Multi-chip modules (MCMs) are packaging technologies that integrate multiple semiconductor integrated circuit components as bare chips onto different types of wiring boards, achieving overall encapsulation. Compared to single-chip packaging, MCMs can improve the integration density of circuits per unit volume, which is beneficial for the development of electronic devices towards higher speeds, more functions, and smaller sizes. However, with the increasing integration density and shrinking size of MCMs, especially for MCMs integrating high-power chips, multiple heat sources are present internally. The thermal interaction between these heat sources is strong, resulting in high power consumption per unit volume, leading to significant chip thermal failure and degradation. Data shows that for every 10°C increase in operating temperature, the failure rate of a device increases by a factor of 1. Therefore, accurately simulating the three-dimensional temperature field distribution of high-power MCM modules and analyzing their thermal characteristics is beneficial for guiding the selection of MCM thermal design schemes and is of great significance for improving the reliability of high-power MCMs. Based on the internal structure, dimensions, and materials of the MCM manufactured by Atmel, this paper simulates the internal and package surface temperature field distribution of the MCM under typical operating modes and natural convection environments. The paper also analyzes the heat dissipation ratio of each part of the MCM during operation and the influence of the thermal conductivity of each part of the MCM on the internal temperature. 1. Calculation Model 1.1 Actual MCM Structure Figure 1 illustrates the internal structure of a certain model of MCM manufactured by Atmel. This MCM contains three chips mounted on an Al2O3 substrate using flip-chip bonding (with a 0.29mm thick adhesive layer between the chips and the substrate). The chip on the left is the CPU, measuring 8.5mm x 7.62mm x 0.65mm. The chips on the right are two memory chips of equal size, measuring 9.5mm x 6.82mm x 6.5mm. The substrate measures 25mm x 21mm x 2.2mm, and its back side is connected to the PCB by an array of 255 solder balls. The solder balls have a diameter of 0.8mm and a center-to-center distance of 1.27mm. The PCB measures 90mm x 50mm x 1.5mm. 1.2 Finite Element Model To facilitate calculation and analysis, the model is simplified and assumed as follows: (1) When the MCM is working, its internal power devices are in thermal equilibrium and its junction temperature distribution is stable. (2) The CPU and memory inside the MCM are the main heat sources. The Joule heat generated when the current flows through the resistance is ignored. (3) The temperature of the bottom surface of the MCM is assumed to be constant, which facilitates the measurement by the infrared thermal imager. (4) The heat exchange coefficient between the package surface and the PCB surface and the surrounding air is assumed to be constant. The model is built using ANSYS based on the structure described in Figure 1, and then the model is meshed. The author used the "Mesh Tool" provided by ANSYS for mesh generation, selecting SOLID70 mesh elements and using "Smart Size" to control the mesh size. The mesh quality is controlled by the size level within "Smart Size," with size level 1 providing the finest mesh and size level 10 providing the coarsest mesh. Considering the relatively simple structure and low heat flux density of the PCB, a size level of 8 was chosen, while the more complex structure and higher heat flux density of the package were handled with a size level of 6. The internal structure of the meshed finite element model is shown in Figure 2. The package finite element model consists of 209,803 elements and 45,222 nodes, while the PCB finite element model consists of 235,284 elements and 51,317 nodes. The materials used in the models and their thermal conductivity are shown in Table 1. [align=center]Figure 2 Finite Element Model of MCM Internal Structure[/align] 1.3 Model Boundary Conditions The thermal boundary conditions of the model are as follows: The heat source is the heat generated when the chip is working, where the heat generation power of the CPU is 2.6W, and the heat generation power of the two memory chips is 15mW respectively; heat is transferred between the materials inside the model through conduction, which obeys Fourier's law of heat transfer; the outer surface of the model dissipates heat through convection and radiation with the air. Convection heat transfer obeys Newton's law of cooling, and thermal radiation obeys the Stefan-Boltzmann law. The external ambient temperature is 16℃, and under natural air convection, the convective heat transfer coefficient is taken as 25W/(m2·K). The emissivity of the PCB is 0.9, and the emissivity of the encapsulation surface is 0.8. 