ISTP-16, 2005, PRAGUE 16 TH INTERNATIONAL SYMPOSIUM ON TRANSPORT PHENOMENA IMPROVEMENT ON MOUNTING THERMAL RESISTANCE BETWEEN A CIRCUIT BOARD WITH MANY COMPONENTS AND A LIQUID-COOLED COLD PLATE Masao Fujii *,Yuu Seshimo ** * Kinki Univ., ** Mitsubishi Electric Corp. fujii@waka.kindai.ac.jp, +81-736-77-0345,+81-736-77-4754 Keywords: Electronic equipment cooling, Mounting thermal resistance, Interstitial material, Cold plate, Flexible thermal contactor Abstract This paper describes the control of mounting thermal resistance using interstitial materials, which exists between the circuit board with many components and the liquid-cooled cold plate. Grease, rubber and newly developed flexible thermal contactor are used to improve the mounting thermal resistance as the interstitial material. There are two control s in the mounting thermal resistance: and Interstitial material control. The flexible thermal contactor (FTC) has the feature of reducing the temperature discrepancy of s set on the circuit board. 1. Introduction The increasing power density of electronic systems is reaching the point at which it is no longer possible to adequately cool the individual components by direct air cooling. There is also often insufficient space to mount a large enough air-cooled heat sink directly on the components. A liquid-cooled cold plate has been used to cool the components, which has a thin plate with multiple straight fins covered by a housing allowing water to be circulated between fins [1]. When many components with regular size such as devices are set on a circuit board and cooled with the liquid cold plate, many visible air gaps exist between the cold plate and the individual components to make high temperature drops at the air gaps as shown in Fig.1. In a module, it is also necessary to Fig.1 Air gaps between s and a cold plate. allow for variations in chip heights and locations resulting from the manufacturing process so that the concept of a spring-loaded mechanical piston touching each chip has been provided [2]. All of the components set on the circuit board are needed to keep under allowable temperature and minimize the temperature distribution of them in order to achieve a smooth operation of electronic systems. This paper describes the control of mounting thermal resistance with interstitial materials, which exists between the circuit board with many components and the liquid-cooled cold plate to make high temperature drops. Mounting thermal resistance is governed by three factors: thermal contact resistance between the component and the interstitial material, thermal contact resistance between the cold plate and the interstitial material and thermal conduction resistance of the interstitial material. In this study, Grease, rubber and newly developed flexible thermal contactor are used to improve the mounting thermal resistance as the interstitial material. 2. Mounting thermal resistance Mounting thermal resistance, R m, is defined as the ratio of the additional temperature drop (Δ T) due to the presence of the imperfect joint to the total heat flow, Q. 1
R m =ΔT /Q (1) The mounting thermal resistance can be separated into the thermal conduction resistance of the interstitial medium, R k, and the thermal contact resistance. The thermal contact resistance exists at two places; between the electrical component and the interstitial material, R 1, and between the interstitial material and the cold plate, R 2, as shown in Fig.2, and hence the mounting thermal resistance is shown below: R m =R 1 +R k +R 2 (2) Fig.2 Mounting thermal resistance. 3. Experimental apparatus and procedure The experimental apparatus used in this study consists of a liquid-cooled cold plate with a chilled water system and 16 components set on a circuit board as shown in Fig.3. 10 Cold plate Air gap Thermal contact resistance R 2 Interstitial material Air gap Circuit board Cooling water inlet Insulation T h Insulation Ts (surface temperature of cold plate) Thermal conduction resistance R k Fig.3 Experimental apparatus. Thermal contact resistance R 1 Tw( surface temperature of ) Weight Thermocouple locations Tc Ts Cooling water outlet Mounting thermal resistance R m Cold plate Interstitial material Rubber heater Circuit board (Unit:mm) The circuit board is made of electrical insulated substrate with 250 x 250 mm 2 surface area and 1.7mm thickness, which has 16 rubber heaters. Each rubber heater has the maximum heat transfer rate of 20W. Each component is made of aluminum of dimensions 30mm long, 30mm wide and 10mm high, which is like device. The component is thereby called in this study. Each is set on the each rubber heater