Available online at www.sciencedirect.com Experimental Thermal and Fluid Science 32 (2008) 1090 1095 www.elsevier.com/locate/etfs Micro capillary pumped loop system for a cooling high power device Chin-Tsan Wang a, *, Tzong-Shyng Leu b, Tsai-Ming Lai b a Department of Mechanical and Electro-Mechanical Engineering, National I-Lan University, 1, Sec. 1, Shen-Lung Road, I-Lan 26047, Taiwan, ROC b Department of Aeronautics and Astronautics, National Cheng Kung University, Tainan 701, Taiwan, ROC Received 8 February 2007; received in revised form 6 August 2007; accepted 8 January 2008 Abstract This work discusses the operation of a capillary-driven two-phase loop, configured on a micro capillary pumped loop (MCPL) system without an external power supply but capable of automatic heat transmission. The MCPL device, fabricated using MEMS (microelectricomechanical system) technology, was tested and yielded the following results: first, the proposed design of a new MCPL system with a water reservoir operating at low pressures is feasible and requires no additional power supply and instead relies on automatic heat transmission. Second, the issue of depriming in a MCPL was effectively controlled, the endurance of MCPL for the depriming problem can be executed by yielding input heat fluxes of 185.2 W/cm 2 at an evaporator temperature of 165 C, thus revealing that this model provides excellent cooling performance. Third, the effective operation range was determined and its successful operation was confirmed for MCPL. The ease of starting up increased with the temperature of the reservoir. Finally, two-phase tension that originated in the groove structures in the evaporator and condenser was confirmed to control the movement of the fluids throughout the system and verified to be effective in improving cooling efficiency. Ó 2008 Elsevier Inc. All rights reserved. Keywords: Micro capillary pumped loop; Two-phase flow; Surface tension 1. Introduction As the power supplied to electronic packages and microelectromechanical systems (MEMS) increases, thermal management becomes increasingly problematic. People are using smaller devices that emit large amounts of thermal energy and thus seriously damage electronic performance. Conventional methods of heat removal are simply not capable of dealing with the high thermal gradients that cause material failure. Traditional thermal management schemes must be improved upon and new approaches to cooling micro-devices must be developed. Many potential configurations have been considered in developing micro-coolers. A micro-cooler may exploit a micro-scale phase change or single-phase heat transfer to dissipate heat from electronic packages, which includes * Corresponding author. Tel.: +886 3 9357400x686; fax: +886 3 9311326. E-mail address: ctwang@niu.edu.tw (C.-T. Wang). the use of miniature heat pipes [1,2]. Regardless of configuration, however, an effective micro-cooler will have to be able to fully control the motion of fluids throughout the system. The combination of micro-scale heat transfer and fluid dynamics along with high surface-to-volume ratios makes the development of an efficient micro-cooler challenging. Kirshberg et al. was the first to present the conceptual design and fabrication of a micro-cpl [3]. The initial design involved a completely-passive three-port micro- CPL, and detailed schematics were provided and addressed in their work. Due to their small size, micro-devices must support extremely large gradients of pressure. The evaporator region was to be placed in direct contact with the micro-processor, sensor, or other electronic chip for which cooling is required and used to maintain an optimal temperature. There are certain advantages to this approach, the first being that the temperature is precisely controlled at the chip-level. Second, the overall cooling is more efficient for specific heat sources in the electronic package. 0894-1777/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.expthermflusci.2008.01.001
C.-T. Wang et al. / Experimental Thermal and Fluid Science 32 (2008) 1090 1095 1091 Third, the overall size of the electronic system can be kept small. Another distinct advantage of the CPL is the potential to further enhance the device s performance. Currently, major components of the system include a new family of pumps and valves that control the liquid or vapor transport in the micro-cpl, as well as innovative micro-scale heat acquisition and heat rejection regions. These regions may incorporate micro-mixers and use enhanced surfaces to promote heat transfer, condensation and boiling. In terms of design, capillary pumped loops are similar to heat pipes. The effective thermal conductivity of a CPL is derived