Development of Functional Porous Heat Sink for Cooling High-Power Electronic Devices

Transcription

1 Yuki and Suzuki: Porous Heat Sink for High-Power Electronic Devices (1/6) [Technical Paper] Development of Functional Porous Heat Sink for Cooling High-Power Electronic Devices Kazuhisa Yuki and Koichi Suzuki Tokyo University of Science, Yamaguchi, Daigaku-dori 1-1-1, Sanyo-Onoda, Yamaguchi , Japan (Received August 18, 2012; accepted October 29, 2012) Abstract A sub-channels-inserted porous evaporator is proposed as a heat sink for future power electronic devices with a heat load exceeding 300 W/cm 2. The porous medium is made by sintering copper particles of micrometer size in diameter and has several sub-channels to enhance discharge of generated vapor outside the porous medium. This porous heat sink is attached to the backside of a heating chip and removes the heat by evaporating a cooling liquid passing through the porous medium against the heat flow. In order to prove the validity of the sub-channels, the heat transfer characteristics of this porous heat sink are evaluated experimentally. The result shows that the heat transfer performance of a sinteredcopper particles porous medium with sub-channels enables the removal of much higher heat flux under a lower flow rate of cooling water and a lower wall superheat conditions than those of a normal porous heat sink. The removal heat flux, 810 W/cm 2, is 1.8 times higher than that of a normal porous heat sink at a wall superheat of 50 K. Furthermore, it is clarified that even with a heat flux up to 810 W/cm 2, it is possible to sufficiently cool the SiC-based chip in practical use. Keywords: Cooling, High Power Electronic Devices, Porous Heat Sink, Sub-Channels, Phase-Change, High Heat Flux 1. Introduction With the rapid development of electric vehicles, the heat-power density generated in an inverter, i.e. the heat flux, has reached hundreds of W/cm 2. For instance, the heat flux for recent hybrid cars exceeds 100 W/cm 2 and this value will conceivably reach or exceed 300 W/cm 2 for future electric cars. Taking safety into account, it s a critical issue to develop a high heat flux heat sink capable of near 500 W/cm 2 of cooling performance, with a low-power pumping system in order to reduce power consumption. However, there is a possiblity that the heat spreader usually installed onto a heating chip to ease the heat flux isn t available for cooling this kind of high-power electronic device. This is due to the large thermal resistance, i.e., the large temperature difference generated within the spreader, which occurs under high heat flux conditions. Figure 1 shows the heat flux decline and the temperature difference in a copper heat spreader with a spread angle of 45 and a thickness of 5.0 mm set on to a cm 2 heating chip with a heat generation of 500 W/cm 2. In this estimation, the heat generated at the chip is assumed to Fig. 1 Heat flux decline and temperature difference in a copper heat spreader. 69