by filling the interstitial region with silicone grease to reduce the mounting thermal resistance, and heated at the same electric power. Figure 4 shows the layout of the 16 s set on the circuit board. The heat transfer rate (Q p ) of the individual is supplied with the rubber heater. The temperature of each (T h ) is measured with a copper-constantan thermocouple, 0.2mm outer diameter, which is located at the center of the with a depth of 7.6mm from the upper surface of the. The temperature of the upper surface of the (T w ) is determined by calculating the temperature drop between the upper surface of the and the measured point of the. The temperature T h is varied from 35 10 No.14 No.13 No.12 No.11 No.24 No.23 No.22 No.21 250 Fig.4 Layout of s. 20 30 35 The cold plate is made of aluminum, which is a thin plate with multiple straight fins covered by a housing allowing water to be circulated between fins. The thickness of the cold plate is 9mm and the weight is 1.9kg with water. The surface temperature (T s ) of the cold plate is uniformly kept constant by controlling a chilled water flow rate. The cold plate is set on the 16 s arranged on the circuit board with interstitial material or without it. Grease, rubber and newly developed flexible thermal contactor are examined to reduce the mounting thermal resistance as interstitial materials between the cold plate and the s. Physical properties of commercial interstitial materials used in this study are shown in Table.1. The newly developed flexible thermal contactor will be described after. No.34 No.33 No.32 No.31 No.44 No.43 No.42 No.41 Table 1. Physical properties of commercial 35 30 20 30 35 Unit:mm 250 2
interstitial materials. Thermal conductivity [W/(m K ] Thickness[mm] Thermal conductance[w/(m 2 K)] Expansion[ ] Adhesion Rubber A P P P 10-3 P U S Soft Large Rubber B Q O T 10-3 T Rather Hard Small Grease P Large Contact pressure range is 0.30 kpa to 2.65 kpa by changing the weight (1.9 kg to 16.9 kg) set on the cold plate. This contact pressure range is rather small compared to that presented by other researchers [3]. In many electronic systems, the joint between the liquid-cooled cold plate and the circuit board is not permanent, because of servicing or other consideration, and the circuit board and electronic devices are delicate. The electronic systems can be treated at relatively low contact pressures accordingly. 4. Experimental Results and Discussion The mounting thermal resistance of the individual, R mp, is defined as the ratio of the heat transfer rate (Q p ) to the additional temperature drop (T w -T s ) due to the presence of the imperfect joint and the interstitial material. R mp = (T w - T s )/Q p (3) 4.1 Effect of contact pressure on mounting thermal resistance without interstitial material Figure 5 shows the effect of the contact pressure on the mounting thermal resistance of the individual without interstitial material (in air). As the contact pressure increases, the mounting thermal resistance decreases. The mounting thermal resistance of the individual is not equal. The dispersion of the mounting thermal resistance is due to the various air gaps existing between the individual and the cold Mounting thermal resistance @R mp [K/W] 1 1 8.00 0.30kPa 0.61kPa 1.08kPa 1.55kPa No.11 No.12 No.13 No.14 No.21 No.22 No.23 No.24 No.31 No.32 No.33 No.34 No.41 No.42 No.43 No.44 Number of Fig. 5 Effect of contact pressure on mounting thermal resistance without interstitial material (air). plate. The mounting thermal resistance of No.11 with the smallest mounting thermal resistance is not very influenced by the contact pressure, while that of No.22 et al. with large mounting thermal resistance are much influenced by the contact pressure. s with large mounting thermal resistance have many large air gaps so that the air gaps decrease with an increase in contact pressure. 4.2 Effect of interstitial material on mounting thermal resistance Figure 6 shows the effect of the interstitial material on the mounting thermal resistance at a contact pressure of 1.55 kpa. The mounting thermal resistance of the bare contact in air is decreased by the insertion of the interstitial material. Although rubber B has large thermal conductivity and small thickness compared to those of rubber A, it is found that the insertion of rubber B actually increases the mounting thermal resistance as shown in of No. 34, No. 42, No. 43 and No. 44, whereas the other interstitial materials decrease the mounting thermal resistance. Rubber A is softer and more adhesive than Rubber B. It is therefore confirmed that the rubber softness and adhesion is more effective in