from the vaporization and condensation of its working fluid. The main advantages of CPLs are that it can increase thermal isolation between the vapor and the liquid [4,5]. Also, it can significantly improve geometric freedom by separating the evaporator from the condenser. Finally, since the wicking structure has been removed from a majority of devices, pressure drops are reduced along the vapor and liquid lines, allowing for larger mass flow rates under the capillary pumping limit. Research on CPLs began in 1961 when Laub and McGinness began to study a two-phase thermal control system called a capillary pumped loop [6]. They only investigated a capillary pumped vapor generator. Stenger expanded their work in 1966, reporting on two CPLs that could transport more than 800 W over 15.24 m [6,7]. Since then, CPLs have been rigorously investigated and observed to be governed by the same limits as heat pipes. Onedimensional paradigms for designing and predicting their performance have been established [5]. The transition of MEMS technology to this specific application faces many challenges. Some of the challenges relate to the micro-fabrication design, and include material and geometric characteristics as well as overall flow dynamics. Several technical challenges raise extremely interesting problems relating to the physics of micro-heat transfer and fluidic mechanics. A few of these issues that have been identified relate to the physics of micro-scale heat transfer, and such phase change phenomena including interfacial dynamics. The development and optimization of such systems will require better understanding of these physical issues and will be discussed later. Eventually, a novel thermal management system must be designed to control liquid and vapor transport, in addition to having innovative micro-scale heat acquisition and rejection regions. 2. Micro-CPL design The five main components of the MCPL system are the evaporator, the condenser, the liquid line, the vapor line, and the reservoir. They were designed and utilized to examine the start-up of heat transmission because it is a key problem for MCPL systems. The micro-cpl system was fabricated using standard MEMS technology to the dimensions of the parameters described in Table 1. Fig. 1 shows the schematics for the micro-cpl and the parts that were fabricated by MEMS to study the start-up of MCPL. Table 1 Parameters of MCPL system Evaporator dimensions 3.42 mm 1.5 mm Condenser dimensions 7.46 mm 5mm Vapor line dimensions 20 mm 460 lm Liquid line dimensions 30 mm 160 lm Reservoir 3 mm 3mm Vapor/liquid line deep length 150 lm Groove structure dimension of evaporator 4.5 mm 30 lm Groove structure dimension of condenser 7.46 mm 30 lm Depth of groove 30 lm Groove number of evaporator 8 Groove number of condenser 25 Condenser Groove Vapor Line Insulation chamber Liquid Line Evaporator Reservoir Liquid line +Groove Fig. 1. Fabrication of MCPL system. Two pieces of glass, specifically Pyrex 7740 each measuring 76 mm 26 mm with a thickness of 1 mm, were utilized as the substrate material and applied to visualize flow motion in the MCPL device, especially in the groove region of the evaporator. The wicking groove structures, measuring 4.5 mm by 30 lm, were used to generate the driving force for the transmission. It should be noted here that the dimension of the wicking structure should be larger than the dimension of the evaporator because it must extend into the liquid line and the two side channels adjacent to the evaporator connecting it to the reservoir. Although increasing the number of grooves will enhance bubble generation and strengthen the capillary force, the induced flow resistance will also clog the flow motion in the groove channel and result in depriming. Limitations in the wire line width for fabrication of the MCPL device prompted technical issues. Initially eight grooves were used in the evaporator, and another 25 grooves were used in the condenser for increasing the velocity of flow after the condensation of steam into sub-cooled flow. Although there was an inversion of the meniscus in the condenser s groove relative to the meniscus in the wicking structure of the evaporator, capillary pressure in the condenser could not overcome the opposing flow induced by the capillary pumping in the evaporator and was confirmed by flow visualization. The next device incorporated an evaporator measuring 3.42 mm by 1.5 mm (displayed in Fig. 1), and was improved by adding a port specifically for filling characteristics. It is connected to a 3 mm by 3 mm reservoir, whose main purpose is to stabilize temperature and pressure in the evaporator and the loop of MCPL system as well as providing access for thermocouples [8].
