Experimental research on the thermal performance of the dry conduction plate for multi-point electronic chips

This paper designs a dry conductive heat dissipation method using several parallel heat pipes to solve the difficulty of the thermal management problem of multi-point electronic chips. The heat sources are distributed along the heat pipe, whose heat is efficiently conducted from the middle to the cold plate on the two sides. The effect of gravity, flow rate, and heat source distributions on the heat transfer performance is studied. The testing data indicated that gravity has less effect on the conduction performance of heat pipes for the cooling on two sides. As the heat sources concentrate in the centre, the heat pipe temperature difference between the centre and the side increases under the same transmission power. The results demonstrated that the two-side cooling and multi-point distribution heat conduction method is conducive to improving the thermal management capability under different inclination angles.


Introduction
With the fast growth of communication technology, the processing speed of electronic devices has been significantly improved, resulting in extremely high heat generation.If the heat cannot be removed effectively, it will endanger the stable operation.Because of the simple structure, mature technology, and low cost, heat pipes are extensively applied in electronic equipment [1].Wang et al. [2] designed a CPU cooler embedded in two heat pipes.At the 140 W dissipating power, 36% of which is transferred through heat pipes, the thermal resistance is 0.12℃/W.When four heat pipes were embedded, the overall heat transfer performance was further increased [3].Wang et al. [4] showed that the cooling capacity of H-shaped heat pipes was 22.5% higher than that without heat pipes.Kim et al. [5] investigated the thermal management of LED arrays using heat pipes.The LED junction temperature is reduced by about 24℃ after using the heat pipes.
The conduction performance relates to heat pipe inclination angles [6][7].Zhang et al. [8] tested the capability of grooved heat pipes under different inclination angles.It is concluded that an appropriate increase in the inclination angle is beneficial to enhancing the heat pipe performance.In addition to the inclination angles, the distribution of heat sources will change the heat pipe evaporation section length, affecting the performance [9].
The above literature demonstrates the heat dissipation performance through experiments.However, current analytical models make it difficult to accurately predict the temperature distribution and capacity of heat pipes in different scenarios [10][11].In addition, most research focuses on the heat transfer process from the heat to the cold side.However, the heat pipe performance at multi-point chips has rarely been reported.

2
In this paper, a conductive heat dissipation structure with several parallel heat pipes is designed aiming at the heat dissipation problem of multi-point electronic chips.The heat is transferred from the middle to the two sides to shorten the transmission distance and improve the heat dissipation performance.The influence of gravity, flow rate, and distribution of heat sources on the performance of the dry conduction plate were tested.

Experimental system and data processing
As shown in Figure 1, the test device comprises a heat pipe conduction plate, liquid cold plate, heating resistances, DC power supply, flowmeter, liquid cooling source, thermocouples, and data acquisition.The dry conduction plate is 340 mm × 275 mm ×7.5 mm.Several heat pipes with a cross area of 7.1 mm × 4.5 mm are sandwiched inside the conduction plate using vacuum welding.There are 128 heat dissipation bulges with a cross area of 15 mm ×15 mm arranged in 8 columns and 16 rows on the heat pipe conduction plate, which correspond to the position of the heat sources.Two heat pipes are embedded under each column of bulges to transfer heat from the middle to the two sides.The cold plate is installed on two sides of the heat pipe conduction plate.Each side of the cold plate is covered with two rows of bulges.The serpentine channel and fins are machined to enhance the temperature uniformity and convection heat transfer.The heating resistances are used as heat sources.The silicone thermal conductive padding is filled between the heating resistance and the heat dissipation bulge, and the graphite interface material is filled between the flow channel and the heat pipe conduction plate to decrease the temperature difference.The cooling medium is provided by a liquid cooling source to the cold plate, and the flow rate is measured through a flowmeter.48 thermocouples are attached to the base of bulges to record the temperature distribution.The cooling medium is 65# ethylene glycol aqueous solution with 35℃ temperature and 100~250 L/h flow rate.The heat pipe conduction plate is installed horizontally or vertically.At the beginning of the test, we turn on the liquid cooling source.When the flow rate and temperature of the cooling medium reach the set value, the power on the DC supply gradually increases.When the temperature of the conduction plate is stabilized, we record the corresponding heat power and temperature values.Then, we continue to increase the heat power until the temperature rises rapidly, which means the transmission limit of the conduction plate is reached.
The thermal performance of the dry conduction plate is closely associated with the internal heat pipe performance, which the equivalent thermal conductivity and conductive thermal resistance can assess.The following equation can obtain the equivalent thermal conductivity. ( ) Q is the transmission power of the half heat pipe Q=UI/32, where U and I are the recorded voltage and current, respectively.Le is the effective transmission length.A=wt is the heat pipe cross area, where w and t are the width and thickness, respectively.Te is the average temperature of the evaporation area , where j is the column number.Here, temperature is approximated by the base of the bulges.
The conduction thermal resistance is obtained from the following equation.
The uncertainty of heating power is composed of voltage (0.1%) and current (0.4%) from the DC power supply, which is less than 0.41%.The error of Le is 0.5 mm, which leads to an uncertainty of 0.36%.The uncertainty of A is composed of width (7.0%) and thickness (2.2%) due to the manufacturing error of the heat pipe, which is less than 7.3%.The error of temperature is 0.5℃, and the error of average temperature Te and Tc is 0.125℃.The uncertainty of the temperature difference is less than 8.8%.Therefore, the uncertainty of equivalent thermal conductivity and thermal resistance is smaller than 11.4% and 8.81% at low heat flux, according to the uncertainty analysis.

