Microchannel Size Effects on Two-Phase Local Heat Transfer and Pressure Drop in Silicon Microchannel Heat Sinks with a Dielectric Fluid

Transcription

1 Purdue University Purdue e-pubs Birck and NCN Publications Birck Nanotechnology Center -- Microchannel Size Effects on To-Phase Local Heat Transfer and Pressure Drop in Silicon Microchannel Heat Sinks ith a Dielectric Fluid Tannaz Harirchian Birck Nanotechnology Center, Purdue University Suresh V. Garimella Birck Nanotechnology Center, Purdue University, sureshg@purdue.edu Follo this and additional orks at: Part of the Nanoscience and Nanotechnology Commons Harirchian, Tannaz and Garimella, Suresh V., "Microchannel Size Effects on To-Phase Local Heat Transfer and Pressure Drop in Silicon Microchannel Heat Sinks ith a Dielectric Fluid" (). Birck and NCN Publications. Paper This document has been made available through Purdue e-pubs, a service of the Purdue University Libraries. Please contact epubs@purdue.edu for additional information.

2 Proceedings of IMECE ASME International Mechanical Engineering Congress and Exposition November -,, Seattle, Washington, USA IMECE- Microchannel Size Effects on To-Phase Local Heat Transfer and Pressure Drop in Silicon Microchannel Heat Sinks ith a Dielectric Fluid Tannaz Harirchian and Suresh V. Garimella School of Mechanical Engineering and Birck Nanotechnology Center Purdue University West Lafayette, IN 9- USA ABSTRACT To-phase heat transfer in microchannels can support very high heat fluxes for use in high-performance electronics-cooling applications. Hoever, the effects of microchannel crosssectional dimensions on the heat transfer coefficient and pressure drop have not been investigated extensively. In the present ork, experiments are conducted to investigate the local flo boiling heat transfer in microchannel heat sinks. The effect of channel size on the heat transfer coefficient and pressure drop is studied for mass fluxes ranging from to kg/m s. The test sections consist of parallel microchannels ith nominal idths of,,,, and µm, all ith a depth of µm, cut into. mm. mm silicon substrates. Tenty-five microheaters embedded in the substrate allo local control of the imposed heat flux, hile tenty-five temperature microsensors integrated into the back of the substrates enable local measurements of temperature. The dielectric fluid Fluorinert FC- is used as the orking fluid. The results of this study serve to quantify the effectiveness of microchannel heat transport hile simultaneously assessing the pressure drop trade-offs. Keyords Microchannel, boiling, dielectric liquid, size effect INTRODUCTION Boiling in microchannel heat sinks is extremely attractive for electronics cooling due to the high heat transfer rates that can be achieved, hile at the same time, the all temperature is maintained relatively uniform. Many studies in recent years have attempted to better understand the flo patterns of boiling in microchannels using different orking fluids [, ]. Peng and Wang [3] experimentally investigated singlephase convection and flo boiling of ater in microchannels of cross section µm. They shoed that unlike for conventional sized tubes, partial nucleate boiling as not observed in the subcooled region and fully developed boiling as induced much earlier than at the macroscale. Ke and Cornell [] experimentally studied to-phase flo and heat transfer ith refrigerant Rb in circular tubes of diameter.39 to 3.9 mm. They proposed a threshold hydraulic diameter belo hich conventional correlations ere not suitable for the prediction of flo boiling heat transfer. Qu and Mudaar [] studied convective boiling heat transfer, to-phase flo patterns, and pressure drop in parallel microchannels of 3 µm idth and µm depth using ater as the orking fluid. They argued that unlike in the case of fluorochemicals, it is difficult to sustain bubbly flo regimes in boiling of ater in microchannels because of the high surface tension and large contact angle of ater; a slug flo regime as found to develop shortly after incipience of boiling. Also, the flo patterns ere strongly influenced by the applied heat flux. Jiang et al. [] carried out experiments ith ater in triangular silicon microchannels of hydraulic diameter and 3 µm and observed that annular flo developed at a relatively lo heat flux, such that evaporation at the liquid film/vapor core interface as the dominant heat transfer mode over a ide range of input poers. Hence, no boiling plateau as observed in the boiling curve and the all temperature increased as the heat flux as increased. Use of