decreasing the mounting thermal resistance than the rubber thermal conductance based on its conductivity and thickness. The grease has the most expectable effect to reduce the mounting thermal resistance. It is observed that the grease drains from sides sandwiched between and the cold plate as the contact pressure increases. Grease can fill the visible air gaps so that grease-filled joints would be less sensitive to various kind of air. Mounting thermal resistancermp mk/w n 9.00 gaps. 8.00 7.00 5.00 3.00 1.00 Air Rubber B Rubber A Grease 11 No.11 12 No.12 13 No.13 No.14 14 21 No.21 22 No.22 23 No.23 24 No.24 31 No.31 32 No.32 33 No.33 34 No.34 41 No.41 42 No.42 No. Number of Fig. 6 Effect of interstitial material on the mounting thermal resistance at a contact pressure of 1.55 kpa. 3
However, it may be noted that the thermal resistance of the grease may deteriorate with time due to the loss of the volatile constituents in it [3]. 4.3 Effect of changes of contact pressure on mean mounting thermal resistance Figure 7 shows the effect of the contact pressure on the mean mounting thermal resistance (R ma ) of 16 s. The contact pressure is continuously changed from left side to right side in Fig.7. The mounting thermal resistances with rubber B and without interstitial material (in air) react significantly to the changes of the contact pressure. These results show that the air gaps reduce as contact pressure increases, while the air gaps increase as contact pressure decreases. This of the mounting thermal resistance is called. Rubber B has the same air gap length as the case without interstitial material, as mentioned in section 4.4. Mean mounting thermal resistance Rma[K/W] 8.00 7.00 5.00 3.00 1.00 Air RubberB RubberA Grease 0.30 1.9 0.61 3.9 6.9 1.08 9.9 1.55 6.9 1.08 4.9 0.77 3.9 0.61 0.45 2.9 0.30 1.9 6.9 1.08 9.9 1.55 1.9 0.30 Contact presssure[kpa] Fig. 7 Effect of contact pressure on mean mounting thermal resistance. In the case of grease, the mounting thermal resistance decreases as the contact pressure increases from 0.30kPa to 1.55kPa, and then it is almost kept constant in spite of contact pressure changes. This means that the grease can fill the air gaps and hence two thermal contact resistances, shown in Fig. 2, disappear. This of the mounting thermal resistance is called Interstitial material control. In the case of rubber A, the mounting thermal resistance decreases as the contact pressure increases from 0.30 kpa to 1.55 kpa. After that the mounting thermal resistance reacts a little to the change of the contact pressure, but it is kept smaller value than original one. Rubber A is soft and adhesive to keep good contact between and the cold plate, but still has some air gaps, as shown in section 4.4. 4.4 Analysis of gap length Figure 8 illustrates the concept of two s; and Interstitial material control. The has air gaps and the Interstitial material control has no air gaps. Air 1 Rubber B (rather hard) 1 Rubber A (soft) Fig. 8 Concept of two mounting thermal resistance s. Grease 3 Interstitial material control The air gap length (δ) in the is given by δ=ka(r mp -R k ) (4) where k = thermal conductivity of air [W/(mK)], A = nominal area, 30 x 30 x 10-6 [m 2 ], R mp = mounting thermal resistance of the individual [K/W], R k = thermal resistance based on thermal conductivity of interstitial material [K/W]. Table 2 gives estimated results at a contact pressure of 1.55kPa. The estimated air gap lengths of the rubber B and air are almost the same. The thermal resistance R k of the rubber B is 0.216 K/W and very small, compared to the total mounting thermal resistance. The mounting thermal resistance of the Rubber B is governed by the contact thermal resistances based on air gaps. It is difficult for the rubber B to reduce the mounting thermal resistance of a circuit board with many components. Assuming that the rubber A has no air gaps because of its softness and adhesion, the gap length can be estimated 2.95 mm at the minimum mounting thermal resistance of 1.82 K/W and 8.65 mm at the maximum mounting thermal 2 4
