Direct Cooled Ceramic Substrate for Thermal Control of Automotive Power Electronics

As electric vehicle technology develops, manufacturers would like to move toward hotter coolants for power electronic components to reduce system level costs. Thus, unique designs of inverter designs were sought to enable operation with 105°C coolant. The proposed solution in this research incorporated flow channels into the ceramic layer of the direct-bonded copper substrate typically found in power electronic packages. The focus of this research details the design and analysis of the direct cooled ceramic substrate from the perspective of its thermal performance and innovative packaging concept.

The research was directed to pursue alumina as the substrate ceramic because of its low cost. Alumina, which has the lowest thermal conductivity among four materials considered, requires a larger substrate cross-sectional area to result in a viable design. Based on preliminary model parameters, two flow channel designs with larger alumina substrates were shown to meet the design goals. Experiments were conducted to characterize the pressure drop across metal foam inserts which were used to enhance the heat transfer in the flow channels. Other experiments were conducted to validate the thermal performance and model configuration. The results of thermal validation experiment showed that the assumed effective thermal conductivity of the metal foam – fluid matrix was too large. The small contact area between the metal foam inserts and ceramic substrate reduces the effective thermal conductivity.

Based on the data reduction method, the model parameters were modified to produce temperature distributions that better reflected the experimental data. Simulations were updated with the modified model parameters. These models showed that the cross-sectional area of the alumina substrate had to increase further in order to adequately manage the heat load.

In parallel efforts, the overall inverter package was considered. A linear manifold package resulted in the highest power density. Technical review of the inverter package raised concerns about stray inductance. Incorporating the entire inverter leg on one substrate would alleviate these losses. Future research can use the parameters determined in this work to more confidently predict the performance of direct cooled ceramic substrate designs.