The impact of gravity on the thermal performance
Capillary wicks with efficient liquid transport performance are critical to operating steadily for heat pipes.At ground conditions, the liquid transport is affected by gravity.At 200 L/h flow rate and 1500 W heating power, temperature distributions at the bulge base for the horizontal and vertical placement are shown in Figure 2.Under the horizontal condition, the mean temperature difference at Rows 1~8 is 4℃, and 1.9℃ at Rows 9~16.The reason may be the difference in the welding quality between the two sides.When the heat pipe conduction plate is vertically installed, the mean temperature difference of the two sides changes.The mean temperature difference is 4.8℃ for the bulges at Rows 1~8 in the lower part.The average temperature difference in the upper part at Rows 9~16 is 1℃.For the vertical placement, the upper part of the heat pipes works in the gravity-assisted state.The evaporated gaseous working medium condenses at the upper condensation section.The condensed liquid is transported downward with the assistance of gravity, which is beneficial to the liquid flow.Therefore, the thermal performance is enhanced, and the temperature difference is reduced.For the lower part of heat pipes working in the anti-gravity state, the gaseous working medium needs to spread downward, and the condensed liquid needs to overcome gravity when transporting upward, which results in the deterioration of the performance and the rise of temperature difference.For entire heat pipes, due to the heat transfer from the middle to the two sides, the comprehensive impact of gravity on heat pipes is small.Mean temperature change is within 1℃ for the vertical placement compared to the horizontal placement.Consequently, the heat conduction performance of heat pipes is relatively stable at different inclination angles.When the heating power is 1500 W, 1720 W, and 1800 W, the average transmission power on each half heat pipe is 46.9 W, 53.8 W, and 56.3 W, respectively.The temperature difference on each side is illustrated in Figure 3.As transmission power increases, the mean temperature difference also increases.Compared to horizontal placement, the mean temperature difference under vertical conditions changes within 1℃.

The impact of flow rate on the thermal performance
The flow rate will affect the convective heat transfer and change the condensation temperature, which influences the overall performance of the conduction plate.The variation of the maximum temperature rise with the flow rate under different heat dissipation powers is presented in Figure 4.With the increase in flow rate, the maximum temperature rise of the bulges decreases.However, the reduction of maximum temperature rise does not exceed 1℃ for the flow rate larger than 200 L/h.At the low heat dissipation power of 500 W, the heat transmission for the half heat pipe is 15.6 W. The highest temperature of the conduction plate corresponding to different flow rates does not exceed 2℃, which means the flow rate has little impact on the convective heat transfer.At the 2100 W dissipation power corresponding to the 65.6 W heat transmission for the half heat pipe, the temperature decreases by 3.7℃ for the flow rate, increasing from 100 L/h to 200 L/h.As the flow rate increases to 250 L/h, the highest temperature is reduced by only 0.2℃.Under the condition of a high flow rate and low heating power, the convection temperature rise is small.Consequently, further increasing the flow rate in the channel has limited improvement in the heat dissipation performance.

