlations as well as locally measured variations in the heating fluid temperature are used to predict the effective wetted area. The average wetted area of the heat and mass exchanger ranged from 0.25 to 0.69 over the range of conditions tested in this study. Local mass transfer results indicate that water vapor is absorbed into the solution in the upper stages of the desorber leading to higher concentration ammonia vapor and therefore reducing the rectifier cooling capacity required. These experimentally validated results indicate that the microchannel geometry is well suited for use as a desorber. Previous experimental and analytical research has demonstrated the performance of this microchannel geometry as an absorber. Together, these studies show that this compact geometry is suitable for all components in an absorption heat pump, which would enable the increased use of absorption technology in the small capacity heat pump market.

Over the last few decades, miniaturization of the product became a necessity for many industries to achieve successful technological development, satisfy customer needs, and stay economically competitive in the market. Thus, many medical, aerospace, and electronic devices tend to decrease in size. Along with the strong demand for miniaturization, new cutting-edge micromanufacturing techniques are developing in order to produce microcomponents with a smooth surface finish and high dimensional accuracy. In the medical industry, some devices require manufacturing of fluidic microchannels on biocompatible materials for transportation of exact amount of medicine to a defined location. Often such microchannels must be manufactured to achieve a high aspect ratio, a submicron surface finish, and an anisotropic controlled profile. The fabrication of such channels on biocompatible materials still poses a challenge. This study developed micromanufacturing technique to produce the microchannels and satisfy all the requirements listed above. Computer controlled micromilling on a high speed machine system in minimum quantity lubrication was used to remove most materials and define a channel pattern. Microchannels were machined with ball end mills of diameters from Ø152μm to Ø198μm on NiTi alloy, 304 and 316L stainless steels. Assessment of microchannel was performed with optical microscopy, scanning electron microscopy, and white light interferometry. The theoretical surface roughness in ball end milling was derived using geometrical approach. The theoretical surface finish model was compared and validated with the experimental surface finish data. Meso- and macro-scale milling confirmed the validity of the model, but surface finish in micro-scale milling was measured to be a few orders of magnitude higher due to size effect and build-up edge. The build-up-edge was reduced when using AlTiN coated tools and milling in minimum quantity lubrication. The empirical surface roughness model obtained in this study shows the dependence of surface finish on chip load in micromilling. In order to further enhance the surface finish of milled microchannels additional finishing technique was identified. A separate study developed an effective electrochemical polishing technique to remove burrs and enhance surface finish of milled microchannels. When applying to 304, 316L stainless steel alloys and NiTi alloy, this hybrid technique can repeatedly produce microchannels with an average surface finish less than 100nm.