(Invited) Baseplate Materials for Securing Reliability of Wide Band Gap Power Semiconductor Module Operating at High Temperatures

<Abstract>

Wide Band Gap (WBG) power semiconductors such as SiC and GaN are currently in the spotlight for realizing high power density converters. To derive their characteristics, one of the important issues of power module technology is high temperature operation. It is well known that a power module equipped with the WBG power semiconductors can operate in high T_j (Junction Temperature) as over 200 degrees C. This device feature can achieve great improvements for downsizing of cooling system of converter compared with the conventional Si power device. However the conventional power module technologies are designed to be reliable up to T_j=175 degrees C. Therefore, it is important to develop high temperature capable power module technologies which can operate in T_j=250 degrees C. At this temperature, there are no standard assembly parts as solder, seal resin and ceramic. In addition, high temperature makes assembly process harder because of huge thermal stress. Furthermore, high temperature operation induces large thermal resistance failure after long-term thermal cycle test. Regarding this background, this study aims to investigate the coefficient of thermal expansion (CTE) mismatch behavior between the baseplate and the ceramic substrate at T_j=250 degrees C. CTE mismatch causes thermal cycle reliability failure at the soldered joint between the baseplate and the ceramic substrate. Therefore, it is important to observe soldered joint condition during thermal cycle test. In particular, we conduct a series of verification tests in order to investigate the soldered joint crack propagation caused by CTE mismatch.

In this test, the samples are shown in Fig. 1. The Cu-metalized Si₃N₄ ceramic substrates are attached to the baseplates with Au–Ge solder. As the baseplate material, Al (pure aluminum) and Cu (oxygen-free copper) are selected which are used for the conventional power modules, in addition, SUS410 (chromium stainless steel) and W-Cu (89W–11Cu) that CTEs are similar to those of the Cu-metalized Si₃N₄ ceramic substrate are selected. The samples undergo the thermal cycle load of –40 to 250 degrees C and the crack propagation of the soldered joint areas are observed by Scanning Acoustic Tomography (SAT).

Fig. 2 shows SAT images of the joint areas and Fig. 3 shows the remaining joint area rate (= remaining joint area / initial joint area) as a function of the number of thermal cycles. Consequently, the smaller mismatch of CTE between the baseplate and the ceramic substrate contribute to less crack propagation of joint area in the case of Cu, SUS410, and W–Cu samples. On the other hand, the crack propagation of Al samples is unexpectedly small in spite of the largest CTE mismatch.

The discussion of the result is as follows. It is well known that the dominant factor of the plastic deformation mechanism is grain boundary sliding not dislocation in the uniform temperature, T_H = T/T_m (absolute melting temperature), exceeds 0.4. Therefore, creep deformation easily occurs in the case. While Al’s 0.4 Tm is 100 degrees C, the other materials' are higher than 250 degrees C. Therefore, in particular, plastic deformation easily occurs for Al, compared with the other three materials in –40 to 250 degrees C. This implies that the Al baseplate plastic deformation caused by thermal stress of CTE mismatch ends up with the stress relief of the soldered joint. At the result of the thermal cycle load, the existence of a fine wavy deformation trace on the rear of the Al baseplate is observed as shown in Fig. 4. That should almost be equal in area to the soldered joint of the baseplate. This physical change implies that repetition plastic deformation should occur to the Al baseplate because the wavy trace is not shaped on the baseplates of the other three materials.

In conclusion, we verify the effects of CTE mismatch between the baseplate and the ceramic substrate of the soldered joint in –40 to 250 degrees C.

As the result, it is clarified that similar CTE between the baseplate and the ceramic substrate is important in order to realize long-term reliability. On the other hand, it is also clarified that the baseplate plastic deformation contributes to better thermal cycle capability even if the CTE mismatch is large. This study will contribute to realizing the power module which is able to operate in T_j=250 degrees C.

<Acknowledgement>

This study was supported by the Council for Science, Technology and Innovation (CSTI), Cross-ministerial Strategic Innovation Promotion Program (SIP), “Next-generation power electronics/Consistent R&D of next-generation SiC power electronics” and “the Novel Semiconductor Power Electronics Project Realizing Low Carbon Emission Society” (funding agency: NEDO)