dielectric liquids in microchannel heat sinks has dran recent attention since the orking fluid in the microchannel heat sinks can be in direct contact ith the electronic chips. Although there have been a number of studies on pool boiling of perfluorocarbon liquids and the effect of surface enhancement [], investigations of flo boiling in microchannels using perfluorocarbon liquids have been limited. Chen and Garimella [] conducted experiments to study the physics of boiling in parallel silicon microchannels ith a cross section of 39 µm 39 µm using FC- as the orking fluid. They tested three different flo rates and observed bubbly and Copyright by ASME Donloaded From: on //3 Terms of Use:

3 slug flo patterns at loer heat fluxes and ispy-annular and churn-flo regimes at higher heat fluxes. Zhang et al. [9] performed an experimental study of flo boiling of FC- for three different orientations of a microchannel heat sink: vertical up-flo, vertical don-flo, and horizontal flo. The heat sink consisted of parallel microchannels µm ide and mm deep. They observed boiling to occur in isolated bubble flo and slug flo regimes. Annular flo as not observed due to the short length of the channels. Warrier et al. [] performed singlephase and boiling experiments in horizontal parallel aluminum microchannels of hydraulic diameter µm ith FC- as the test fluid. They tested different mass flo rates and inlet subcooling and proposed correlations for both pressure drop and heat transfer. The effect of flo rate on microchannel flo boiling has been considered in a number of studies. Hoever, fe have arrived at clear conclusions on the effect of mass flux on boiling heat transfer and pressure drop. Chen and Garimella [] compared the boiling heat transfer coefficients and pressure drops for flo rates of, 39, and kg/m s. Liu and Garimella [] carried out an experimental study of flo boiling heat transfer in microchannels ( µm 3 µm) cut into a copper block, using DI ater ith mass fluxes ranging from 3 to 93 kg/m s. Chen and Garimella [] performed an experimental study of flo boiling of FC- in a copper microchannel ( channels, µm. mm) heat sink at flo rates ranging from to 33 kg/m s. Pate et al. [3] explored to-phase heat transfer in parallel microchannels ith hydraulic diameter of 3 µm using FC- as the test fluid. Six re-entrant type cavities, spaced evenly on the base of each channel, ere used to promote controlled nucleation. Mass fluxes in the range of 3-3 kg/m s ere considered. All these studies [,,, 3] shoed that beyond the onset of nucleate boiling, the boiling curves collapsed on a single curve for all mass flo rates; as the heat flux as increased further, the curves diverged and became dependent on mass flux. Lin et al. [] performed an experimental study of to-phase flo of refrigerant Rb in circular tubes of diameter of.,., 3. mm and one square tube of mm ith different mass fluxes ranging from to 3 kg/m s. Both nucleate and convective boiling mechanisms ere said to occur in the tubes and the local heat transfer coefficient as found to be a eak function of mass flux hile the mean heat transfer coefficient as independent of mass flux. Although a large number of studies of microchannel flo boiling have been reported, fe have systematically explored the effect of microchannel dimensions on the thermal transport. While a fe studies have considered the effect of microchannel size on flo boiling patterns and the transition beteen different flo regimes, the effect of microchannel size on heat transfer coefficient and pressure drop has been largely unexplored. Lin et al. [] shoed that the transition from nucleate to convective boiling at high heat flux occurred at higher qualities in the smaller diameter tubes. In the absence of dryout in the saturation boiling region, they found that the mean heat transfer coefficient varied only slightly ith the tube diameter and as mainly a function of heat flux. Zhang et al. [] conducted experiments in single silicon microchannels of - µm hydraulic diameters and different surface roughnesses ith DI ater to study bubble nucleation, flo patterns, and transient pressure fluctuations. They shoed that in channels smaller than µm in hydraulic diameter, the bubble nucleation mechanism as eruption boiling and that mist flo developed almost right after single-phase flo ended because of the large amount of all superheat. Hence, no temperature plateau as observed. The boiling mechanism as found to be determined by the all surface conditions rather than the channel dimensions. Bergles et al. [] also performed