The impact of heat source distribution on the thermal performance
The conventional application of heat pipes is to absorb heat at one end, transmit through the middle adiabatic section, and condense at the other.In this paper, multi-point heat sources are distributed on the entire heat pipe.The experimental results of Section 3.1 showed that the heat dissipation performance is better for the multi-point distribution, which can reduce the temperature difference among different chips.Consequently, heat source distributions have a significant impact on the thermal performance of heat pipes.The thermal performance of heat pipes under horizontal placement for five different heat source distributions is studied in this section, as shown in Figure 5. Condensation sections are located on the two sides with the same contact area.When heat sources are uniformly distributed along the heat pipe, the entire heat pipe acts as the evaporation section, overlapping with the condensation section on two sides.As the number of heat sources decreases and concentrates towards the centre of the heat pipe, the evaporation section length decreases while the adiabatic section increases.When the heating power is 1500 W (47 W transmission power for the half heat pipe), the temperature distribution of the bulges corresponding to different heat source distributions at 200 L/h flow rate is shown in Figure 6.Due to the power limitation of the experimental device, the 8-point heat source distribution (Columns 5~8, Rows 8~9) is set to test the ultimate conduct performance.With the reduced number and concentrated distribution of heat sources, the heat pipe temperature difference continues to increase.The condensation temperature is unchanged, while the temperature of the middle area gradually increases.For the 128-point heat sources uniformly distributed along the heat pipe, the max temperature difference between the evaporation and condensation was 2.9℃.The temperature distribution is relatively flat.For the 16-point heat sources concentrated in the centre of the heat pipe, the max temperature difference increased to 10.1℃.The centre temperature rises sharply, while the change of the condensation and adiabatic temperature is not significant, which results in a peak distribution of the temperature.When heat sources are evenly distributed throughout the entire heat pipe, local heat flux is small, and the temperature increases slowly under the same heating power.The liquid evaporation occurs over the entire heat pipe.The vapor condensation only exists on the condensation sections on the two sides.The liquid is continuously evaporated with high efficiency during the transportation from the two sides to the middle, and the temperature increases slowly.The required liquid supply is small for the evaporation at the centre heat pipe, which can achieve a higher conduction performance.However, when the heat sources are concentrated at the centre, the single-point heat flux is larger, and the temperature increases sharply at the same transmission power of the heat pipe.All the condensed liquid needs to be transported over a longer distance to supply the middle heat sources, resulting in the deterioration of the heat transfer performance.Variation of the half heat pipe thermal resistance with power can be seen in Figure 7.As the heat pipe transmission power increases, the thermal resistance decreases and becomes stable.At the same transmission power, the more dispersed the heat sources distribution is, the smaller the thermal resistance is.It means the small temperature difference of heat pipes and the uniform temperature for different heat sources.Under the condition with 128 heat sources distributed evenly on the conduction plate, the thermal resistance is reduced to 0.06℃/W with the transmission power increase.The variation of thermal resistance is only 8%, indicating that the heat pipe has stable heat dissipation performance for distributed heat sources.For single-point heat sources, the heat transmission power can reach 75 W with a thermal resistance of 0.215℃/W.The temperature difference can reach 16℃.The equivalent thermal conductivity is about 14000 W/(m• K).

Analysis of the dry conduction plate performance
According to the above discussion, a simplified conduction model is established to analyse the temperature variation for different heat source distributions.The conduction model for a half heat pipe is shown in Figure 8, where xL is the length of half heat pipe, xc is the length of the condensation section, xe-xc is the length of the adiabatic section, and xL-xe is the length of the evaporation section.According to Fourier's law of thermal conductivity, local temperature difference at a certain position x at the heat pipe can be described by: where Ke is the equivalent thermal conductivity with 14000 W/(m• K).Qx is the transmission heat power at position x.The heat power at the condensing section is assumed as 0, while at the adiabatic section is Q.The heat power linearly increases along the evaporation section from 0 in the middle to Q at the adiabatic section.T is the heat pipe temperature difference between the evaporation and condensation.
When the heat power applied to the half heat pipe is 49 W (the whole dry conduction plate power is 1500 W), the temperature distribution is shown in Figure 9.As the distribution of heat sources changes from dispersed to concentrated, the temperature difference between the evaporation and condensation gradually increases, consistent with experimental results.The transmission power through the adiabatic section is constant, and the temperature curve is an inclined straight line.At the evaporation section, the heat transmission power decreases, and the temperature curve slope decreases continuously.The transmission power at the centre is 0 with a plain temperature curve.For the concentrated distribution of 16 heat sources, the heat pipe temperature difference was 11.2℃, which is consistent with the experimental results with a 10.9% error.For the dispersed distribution of 96 heat sources, the temperature difference is 6.1℃ with an error of 29.8% compared with the experimental results, which indicates that the equivalent thermal conductivity is higher for the dispersed distribution.The reason is that the heat transfer efficiency improves when the phase change region increases for the dispersed distribution of heat sources.Besides, the local heat flux is lower under the same heat transmission power for a dispersed distribution, which can achieve higher heat transfer performance.

Conclusions
In this paper, a heat pipe conduction plate is designed for the heat dissipation of multi-point distributed heat sources.With the help of heat pipes possessing the advantage of high thermal conductivity, the heat is conducted from middle to liquid cold plates on the two sides.The thermal performance of the dry conduction plate for different inclination angles, liquid flow rates, and heat source distributions is experimentally studied.
(1) The average temperature difference between evaporation and condensation at vertical placement changes within 1℃ compared with the horizontal placement.The upper half of the heat pipe works in a gravity-assisted state, which is conducive to liquid reflow.The thermal conduction performance is enhanced for heat transmitted from the middle to the two sides.
(2) For the high flow rate and low heat dissipation power condition, the convection temperature rise is small.The enhancement of the thermal performance through increasing flow rates is limited.
(3) The temperature distribution of heat pipes was quantitatively analysed using the equivalent thermal conductivity.The calculated results were the same as the experimental data.It indicated that the heat pipe equivalent thermal conductivity is higher for a dispersed distribution, which is conducive to achieving higher heat dissipation performance.

Figure 2 .
Temperature distribution of conduction plates under different placement conditions.

Figure 3 .
Figure 3.The variation of heat pipe temperature difference with transmission power.

Figure 4 .
Figure 4.The impact of flow rate on the maximum temperature rise.

Figure 7 .
Figure 7. Variation of thermal resistance with transmission power.

Figure 8 .
Figure 8. Thermal conduction model of a half heat pipe.

Figure 9 .
Figure 9.The variation of temperature with heat pipe length at different heat source distributions.