2 Simulation Results and Analysis After simulation by ANSYS, the overall temperature distribution of the model is shown in Figure 3. It can be seen from the figure that the highest temperature point of the model is located at the CPU, with a value of 60.249℃. The surface temperature of the MCM gradually decreases outward from the CPU, forming an arc-shaped temperature profile. The lowest temperature point of the MCM is located at the two corners on one side of the memory. The MCM's heat generation has a very small impact on the PCB temperature range. Within this range, the PCB temperature gradually decreases outward in an arc shape, centered on the MCM. The highest temperature of the MCM under typical operating mode, measured by an infrared thermal imager, is 62.67℃. Using Equation 1, the error is calculated to be 5.19%, which is relatively small. This indicates that the model and finite element analysis method can accurately reflect the temperature distribution of the MCM and can be used for temperature analysis. Figures 5 and 6 show the temperature distribution of the solder balls and the substrate, respectively. As can be seen from the circles, the temperature distribution of the solder balls and the substrate is basically the same. This is because the solder balls and the substrate are connected, and the heat emitted by the chip is conducted to the solder balls through the substrate. The area of ​​the substrate in contact with the air is relatively small, resulting in minimal heat dissipation through convection and radiation; most of the heat is transferred to the solder balls through conduction. Based on the above simulation analysis results, the heat dissipation power of the PCB, encapsulation, and substrate was calculated separately under the condition of no heat sink. The results show that: the heat dissipation of the PCB (including convection and radiation) contributes the most to the heat dissipation, accounting for 50.76% of the total heat dissipation; followed by the heat dissipation of the encapsulation (including convection and radiation), accounting for 29.8% of the total heat dissipation; and then the heat dissipation of the substrate (including convection and radiation), accounting for 17.92% of the total heat dissipation. 3. Analysis of the influence of material thermal conductivity on the internal temperature of the MCM From the above analysis results of the heat dissipation distribution of each part, it can be seen that most of the heat in the model is transferred from the PCB, encapsulation, and substrate. The heat dissipated by the encapsulation is generated by the chip, conducted through the encapsulation to the encapsulation surface, and then transferred to the surrounding air by convection and radiation; the heat dissipated by the PCB is generated by the chip, conducted through the substrate and solder balls to the PCB, and then transferred from the PCB surface to the surrounding air by convection and radiation. Therefore, when the heat transfer coefficient and emissivity are constant, the thermal conductivity of the encapsulation and base affects the heat dissipation of the model. Based on the above model and finite element analysis method, the influence of the thermal conductivity of the encapsulation and substrate on the internal temperature of the MCM is studied by changing the thermal conductivity of the encapsulation and substrate respectively. Figure 7 shows the change curve of the highest internal temperature of the MCM when the thermal conductivity of the substrate increases from 0.5 to 125, while other parameters remain unchanged. As can be seen from Figure 7, the highest internal temperature of the MCM decreases significantly when the thermal conductivity of the substrate increases from 0.5 to 20, while the temperature decrease tends to be stable when it increases from 20 to 125. Figure 8 shows the change curve of the highest internal temperature of the MCM when the thermal conductivity of the encapsulation increases from 0.1 to 25, while other parameters remain unchanged. As can be seen from the figure, when the thermal conductivity of the encapsulation increases, the highest internal temperature of the MCM generally shows a decreasing trend, but when the thermal conductivity increases from 0.1 to 0.8, the temperature shows an increasing trend. The analysis results of the heat dissipation ratio of each part of the model and the influence of the thermal conductivity of each part on the internal temperature of the MCM are basically consistent. The reason for this is that the heat dissipation of the PCB and the encapsulation accounts for a very large proportion of the total heat dissipation. Increasing the thermal conductivity of the substrate makes it easier for the heat emitted by the chip to be transferred to the PCB, while increasing the thermal conductivity of the encapsulation is beneficial to transferring the heat emitted by the chip to the packaging surface, where it is dissipated through convection and radiation. 4 Conclusions Taking the MCM produced by ATMEL as the research object, the temperature distribution of the MCM was simulated and analyzed using the finite element method. The results show that: (1) The established model and finite element analysis method can accurately simulate the temperature field distribution of the MCM, providing an effective method for the thermal simulation of the MCM. (2) The heat dissipation ratio analysis results of each part of the model show that, in the absence of a heat sink, the heat dissipation of the PCB contributes the most to the heat dissipation, followed by the heat dissipation of the encapsulation. (3) The analysis results of the influence of the thermal conductivity of each part on the internal temperature of the MCM show that increasing the thermal conductivity of the encapsulation and the substrate can effectively improve the heat dissipation effect of the MCM and reduce the internal temperature of the MCM.