an analysis and shoed that incipience of boiling in a subcooled flo as most likely governed by the nucleation sites in the alls and not by the channel size. Saitoh et al. [] experimentally investigated the boiling heat transfer of refrigerant R-3a flo in three horizontal tubes of diameter.,., and 3. mm and mass fluxes ranging from to kg/m s. Their study shoed that the local heat transfer coefficient increased ith increasing mass flux in larger tubes but as not significantly affected by mass flux in smaller tubes. The heat transfer coefficient increased ith increasing heat flux in all three tubes. The contribution of forced convective evaporation to the boiling heat transfer decreased ith decreasing tube diameter. Lee et al. [] investigated the effect of microchannel height on nucleationsite activity and bubble dynamics using three fluids: ater, methanol and ethanol. The heat sink consisted of ten parallel microchannels fabricated in a silicon afer, ith idth ranging from to 9 µm and height ranging from to µm. They found that the bubble nucleation activity as dependent on channel height for all fluids tested. Lee et al. [9] explored the effect of channel shape on to-phase flo patterns in DI ater. The microchannel heat sink they tested consisted of ten shallo, nearly rectangular silicon microchannels of µm idth and µm depth. They studied bubble dynamics and to-phase flo patterns and compared the results to similar ork by Jiang et al. [] in triangular microchannels ith hydraulic diameter of µm. Dupont and Thome [] studied the effect of diameter on flo boiling heat transfer and transition from macro- to microchannel evaporation using a three-zone flo boiling model based on evaporation of elongated bubbles in microchannels. The model predicted an increase in heat transfer coefficient ith a decrease in diameter for lo values of vapor quality and a decrease in heat transfer coefficient for large vapor qualities. In vie of the absence of any systematic studies in the literature on the effect of microchannel dimensions on heat transfer coefficient and pressure drop, the objective of the present ork is to investigate the effect of mass flo rate and microchannel size on boiling heat transfer ith a dielectric fluid, FC-. Boiling curves, heat transfer coefficients, and pressure drops are presented and compared for a range of flo rates and Copyright by ASME Donloaded From: on //3 Terms of Use:

4 microchannel sizes, and design criteria for microchannel heat sinks discussed. NOMENCLATURE A A b,c C heat sink base area total etted area of microchannels c constants in heat loss-temperature relation p G h specific heat of FC- mass flux heat transfer coefficient h latent heat of FC- fg H k si L ṁ N P p q net loss microchannel height thermal conductivity of silicon microchannel length mass flo rate number of microchannels in a heat sink pumping poer pressure drop heat dissipated to the fluid q heat loss q q b q R t T T heat dissipated from the heat sources base heat flux all heat flux d in sat resistance heat sink thickness temperature measured by diode sensors inlet fluid temperature T fluid saturation temperature T V x η e f microchannels bottom all temperature voltage microchannel idth fin idth exit vapor quality fin efficiency of microchannel heat sink EXPERIMENTAL SETUP AND PROCEDURES Flo loop A schematic diagram of the experimental test loop is shon in Figure. A magnetically coupled gear pump drives the dielectric liquid, FC-, through the closed loop. A preheater installed upstream of the test section heats the coolant to the desired subcooling temperature at the entrance of the microchannels, and a liquid-to-air heat exchanger located donstream of the test section cools the fluid before it enters the reservoir. The liquid is fully degassed before initiating each test using the to degassing ports and the expandable reservoir. Details of the expandable reservoir design and the degassing procedure are available in Chen and Garimella []. A flo meter ith a measurement range of - ml/min monitors the flo rate through the loop and five T-type thermocouples measure the fluid temperature before and after the preheater, before and after the test section, and after the heat exchanger. The pressure in the outlet manifold of the test section is maintained at atmosphere. The pressure at the inlet manifold and the pressure drop across the microchannel array are measured using a pressure transducer (Gems Sensors, series) and a differential pressure transducer (Omega, PX3 series), respectively. Degassing port P T Expandable reservoir Filter Heat Exchanger Pump T T3 Test section dp P Filter Flo Meter P3 Preheater T T P Figure. Schematic illustration of the flo loop. Test Section Fluid outlet Pressure ports Temperature ports Fluid inlet Microchannel heat sink Figure. Microchannel test section. P Degassing port 3 Copyright by ASME Donloaded From: on //3 Terms of Use:

5 The test section shon in Figure consists of a. mm. mm silicon microchannel heat sink mounted on a printed circuit board (PCB). The PCB is installed on a quick-connect board ith an insulating G piece in beteen. A polycarbonate top cover above the test piece provides an enclosed passage for the liquid and is sealed ith an o-ring. To avoid melting of the polycarbonate at high die temperatures, a. mm. mm Pyrex sheet of thickness. mm ith a high melting point is sandiched beteen the silicon die and the top cover and forms the top all of the microchannels. As detailed in Figure 3, parallel microchannels are cut on the top surface of the silicon chip using a dicing sa. The idth, depth, and number of microchannels in each heat sink are listed in Table. The bottom all of the µm ide microchannels has an average roughness of. µm. Wider microchannels ere made ith a number of cuts; this process imparts a aviness to the bottom surfaces resulting in an average roughness of. to. µm for the different test pieces; the average roughness in the region of a single cut is. µm. Flo Flo Tenty five individual current sources are used to apply µa to each diode and the voltage drop across each diode is measured at seven temperatures ranging from C to C. The calculated local heat transfer coefficients presented in this paper are based on measurements from the temperature sensor at location 3 in Figure 3, hich is along the centerline of the test piece near the exit. Table. Microchannel dimensions and mass fluxes (the four mass fluxes are referred to in the rest of the paper by the nominal values of,, and kg/m s). (µm) (nominal values). () 39. () 9. ().3 () 9. () f (µm) (nominal values).9 (). (). (). ().3 () H (µm) (nominal values) 39. () 3. () 3.9 () 3. () 33. () N G (kg/m s) 3,,,,,, 9,,,,,,,,, Maximum x e (%),, 3, 93,,, 9, 9,, 3,,, 9, 9,, More details of the individual heaters, diode temperature sensors, calibration procedures, test section assembly, and test loop are available in Chen and Garimella []. Not to scale Figure 3. Integrated heaters and temperature sensors in the microchannel test piece. A array of resistance heat sources and a like array of temperature sensing diodes are fabricated on the other side of the chip, providing uniform heat flux to the back side of the microchannels and local measurements of the base temperature. The resistance of each heating element is measured at temperatures up to 3 C. Since the resistances of all the heat sources are almost identical, they are connected in parallel and are supplied ith a single DC voltage in order to provide a uniform heat flux to the back of the microchannels. The heat generated by each source is obtained from the calibrated resistance of the corresponding element and the applied voltage. For a given current passing through a diode temperature sensor, the voltage drop across the diode depends on the temperature. To obtain the voltage-temperature relationship for each diode, the test piece is calibrated in a conventional oven. Experimental Procedures Experiments are conducted for five test pieces ith a constant channel depth of µm and different channel idths ranging from µm to µm to study the effect of microchannel idth on the boiling heat transfer coefficients and pressure drops as a function of mass flux; four mass flux values ranging from to kg/m s are investigated for each test piece to map the effect of flo velocity on boiling. The actual and nominal dimensions of the test pieces and the flo rates are summarized in Table. Before initiating each test, the liquid in the test loop is fully degassed. It is then driven into the loop at a constant flo rate and preheated to approximately 9 C, providing C of subcooling at the inlet of the channels. For each test, the flo rate and the inlet fluid temperature are kept constant throughout the test and the heat flux to the chip is increased from zero to the point at hich the maximum all temperature reaches C, hich is a safe operating temperature for the integrated heaters and temperature sensors. Heat flux values approaching critical heat flux are avoided since the corresponding temperatures could cause the solder bump s in the test chip to melt. Copyright by ASME Donloaded From: on //3 Terms of Use:

6 DATA REDUCTION The heat transfer rate to the fluid, q net, in the microchannels is obtained from the energy balance for each heating element: q = q q () net in hich q is the total heat dissipated from each heat source and is obtained from q = V / R loss, here V is the applied voltage to the heating element and R is the calibrated resistance. The heat loss, q loss, consists of losses through natural convection, radiation, and conduction through the PCB and the top cover and is determined as follos. Before the test section is charged ith liquid, a constant voltage is applied to the heaters. When the readings of the diode temperature sensors reach a steady state, the temperature of each sensor is recorded and correlated to the heat dissipated from the corresponding heater at that location. This procedure is repeated for several levels of input poer and a linear relation in the form of q loss = ct d + c is obtained, here c and c are constants and are slightly different at each location and for different test pieces. The local heat transfer coefficient is then calculated from here h = T q T T is the local microchannel all temperature and is corrected for the microchannel base thickness using T sat ( ) q t H () b = Td (3) ksilicon here T is the diode temperature at location 3 in Figure 3, and d t is the total thickness of the silicon die. The heat flux used in Eq. () is the all heat flux and is defined as here ( ) q = q / A / () net A is the total etted area of the microchannels ( η ) A = N + H L () The heat flux used in finding the all temperature in Eq.(3) is the applied heat flux and is calculated using the chip base area, A, hich is the same as the combined area of all the heat sources ( ) q = q / A / () b net b The exit vapor quality is calculated from an energy balance as follos q x C T T net e = p sat in h fg m ( ) b () Measurement Uncertainties The measurement uncertainties for the flo meter and the pressure transducers are % and.% of full scale, respectively. The uncertainties in the measurements of the channel dimensions, the T-type thermocouples and the diode temperature sensors are ± µm, ±.3 C and ±. C, respectively. Folloing a standard uncertainty analysis [], the uncertainties associated ith the all heat flux and the heat transfer coefficient are obtained to be to % and to.%, respectively, for the cases considered. RESULTS AND DISCUSSION Boiling Curve Figure shos the boiling curves at four different mass fluxes for the test piece ith microchannels of idth µm. Both single-phase and to-phase regions are included in this plot. As can be seen from this figure, the onset of boiling is associated ith a sharp drop in the all temperature. For sufficiently small heat flux increments, this temperature overshoot as observed ith all the test pieces and at all mass fluxes. The maximum all temperature drop observed as C for the µm ide microchannels, at a mass flux of kg/m s. This is consistent ith the results obtained by Pate et al. [3], ho also observed that the all temperature overshoot at the incipience of boiling decreased ith increasing mass flux. q" 3 3 = µm kg/m s kg/m s kg/m s kg/m s 9 3 T (C) Figure. Effect of mass flux on boiling curves. Beyond the onset of nucleate boiling, curves for all mass fluxes collapse to a single curve, indicating the dominance of nucleate boiling. At higher heat fluxes, as the convective heat transfer starts to dominate, the curves diverge and, as in the single-phase region, more heat is dissipated as mass flux increases at a fixed all temperature. Figure also shos that Copyright by ASME Donloaded From: on //3 Terms of Use:

7 the onset of nucleate boiling occurs at a loer temperature and a loer heat flux for a loer mass flux. These trends in the boiling curve and the influence of mass flux shon for the µm channels ere also observed ith the other four channel sizes tested. The trends are also consistent ith the results of Liu and Garimella [], Chen and Garimella [], and Pate et al. [3]. At very lo heat fluxes, the boiling curve for the kg/m s mass flux deviates from the others. This is likely due to very high vapor quality in this case, and a possible transition to a convective heat transfer regime at a loer heat flux. This deviation as less noticeable for µm microchannels since the vapor quality at kg/m s mass flux is loer for this large microchannel size. The effect of microchannel size on the boiling curve for a fixed mass flux of kg/m s is shon in Figure. For microchannels of idth µm and larger, the boiling curves collapse to a single curve beyond the onset of