1092 C.-T. Wang et al. / Experimental Thermal and Fluid Science 32 (2008) 1090 1095 A welding blowpipe with a working voltage of 110 V, electrical resistance R = 635 X and heating power of 20 W was connected to a voltage regulator capable of a range from 72 V to 82 V and used as the heat source in the geometric center of the evaporator to imitate the electrical heating of the device. The heating area was very small and can be seen as a hot-spot because the welding blowpipe was approximately 1 square millimeter for the heat source. The heating power of welding blowpipe can be obtained by the formula P ¼ V 2 rms, where V R rms is the output voltage of voltage regulator and R is the resistance of the welding blowpipe. The effect of heating power associated with evaporation on the startup performance of an MCPL system was studied by controlling the electrical resistance of the blowpipe. An increase in the cooling area corresponds to an increase in the ability to transfer heat from the evaporator to the environment. Hence, a rectangular area measuring 7.46 mm by 5 mm was used, and the wicking groove structures in the condenser area enhanced the transmission velocity of the sub-liquid and pushed the sub-liquid in the liquid line to the evaporator to compensate for the evaporation of liquid and to maintain the cooling capacity to prevent dry-out. The dimensions of the liquid line and vapor line design are 20 mm by 460 lm and 30 mm by 160 lm, respectively. The depth of the flow channel was etched to 150 lm. MEMS was applied to make a prototype MCPL, which was then used in an experiment. Fig. 2 presents a sketch map of the experimental set-up for the micro capillary pump loop system. The working fluid, pure water applied, must operate without impurities as an important condition to avoid the appearance of non-condensable gases (NCG) in the loop. The presence of NCG in a CPL can cause a failure in the capillary evaporator, but it is less likely to occur there than in heat pipes. The presence of NGC can be minimized using materials compatible with the selected working fluid. Ensuring a good vacuum seal in the loop as well as using fluid with a minimum amount of contaminants also minimizes the presence of NCGs. Fig. 2. Sketch of experimental setup. Before start of the experiment, the non-condensable gas embedded in the device of system and reservoir was eliminated using a vacuum pump to prevent residual NCG from impeding flow and prompting the deprime issue. Due to the non-condensable nature of the gas, the pressure of the system device was drawn out using a vacuum pump to maintain a low pressure of around 190 mm-hg. Finally, the liquid line and condenser area had to be completely filled, but only partially in evaporator by filling pure water from the reservoir at a low pressure state. Note here that fluid must be prevented from entering the vapor line. Furthermore, the reservoir employed to maintain the saturation state needed to be preheated by a hot-plate. A thermocouple temperature sensor, K type, and CCD camera were used to visualize the flow. 3. Results and discussion During the start-up testing process, 90% of the volume was maintained and the reservoir preheating temperature was set to 90 C. The heating power was varied such that the heat flux ranged from 156 W/cm 2 to 185.2 W/cm 2 in order to elucidate the effect of heat flux on start-up performance, which is of significant importance in the design of MCPLs. Hence, a cooling fan was bonded to the surface area of condenser by forced convection for the purpose of enhancing its cooling ability. Note that the heat flux is calculated as the ratio of the electric power to the area of the heater, which in this case is comprised of 1 square millimeter of copper and is embedded in the center of evaporator, not on the surface of the evaporator. The working fluid in evaporator was heated to saturation by heating the small piece of copper and utilizing it as a hot-spot. In addition, the fluid in the condenser was also condensed to a saturated condition. Although minor heat loss was inevitable, an insulation chamber situated in the center of device (shown in Fig. 1) was applied to prevent as much heat loss as possible and eliminate error in temperature measurement. Furthermore, in order to determine the maximum heat flux capacity of the device with maximum accuracy, a vacuum chamber was constructed so that the micro-cpl could be tested without convectional heat loss to ambient surroundings. Fig. 3 demonstrates that the MCPL system started up successfully at a constant heat power of approximately 156 W/cm 2 ; the temperature of the evaporator then approached a steady-state temperature of about 135 C. The condenser simultaneously remained in a quasi-steady state at 40 C. This result indicated that transmitted heat accumulated and the heat transfer equilibrium of the system was reached. When the system was heated by heat flux of 166 W/cm 2, the circulation of the system occurred naturally and was maintained continuously at a steady state without an additional power supply. Through the use of a CCD camera and microscope, the successful circulation of the fluid for system was confirmed by the flow image. The start-up time plotted in Fig. 4 was 3 min earlier than in Fig. 3. When the temperature of the evaporator was
C.-T. Wang et al. / Experimental Thermal and Fluid Science 32 (2008) 1090 1095 1093 Start-up works Start-up works 140 180 120 T reservoir 160 140 120 T reservoir 80 80 60 60 40 40 20 0 5 10 15 20 25 30 35 Time(min) Fig. 3. A heating power of 156 W/cm 2 and 90% of the volume were used for the start-up test. 160 140 120 80 60 40 Start-up works T reservoir 20 0 5 10 15 20 25 Time (min) Fig. 4. A heating power of 166 W/cm 2 and 90% of the volume were used for the start-up test. increased to around 145 C and the condenser was increased to a constant temperature of 40 C, the fluid for the system continued to circulate and maintained a steady state. Additionally, the phenomenon of depriming did not occur in evaporator. Fig. 5 demonstrates that the system can also work at a heat flux of 175.4 W/cm 2. The temperature of the evaporator was changed to T = 165 C and the condenser was also increased to T =50 C. The heat transfer equilibrium of the system was maintained under this condition. The CCD camera was used to visualize the flow and the associated two-phase phenomenon in the groove region. The exchange of the two phases (liquid/vapor) at the gas liquid interface of the groove region was clearly observed, and is depicted by 20 0 5 10 15 20 25 Time (min) Fig. 5. A heating power of 175.4 W/cm 2 and 90% of the volume were used for the start-up test. the flow image in Fig. 6. This phenomenon indicated that the liquid in the evaporator was vaporized and then pushed forward into the vapor line by the driving force of the surface tensor, which was generated by a groove structure. At the same time, the sub-liquid of the liquid line that was condensed and pushed out by the surface tension of the groove structures in the condenser area moved to the evaporator to compensate for the vaporization of the liquid. Accordingly, circulation occurred in the closed loop system and the heat transmission was in equilibrium. The groove structure is important in maintaining the circulation of the fluid for MCPL system and the equilibrium of heat transmission. Knowing the capacity of the MCPL system to maintain heat transfer equilibrium without depriming is important because this issue drastically affects cooling of the MCPL. Thus, the system was heated to a heat flux of 185.2 W/cm 2, and depriming occurred about 7 min later (shown in Fig. 7). The circulation of the fluid for the system and the equilibrium of heat transmission failed simultaneously. According to the results on the effect of heat flux related to the start-up performance of the system, the MCPL system could bear a heat flux of 185.2 W/cm 2 and should be regarded as an effective cooling device which can meet the current need for electronic cooling. In addition, the effect of heat power input on thermal resistance is worth Fig. 6. Experimental video near the heating part in the start-up test.
1094 C.-T. Wang et al. / Experimental Thermal and Fluid Science 32 (2008) 1090 1095 deprime 200 150 T reservoi r 50 0 0 5 10 15 20 Time (min) Fig. 7. A heating power of 185.2 W/cm 2 and 90% of the volume were used for the startup test. investigating because it could act as an index of efficiency for cooling. Three kinds of input heating power valued at 8 W, 9 W and 10 W were executed at 90% of the volume and T r =90 C for the reservoir. The results of Fig. 8 shows that the thermal resistance will increase with additional heat power input. In addition, the slope of thermal resistance will drastically increase when the heat power input is in the range between 8.5 W and 9 W. This thermal phenomenon is suggested to be the result of the inlet mass flow rate of the cooling liquid from the liquid line to the evaporator not being able to match the rate of the increased evaporation rate due the additional heat power input, thus stressing the system to its limits and inducing the appearance of depriming. Knowing the thermal fluid dynamic behavior of evaporation/boiling in the grooves of the evaporator is important Thermal Resistance (degrees/w) 13.2 13 12.8 12.6 12.4 12.2 12 11.8 8W 8.5W 9W 11.6 7.8 8 8.2 8.4 8.6 8.8 9 9.2 Power (W) Fig. 8. Input heating power versus thermal resistance at 90% of the volume and T r =90 C for the reservoir. Fig. 9. Appreciate operation range for start up of MCPL. for future development of miniature thermal control systems. From the visualizations shown in Fig. 6, the steady behavior was never observed but unsteady boiling appeared to occur. Researchers at Stanford University have also found unsteady boiling regimes at the micro-scale level [9]. However, despite the unsteady nature of the evaporation process, the resulting temperature of the evaporator remained remarkably steady [10]. Size effects may be responsible to a large degree for the unsteady boiling observed in experiments. However, there seem to be indications that surface tension effects may play an important role in the evaporation dynamics. Although transients exist in the evaporative process, they are much slower and less chaotic and violent. This indicates that one possible factor in the unsteadiness could be thermal generation of surface tension gradients. Temperature gradients will exist in the evaporator as a result of wick geometry and unsteady boiling, and may generate unsteady forces that create unsteady flows that might pass over or directly interact with the nucleate boiling processes. However, the exact thermal dynamic behavior in the grooves of the evaporator and condenser is very complex and needs further study. Finally, further study was done in determining an appropriate range for the system to operate effectively. The area marked in color 1 in Fig. 9 indicates the effective operating range of the MCPL system. Fig. 9 also reveals the ease of start-up increasing with the temperature of the reservoir. Care must be taken to control the temperature of the reservoir. 4. Conclusions A micro capillary pumped loop system to maintain the equilibrium of heat transmission was designed, fabricated and tested in a series of experiments. The following important findings were presented and addressed. 1 For interpretation of color in Fig. 9, the reader is referred to the web version of this article.
C.-T. Wang et al. / Experimental Thermal and Fluid Science 32 (2008) 1090 1095 1095 First, the MCPL system must be at low pressure and maintain automatic circulation for preheating to ensure the equilibrium of heat transmission without an additional power supply. Second, the system starts up successfully and depriming does not occur until the heat flux reaches 185.2 W/cm 2. Thus, it can be regarded as an effective cooling device which meets current needs for electrical cooling. Finally, groove design is important in an MCPL system and further effort must be made to improve the cooling performance of the system. Acknowledgement The authors would like to thank the National Science Council of the Republic of China, Taiwan, for financially supporting this research under Contract No. NSC-95-2212-E-197-020. References [1] G.P. Peterson, A.B. Duncan, M.H. Weichold, Experimental investigation of micro heat pipes fabricated in silicon wafers, Journal of Heat Transfer 115 (1993) 751 756. [2] A. Hoelke, H.T. Henderson, F.M. Gerner, M. Kazmierczak, Analysis of the heat transfer capacity of a micromachined loop heat pipe, Proceedings of the ASME Heat Transfer Division, HTD 364-3 (1999) 53 60. [3] J. Kirshberg, K. Yerkes, D. Liepmann, Micro-cooler for chip-level temperature control, in: SAE Aerospace Power Systems Conference, 1999, p. 341. [4] A. Faghri, Heat Pipe Science and Technology, Taylor & Francis, Washington, DC, 1995. [5] J.T. Dickey, G.P. Peterson, Experimental and analytical investigation of a capillary pumped loop, Journal of Thermophysics and Heat Transfer 8 (3) (1994) 602 607. [6] J.H. Laub, H.D. McGinness, Recirculation of a Two-phase Fluid by Thermal and Capillary Pumping, California Institute of Technology, Jet Propulsion Lab., TR 32-196, Pasadena, CA, December 1961. [7] F.J. Stenger, Experimental Feasibility Study of Water-Filled Capillary-Pumped Heat-Transfer Loops, NASA TM-X-1310, NASA Lewis Research Center, Cleveland, OH, 1966. [8] K. Pettigrew, J. Kirshberg, K. Yerkes, D. Trebotich, D. Liepmann, Performance of a MEMS based micro capillary pumped loop for chip-level temperature control, in: 14th IEEE International Conference on Micro Electro Mechanical Systems, Interlaken, Switzerland, 2001, pp. 427 430. [9] L. Zhang, J. Koo, L. Jiang, K.E. Goodson, J.G. Santiago, T.W. Kenny, Study of Boiling Regimes and Transient Signal Measurements in Microchannels, Transducers 01, Munich, Germany, June 10 14, 2001, pp. 1514 1517. [10] J. Kirshberg, K.L. Yerkes, D. Liepmann, Demonstration of a micro- CPL based ones MEMS fabrication technologies, in: 35th Intersociety Energy Conversion Engineering Conference, vol. 2, Las Vegas, NV, USA, 2000, pp. 1198 1204.