2 Transactions of The Japan Institute of Electronics Packaging Vol. 5, No. 1, dimensionally spread through the heat spreader. The heat flux in the spreader is eased to 100 W/cm 2 at a distance, ΔL, of 5.0 mm from the chip surface, but the temperature difference exceeds 30 K. This thermal resistance becomes much larger in consideration of the thermal contact resistance generated at both the sides of the spreader. Furthermore, the heat generated in the chip doesn t spread sufficiently within the spreader under the high heat transfer rate conditions that are inevitably required for cooling high heat flux devices. These facts suggest that direct cooling without a heat spreader might be essential under high heat flux conditions. In that sense, boiling heat transfer that utilizes the latent heat of vaporization is one realistic solution to remove a heat flux on the order of hundreds of W/cm 2. In order to succeed at high heat flux removal with low pumping power, the authors proposed a heat sink that utilizes the phase change of a cooling liquid in a metal porous medium attached to the back of high heat flux equipment firstly such as divertor in a fusion reactor.[1 6] This porous heat sink is a counter type of cooling device that removes heat by evaporating the cooling liquid, which is pumped against the heat flow. In other porous heat transfer devices that actively utilize the latent heat of vaporization, such as heat pipes[7] and vapor chambers,[8] the bottle-neck of vaporliquid exchange is a trade-off relationship between the capillarity and the permeability of the liquid, which makes it extremely difficult to achieve heat flux removal exceeding 100 W/cm 2. Of course, there are many papers reporting on porous heat sinks under flow conditions;[e.g. 9 12] the critical issue is to ease their large pressure loss. To cope with these difficulties, our porous heat sink enables mechanical pumping of a moderate amount of cooling liquid corresponding to the heat input towards the two-phase region and the evaporation of most of the liquid by utilizing the vast heat transfer surface of the porous medium. This system could lead to low flow rate operation and contribute significantly to economic driving of the electric vehicle. Judging from heat transfer experiments and simulations using water as the cooling liquid under atmospheric pressure conditions performed in the past,[5] a heat flux of up to approximately 300 W/cm 2 could be removed at a low wall superheat of 70 K under low flow rate conditions. However, it has also been confirmed that active discharge of the vapor generated in the porous medium could be essential for heat removal over 300 W/cm 2. In order to achieve heat flux removal exceeding 300 W/cm 2 under a high heat transfer rate, functional porous media that enable the active discharge of vapor outside the porous medium must be developed, keeping the large heat transfer surface. The purpose of this study is to evaluate the heat transfer performance and heat transfer characteristics of a new porous heat sink with included sub-channels and to clarify the applicability of this porous heat sink for cooling electronic power devices under heat flux conditions of over 300 W/cm Conceptual Design of Sub-Channels-Inserted (SCI) Porous Heat Sink The greatest concern in utilizing porous media as a heat sink is that a completely dry region is formed in the porous medium. This dry region increases the thermal resistance and reduces the effective thermal conductivity of the porous medium; as a result, it raises the surface temperature of the chip. One solution to this difficulty is the installation of sub-channels in the porous medium toward the outlet as shown in Fig. 2. As the sub-channels are arranged radially along the heat transfer surface of the cylindrical shape of the porous medium, the vapor formed near the center is automatically allowed to discharge outside the porous medium at the moment when the developed vapor-phase region reaches the inlets of these channels. It should be noted that these sub-channels can be installed not only in the above-mentioned direction but also in the axial and other directions. Furthermore, it is desirable to install a barrier wall between the vapor-discharging sub-channels and the porous medium in order to prevent the liquid from directly flowing into the sub-channels. In terms of the sub-channels-inserted (SCI) porous concept, we propose utilizing a pipe as the sub-channel. It might be possible to control vaporization completely, unlike in the usual flow boiling heat transfer. Fig. 2 Sub-channels-inserted porous medium. 70

3 Yuki and Suzuki: Porous Heat Sink for High-Power Electronic Devices (3/6) 3. Heat Flux Removal Experiment Using Sub-Channels- Inserted Porous Media 3.1 Experimental apparatus and procedure Figure 3 shows the test section of the experimental apparatus. The test section consists of a copper heat transfer block which is heated by a plasma arcjet, and a stainless-steal circular pipe with a porous medium inserted into it. Each of the parts is connected using a flange. Of course, we can utilize a high power cartridge heater as a heat source, but it s not possible to realize a heat flux exceeding 500 W/cm 2 for the large heat transfer surface used in this study. The copper heat transfer block is cooled by mechanically supplying water into the porous medium attached to the back of the block with a magnetic pump. The copper heat transfer block is cylindrical, 11.5 mm in thickness and 120 mm in diameter. On the central axis of the block, four K-type sheathed thermocouples, each 1.0 mm in diameter, are installed. The locations of the thermocouples are 1.5 mm, 4.0 mm, 6.5 mm, and 8.5 mm from the interface between the porous medium and the copper heat transfer block. The fluid passing through the porous medium can flow out only from the sub-channels as shown in Fig. 2; it is then discharged through six circular pipes, each with an inner diameter of 8 mm, attached to the connection flange. In this experiment, the flow rate of cooling water is adjusted first; afterwards, the heat transfer block is heated by the plasma arcjet. Heat flux from the plasma strongly depends on the distance between the plasma nozzle and the target. The distances, H, are H = 8.0, 5.0, and 4.0 cm. For each distance, after the temperatures of the copper heat transfer block reach a steady state, 1 min. of temperature data for each thermocouple is obtained at intervals of 1.0 s. From the evaluation of the temperature profile on the central axis of the block, the heat flux, q, is estimated using Fourier s law of heat conduction. The heat transfer coefficient is also evaluated by q/δt sat (ΔT sat : wall superheat). The degree of subcooling of the cooling water is approximately 80 K 85 K. Fig. 4 Photograph of porous medium. 3.2 Porous medium loaded sub-channels Figure 4 shows the sub-channels-inserted porous medium that we fabricated for the present study. In the fabricating process, first, copper pipes are attached onto the bottom surface of the graphite furnace mold, and then copper particles are filled around these pipes and sintered by heating and pressurizing. The average diameter of the copper particles used is 500 μm (the pore size is below 100 μm), and the porosity is approximately 30%. The porous medium is cylindrical, 50 mm in diameter and 20 mm in height. The installed copper pipe sub-channels are oriented in a radial direction along the heat transfer surface. Each sub-channel is 3.0 mm in diameter and 20 mm long, and a total of four sub-channels are installed at 90 angles in the circumferential direction. 4. Experiment Results 4.1 Heat transfer performance and characteristics of SCI porous heat sink Figures 5 and 6 indicate the removal heat flux and the wall temperatures, respectively. The horizontal axis indicates the mass flow flux of cooling water (kg s 1 m 2 ). The inlet pressure is approximately from 10 to 90 kpa. We adjusted the heat input by changing the distance from the plasma nozzle to the heat transfer surface. As shown in Fig. 5, there is a heat removal regime where the heat flux doesn t depend on the flow rate, especially for H = 5.0 and Fig. 3 Experimental apparatus. Fig. 5 Removal heat flux. 71