nucleate boiling. As boiling starts in these microchannels, the all temperature has a eak dependency on the heat flux and is relatively constant; hoever, at high heat fluxes the all temperature becomes more dependent on heat flux and the boiling curves deviate for the different sizes. q" 3 µm µm µm µm G = kg/m s µm 9 3 T (C) Figure. Effect of microchannel idth on boiling curves. For the microchannels of idth and µm, the all temperature increases ith increasing all heat flux and the boiling curves do not collapse on those of the larger microchannels. This may be attributed to the early establishment of annular flo in microchannels of very small diameter [,, and ]. The effect of microchannel size on the boiling curves seen in Figure as also observed in the experiments conducted ith other mass fluxes. Heat Transfer Coefficient Figure illustrates the effect of mass flux on the heat transfer coefficient as a function of the all heat flux. The heat transfer coefficient increases ith mass flux in the single-phase region for a fixed all heat flux. After the onset of nucleate boiling, hoever, the heat transfer coefficient becomes independent of mass flux, and increases ith heat flux. At high levels of all heat flux, as the convective heat transfer dominates the nucleate boiling, the heat transfer coefficient becomes a function of mass flux and increases ith increasing mass flux. Other microchannel sizes tested yielded similar trends for the dependence of heat transfer coefficient on mass flux. These results regarding the dependence of heat transfer coefficient on the flo rate are also consistent ith the findings of Chen and Garimella [, ]. At high heat fluxes, a decrease in heat transfer coefficient is detected that may be attributed to a flo regime transition beteen churn flo and ispy annular flo hich leads to a partial all dryout []. h (kw/m K) 9 3 = µm kg/m s kg/m s kg/m s kg/m s 3 3 q" Figure. Effect of mass flux on local heat transfer coefficient. The influence of microchannel size on the heat transfer coefficient is illustrated in Figure (a). Interestingly, the heat transfer coefficient is independent of microchannel idth for channels of idth µm and larger, and has a eak dependence on channel size for smaller microchannels. At the µm idth, the heat transfer coefficients are slightly loer than for the larger sizes at any all heat flux. For the µm ide microchannels, the behavior is markedly different, ith the heat transfer coefficient being relatively higher at the loer heat fluxes. As the heat flux increases, the curve crosses over and is loer than for the larger microchannels. This is attributed to the high exit vapor quality for the µm microchannels, as illustrated in Figure. As a result of the high vapor qualities in Copyright by ASME Donloaded From: on //3 Terms of Use:

8 the µm microchannels, annular flo commences at loer heat fluxes than at the larger channel sizes, leading to higher heat transfer coefficients at the loer heat fluxes. h (kw/m K) h (kw/m K) 3 G = kg/m s µm µm µm µm µm 3 q" 3 G = kg/m s (a) µm µm µm µm µm 3 q" b (b) Figure. Effect of microchannel idth on local heat transfer coefficient as a function of: (a) all heat flux, and (b) base heat flux. Figure (a) illustrates that increasing the idth of microchannels (for a fixed channel depth) does not affect the heat transfer coefficient for a fixed all heat flux,. Hoever, from a design point of vie, the dependence of heat transfer coefficient on microchannel size should be considered in terms of a given amount of heat dissipation from the chip, i.e., a fixed value of base heat flux,. Plotted in this manner, the heat q b q transfer coefficient (as a function of base heat flux) for different microchannel sizes is presented in Figure (b). It is evident from this figure that for a given amount of heat dissipation from the chip, the heat transfer coefficient increases as the microchannels idth increases. Hoever, the maximum amount of heat that can be removed from the chip increases as the microchannels become smaller due to the larger surface enhancement ith the smaller microchannels. q" b h (kw/m K) x exit G = kg/m s µm µm µm µm µm Figure. Variation of heat transfer coefficient ith exit vapor quality for different microchannel idths. 3 µm µm µm µm G = kg/m s µm 9 3 T (C) Figure 9. Variation of base heat flux ith all temperature for different microchannel idths. Further, if the base heat flux ere plotted versus all temperature as is done in Figure 9, it is seen that for a fixed heat dissipation rate from the chip, the all temperature increases Copyright by ASME Donloaded From: on //3 Terms of Use:

9 ith channel size. In other ords, more heat can be removed from the chip at a given all temperature ith the smaller microchannels. It is emphasized that the number of microchannels incorporated into the chip base area directly affects the base heat flux value; hoever, the heat sinks tested are not optimized in terms of the number of microchannels. The trends discussed above are readily extrapolated for optimized heat sinks by correcting for the number of microchannels present. Pressure Drop The pressure drop as a function of the average all heat flux is illustrated in Figure for the µm-ide microchannels at four different mass fluxes. The to-phase region can be clearly distinguished from the single-phase region by the sharp change in slope of the curves. In the single-phase region, the pressure drop slightly decreases ith increasing heat flux due to the reduction in liquid viscosity as the liquid temperature increases. In the to-phase region, the pressure drop is strongly dependent on heat flux and increases rapidly and almost linearly ith increasing heat flux due to the acceleration of vapor. This trend in pressure drop has been reported in many studies [,,,, and 3]. In both single-phase and tophase regions, the pressure drop increases ith increasing mass flux. Pate et al. [3] obtained similar results for different mass fluxes, but Chen and Garimella [] found that pressure drop as independent of mass flux in the to-phase region. They related this observation to the balance beteen the frictional pressure drop and acceleration pressure drop because of the moderate inlet subcooling in their tests, hich is in contrast to the very modest subcooling used in the current tests. p (kpa) = µm kg/m s kg/m s kg/m s kg/m s 3 q" Figure. Effect of mass flux on pressure drop. In Figure, the pressure drop is presented for different channel sizes at a fixe d mass flux as a function of all heat flux. In both the single-phase and to-phase regions, the pressure drop increases ith decreasing microchannel idth at a given all flux. In the to-phase region, the slope of the line also increases as the channel idth decreases, resulting in much larger pressure drops for smaller channels at higher heat fluxes. P (W) p (kpa) 3 G = kg/m s µm µm µm µm µm 3 q" Figure. Effect of microchannel idth on pressure drop µm µm µm µm µm G = kg/m s 3 q" b Figure. Effect of microchannel size on pumping poer. Figure shos the pumping poer required to manage a required heat sink base heat flux ith the different microchannel sizes. In the single-phase region the pumping poer required is almost constant, independent of the heat flux, hile in the to phase region, the pumping poer increases rapidly ith heat flux. This figure also shos that for microchannels of idth µm and larger, the pumping poer is not a strong function of microchannel idth. For idths belo µm, hoever, the Copyright by ASME Donloaded From: on //3 Terms of Use:

10 pumping poer increases for a given base heat flux. Therefore, for a given pumping poer, more heat can be removed from the heat source ith larger microchannels, although, as discussed in the previous section, using the larger microchannels results in higher all temperatures. CONCLUSIONS The effect of mass flux and channel size on boiling heat transfer of FC- through microchannel heat sinks has been experimentally investigated. For a fixed channel size, boiling curves and heat transfer coefficients are independent of mass flux in the nucleate boiling region. As the heat flux is increased further and convective heat transfer becomes dominant, the boiling curves diverge and more heat is dissipated as mass flux increases for the same all temperatures. The maximum heat transfer coefficient, as ell as the all heat flux at hich this maximum occurs, increases ith increasing mass flux. Pressure drop decreases slightly ith heat flux in the single-phase region and increases rapidly ith heat flux in the to-phase region. For a fixed all heat flux, pressure drop increases ith increasing mass flux and decreasing channel idth. For microchannels of idth to µm, boiling curves are independent of channel size. For a fixe d all heat flux the heat transfer coefficient is almost independent of channel size, hile for a given heat dissipation amount from the heat source, the heat transfer coefficient increases ith increasing channel idth. Although for a fixed pumping poer the base heat flux increases ith increasing the idth of the microchannels, the maximum heat that can be removed from the chip increases