30 Improvement on mounting thermal resistance between a circuit board with resistance of 5.34 K/W. The rubber A has a free thickness of 1.0 mm and can not stretch double and over. The rubber A has some air gaps between the cold plate and the accordingly. The grease will fill the air gaps at relatively low contact pressures so that it can be predicted that the gap length will be equal to that of the air (without interstitial material). The estimated gap length of the grease is however larger than that of the air. It can be considered that the amount of filler like grease influences the mounting thermal resistance at relatively low contact pressures. Table 2. Estimation of gap length. Interstitial material Air Rubber B ( rather hard) Rubber A (soft) Grease Interstitial material control Thermal conductivity k [W/(mK)]/ Thickness[mm]/ Thermal resistancer k[k/w] Mounting thermal resistancer mp(1.55kpa) (Experiment) Gap length [ m] (Estimation) Number of 0.0276/-/0 Minimum 2.35 58 No.11 Maximum 8.18 203 No.32 Average 4.62 115 2.32/0.45/0.216 Minimum 2.85 65 No.11 Maximum 7.54 182 No.43 Average 4.57 108 1.8/1.0/0.617 Minimum 1.82 30 No.11 Maximum 5.34 117 No.32 Average 3.12 62 1.0/-/0 Minimum 0.52 468 No.11 Maximum 1.39 1251 No.23 Average 0.88 792 5. Flexible thermal contactor The flexible thermal contactor (FTC) consists of a copper plate as a thermal conductive material and a stainless steel as an elastic material as shown in Fig.9. The stainless steel set on the inner side of the FTC pushes up the copper plate to increase the contact pressure. The dissipation heat from the is mainly transported through the copper plate. The thickness of the materials is 0.1 mm respectively. The grease is applied on the contact surfaces of the FTC to the cold plate and the to reduce the mounting thermal resistance, but the amount of the grease is very small compared to that mentioned in section 4. It is easy to set the FTCs between the individual components and the liquid-cooled cold plate. Figure 10 shows the thermal path of the FTC with 4 thermal paths. Elastic material 2 Fig.9 Flexible thermal contactor(ftc). 15 Fig.10 Thermal path of FTC. 1 Cold Plate Unit:mm Thermal conductive material 5.1 Mounting thermal resistance The experimental results of the mounting thermal resistance of the FTC are compared to those of other interstitial materials at the contact pressure of 2.65kPa in Fig.11. Mounting thermal resistance[k/w] 5.00 3.00 1.00 Flexible thermal contactor RubberB RubberA Air FTC Grease No.11 No.12 No.13 No.14 No.21 No.22 No.23 No.24 No.31 Number of No.32 No.33 No.34 No.41 No.42 No.43 No.44 Fig.11 Mounting thermal resistances of interstitial materials. The mounting thermal resistances of individual with FTC are nearly equal. The FTC has the feature of reducing the temperature distribution of s set on the circuit board. 5.2 Thermal analysis of FTC An analysis of the mounting thermal resistance of the FTC is carried out by using a thermal network l as shown in Fig.12. The contact resistances between the FTC and the cold plate, and between the FTC and the are assumed to 5
be equal to the minimum mounting thermal resistance (No.11) obtained from the grease in Fig.6. Two parts corresponding to the node of 8 and 15 are assumed not to contact to the cold plate and the. Cold plate Node Thermal resistance Heat source Copper thickness Fig.12 Analytical l of FTC. Node Thermal resistance Heat source 0.1mm Cu 6.3W 14 13 12 11 10 9 8 U 15 7 7 7 7 7 7 22.4 35.5 Mounting thermal resistance Experiment 2.81K/W Estimation 2.76K/W 38.2 Tc Water temperature FTC Tc Water temperature 19.7 27.7 30.2 Thermal path length 1.81mm 30mm is more effective in reducing the mounting thermal resistance. (4) The amount of filler like grease influences the mounting thermal resistance at relatively low contact pressures. (5) The flexible thermal contactor (FTC) has the feature of reducing the temperature discrepancy of s set on the circuit board. References [1] Delia, D.J., Gilgert, T.C., Graham, N.H., Hwang, U., Ing, P.W., Kan, J.C., Kemink, R.G., Maling, G.C., Martin, R.F., Moran, K.P., Reyes, J.R., Schmidt, R.R., and Steinbrecher, R.A., System Cooling design for the water-cooled IBM Enterprise System/9000 processors, IBM J. of Res. And Dev., Vol.36, No.4, pp.791-801, Jul., 1992. [2] Chu,R.C., and Simons,R.E., COOLING TECHNOLOGY FOR HIGH PERFORMANCE COMPUTERS DESIGN APPLICATIONS, Cooling of Electronic Systems, Kluwer Academic Publishers, pp.71-95, 1993. [3] C.V.Madhusudana, Thermal Contact Conductance, Springer-Verlag New York Inc., 1996. Fig.13 Analytical result. The analytical result is shown in Fig.13. The calculated value of the mounting thermal resistance of the FTC is 2.76 K/W, and agrees with the experimental value of 2.81 K/W. Temperature drop through the thermal path is very large. It is hence expected to improve the mounting thermal resistance if the cross sectional area of the copper as thermal conductive material can be made large. 6. Results (1) There are two control s in the mounting thermal resistance: and Interstitial material control. (2) The rubber is not fully effective for filling the visible air gaps to reduce the mounting thermal resistance. (3) Soft and adhesive material such as rubber A 6