4 Transactions of The Japan Institute of Electronics Packaging Vol. 5, No. 1, 2012 H = 4.0 cm, which might correspond to a fully developed boiling regime on the usual boiling curve. In this regime, stable cooling in which no perturbations affect the cooling of the chip is possible. Focusing on the middle heat input case (H = 5.0 cm), the heat flux reaches 810 W/cm 2, which is 1.8 times higher than that of the normal porous heat sink which has the same particle diameter,[5] at a wall superheat of approximately 50 K (a wall temperature of 150 C). As shown in Fig. 7 the heat transfer coefficient reaches Wm 2 K 1 at the same wall superheat, which is 2.4 times higher. These results prove that the heat transfer performance of the newly introduced porous heat sink is dramatically enhanced by inserting the subchannels for discharging vapor. In the high heat input case (H = 4.0 cm), the heat flux almost reaches almost 1000 W/ cm 2 (the maximum heat flux is 960 W/cm 2 ), though the wall superheat exceeds 100 K (a wall temperature of 225 C). This suggests that the present heat sink isn t available for 1000 W/cm 2 class of a high power electronic device because the temperature of the chip excessively exceeds 200 C. Figure 7 also shows that the heat transfer coefficient increases with a decreasing flow rate, which proves that utilization of latent heat becomes more dominant as the flow rate decreases. Of course, we have to be careful of the degradation of the heat transfer coefficient in cases where the flow rate is quite low. On the other hand, the heat transfer coefficient decreases with increasing heat flux. This result doesn t show the degradation of the heat transfer of this heat sink because the data still correspond to that in a fully developed boiling regime where the heat flux doesn t depend on the flow rate. In that sense, this data indicates the increase in the temperature jump due to thermal contact resistance at the interface between the heat transfer block and the porous medium. This temperature jump also depends on the heat flux and could be significant with increasing heat flux. 4.2 Applicability of SCI porous heat sink to high power electronics devices Focusing on the approximately 40 kg s 1 m 2 mass flow flux of the cooling water shortly before the degradation in the heat transfer coefficient, we can obtain a boiling curvelike result for the present SCI porous heat sink. Figures 8 and 9 show the heat flux and the heat transfer coefficient, respectively, for wall superheat. As mentioned above, although the heat transfer coefficient decreases with increasing heat flux, it is possible to cope with this by sintering the porous medium and the heat transfer block together. By utilizing data obtained in this study, we can predict the chip surface temperature without the effect of thermal resistance. Figure 10 shows the chip surface temperature for the heat flux, assuming that the thickness of the copper base of the heat sink is 1.0 mm (see Fig. 11). Fig. 6 Wall temperature. Fig. 8 Boiling curve for porous cooling. Fig. 7 Heat transfer coefficient. Fig. 9 Heat transfer coefficient. 72

5 Yuki and Suzuki: Porous Heat Sink for High-Power Electronic Devices (5/6) Fig. 10 Temperatures of chip. wall temperature was approximately 220 C). (4) With a heat flux from 300 W/cm 2 to 810 W/cm 2 the chip can be sufficiently cooled if the SiC-based chip can operate at up to 200 C. The optimized design of the sub-channels conceivably depends mainly on the heat input, which suggests that the diameter, length, number, and shape of the sub-channels need to be adjusted along with the increase of heat flux. As our next step, we plan to evaluate these impacts on the heat transfer performance and characteristics for the flow rate of cooling water, degree of subcooling, and heat flux. Fig. 11 Prediction of chip surface temperature. Judging from these data, a heat flux from 300 W/cm 2 to 810 W/cm 2 is sufficient to cool down the chip if the SiCbased chip can operate up to 200 C. Although the detailed boiling curve-like data shown in Fig. 8 should be obtained and the thermal contact resistance must be evaluated under high heat flux conditions in order to predict the chip surface temperature more precisely, the present study sufficiently proves the applicability of the SCI porous heat sink to high-power electronic devices. 5. Conclusion In this study, the heat transfer performance and characteristics of a sub-channels-inserted porous heat sink were evaluated. The findings are summarized as follows. (1) The heat transfer performance of a sintered-copperparticles porous medium with sub-channels enables the removal of a much higher heat flux under a lower flow rate of cooling water and a lower wall superheat conditions, compared with normal porous media. (2) A heat flux of 810 W/cm 2 was achieved at a wall superheat of 50 K and a heat transfer coefficient of Wm 2 K 1. (3) The heat flux could reach almost 1000 W/cm 2 although the wall superheat exceeded 100 K (The References [1] K. Yuki, J. Abei, H. Hashizume, and S. Toda, Super- High Heat Flux Removal Using Sintered Metal Porous Media, Journal of Thermal Science, Vol. 14, No. 3, pp , [2] H. Togashi, K. Yuki, and H. Hashizume, Heat transfer enhancement technique with copper fiber porous media, Fusion science and technology, Vol. 47, No. 3, pp , [3] K. Yuki, J. Abei, H. Hashizume, and S. Toda, Numerical Investigation of Thermofluid Flow Characteristics with Phase Change against High Heat Flux in Porous Media, ASME Journal of Heat Transfer, Vol. 130, Issue 1, , [4] A. Matsui, K. Yuki, and H. Hashizume, Dependence of heat transfer coefficient on porous structure in porous media, Proceedings of 2008 ASME Summer Heat Transfer Conference, Paper-no , [5] K. Yuki, H. Hashizume, S. Toda, and K. Suzuki, Key issues to enable heat flux removal exceeding 10 MW/m 2 by use of metal porous media as latent-heat transfer device, Special Topics & Reviews in Porous Media An International Journal, Vol. 1, No. 1, pp. 1 13, [6] K. Yuki, K. Suzuki, J. Abei, H. Hashizume, and S. Toda, Numerical investigation on heat transfer characteristics in metal-particle-based porous heat sink by a two-energy model, Proceedings of the 3rd International Conference on Porous Media and its Applications in Science and Engineering, [7] D. Reay, R. McGlen, and P. Kew, Heat Pipes, Sixth Edition: Theory, Design and Applications, Butterworth-Heinemann, [8] K. M. Kota, Design and Experimental Study of an Integrated Vapor Chamber-Thermal Energy Storage System, Proquest Umi Dissertation,

6 Transactions of The Japan Institute of Electronics Packaging Vol. 5, No. 1, 2012 [9] C. Li and G. P. Peterson, Evaporation/Boiling in Thin Capillary Wicks (I) Wick Thickness Effect, ASME Journal of Heat Transfer, Vol. 128, pp , [10] C. Li and G. P. Peterson, Evaporation/Boiling in Thin Capillary Wicks (II) Effects of Volumetric Porosity and Mesh Size, ASME Journal of Heat Transfer, Vol. 128, pp , [11] Z. Q. Chen, P. Cheng, and T. S. Zhao, An Experimental Study Of Two Phase Flow And Boiling Heat Transfer In Bi-Dispersed Porous Media, Int. Comm. Heat Mass Transfer, Vol. 27, No. 3, pp , [12] G. Hetsroni, M. Gurevich, and R. Rozenblit, Sintered porous medium heat sink for cooling of highpower mini-devices, International Journal of Heat and Fluid Flow, Vol. 27, Issue 2, pp , Kazuhisa Yuki Research on cooling technology of high heat flux equipment such as electronic and energy devices, utilizing metal porous media, micro & minichannel, nano technology etc. ( teacher /t-yuuki.html) Koichi Suzuki Research on boiling heat transfer and advanced high heat flux cooling system for power electronics. A leading researcher on Microbubble Emission Boiling (MEB) in the world. ( teacher/t-suzuki.html) 74