ith decreasing channel idth. Also, for a given amount of heat dissipation from the chip, the all temperature is loer for smaller channels. In ongoing ork, these trends are being related to the tophase flo regimes existing under the conditions of the experiment so that regime-based heat transfer and pressure drop models may be developed. ACKNOWLEDGMENTS Financial support from members of the Cooling Technologies Research Center (.ecn.purdue.edu/ctrc), an NSF Industry/University Cooperative Research Center at Purdue University, is gratefully acknoledged. The authors thank Bruce Myers and Darrel Peugh of Delphi Electronics and Safety, Kokomo, Indiana, for providing the silicon test pieces. REFERENCES [] Garimella, S. V. and Sobhan, C. B., 3, Transport in Microchannels -a Critical Revie, Annual Revie of Heat Transfer, 3, pp. -. [] Sobhan, C. B. and Garimella, S. V.,, A Comparative Analysis of Studies on Heat Transfer and Fluid Flo in Mcrochannels, Microscale Thermophysical Engineering,, pp [3] Peng, X. F. and Wang, B. X., 993, Forced Convection and Flo Boiling Heat Transfer for Liquid Floing Through Microchannels, International Journal of Heat and Mass Transfer, 3 (), pp [] Ke, P. A. and Cornell, K., 99, Correlations for the Prediction of Boiling Heat Transfer in Small-Diameter Channels, Applied Thermal Engineering,, pp. -. [] Qu, W. and Mudaar, I.,, Transport Phenomena in To-Phase Micro-Channel Heat Sinks, Journal of Electronic Packaging,, pp. 3-. [] Jiang, L., Wong, M., and Zohar, Y.,, Forced Convection Boiling in a Microchannel Heat Sink, Journal of Microelectromechanical Systems,, pp. -. [] Honda, H. and Wei, J. J.,, Enhanced Boiling Heat Transfer from Electronic Components by Use of Surface Microstructures, Experimental Thermal and Fluid Science,, pp [] Chen, T. and Garimella, S. V.,, Measurements and High- Speed Visualization of Flo Boiling of a Dielectric Fluid in a Silicon Microchannel Heat Sink, International Journal of Multiphase Flo, 3 (), pp [9] Zhang, H. Y., Pinjala, D., and Wong T. N.,, Experimental Characterization of Flo Boiling Heat Dissipation in a Microchannel Heat Sink ith Different Orientations, Proceeding of th Electronics Packaging Technology Conference, EPTC,, pp. -. [] Warrier, G. R., Dhir, V. K., and Momoda, L. A.,, Heat Transfer and Pressure Drop in Narro Rectangular Channels, Experimental Thermal and Fluid Science,, pp. 3-. [] Liu, D. and Garimella, S. V., Flo Boiling in a Microchannel Heat Sink, ASME Journal of Heat Transfer (In press). [] Chen, T. and Garimella S. V.,, Flo Boiling Heat Transfer to a Dielectric Coolant in a Microchannel Heat Sink, IEEE Transactions on Components and Packaging Technologies, 3 (), pp. -3. [3] Pate, D. P., Jones, R. J., and Bhavnani, S. H.,, Cavity- Induced To-Phase Heat Transfer in Silicon Microchannels, 9 Copyright by ASME Donloaded From: on //3 Terms of Use:

11 Proceedings of the Intersociety Conference on Thermal and Thermomechanical Phenomena and Emerging Technologies in Electronic Systems, pp. -. [] Lin, S., Ke, P. A., and Cornell, K.,, Flo Boiling of Refrigerant RB in Small Tubes, Chemical Engineering Research and Design, 9 (), pp. -. [] Zhang, L., Wang, E. N., Goodson, K. E., and Kenny, T. W.,, Phase Change Phenomena in Silicon Microchannels, International Journal of Heat and Mass Transfer,, pp. -. [] Bergles, A. E., Lienhard, J. H., Kendall, G. E., and Griffith, P., 3, Boiling and Evaporation in Small Diameter Channels, Heat Transfer Engineering, (), pp. -. [] Saitoh, S., Daiguji, H., and Hihara, E.,, Effect of Tube Diameter on Boiling Heat Transfer of R-3a in Horizontal Small- Diameter Tubes, International Journal of Heat and Mass Transfer, (3-), pp [] Lee, M., Cheung, L. S. L., Lee, Y., and Zohar, Y.,, Height Effect on Nucleation-Site Activity and Size-Dependent Bubble Dynamics in Microchannel Convective Boiling, Journal of Micromechanics and Microengineering,, pp. -9. [9] Lee, M., Wong, Y. Y., Wong, M., and Zohar, Y., 3, Size and Shape Effects on To-Phase Flo Patterns in Microchannel Forced Convection Boiling, Journal of Micromechanics and Microengineering, 3, pp. -. [] Dupont, V. and Thome, R.,, Evaporation in Microchannels: Influence of the Channel Diameter on Heat Transfer, Microfluidics and Nanofluidics, (), pp. 9-. [] Chen, T. and Garimella S. V.,, Effect of Dissolved Air on Subcooled Flo Boiling of a Dielectric Coolant in a Microchannel Heat Sink, Journal of Electronic Packaging, (), pp [] Taylor, J. R., 99, An Introduction to Error Analysis, nd Ed., University Science Books. Copyright by ASME Donloaded From: on //3 Terms of Use: