High power IGBT modules: thermal fatigue resistance evaluation of the solder joints

摘要

IGBTs (Insulated Gate Bipolar Transistors) progressively replace thyristors and GTOs (Gate Turn Off thyristors) in traction applications like high speed trains or subways [1]. Corresponding power module packaging and especially solder joints must be optimized from the design because of the very severe environmental and operation conditions, as well as the 30 years lifetime requirement. In addition, legal and technological factors are pressing for alternatives to Pb-based solders which may lead to the exclusive use of Pb-free solder alloys in the next years [2,3]. Such power modules are typically composed of numerous IGBT chips connected in parallel and soldered on a DBC (Direct Bonded Copper) substrate which is soldered on a thick copper baseplate. Two interconnection levels have to be carefully designed: the solder joint between he ceramic substrate and the copper baseplate, and the solder joint between the silicon die and the substrate. Indeed, these solder layers will fatigue and crack under thermal cycling because of the mismatched expansion on either side. Recently, a composite material (AlSiC) has been successfully used [4,5] instead of copper for the baseplate. Its coefficient of thermal expansion (CTE) can been nicely adapted to the ceramic one. Yet this material is still quite expensive. This paper presents an extensive study of the thermal fatigue resistance of several solder alloys including one Pb-free composition. Accelerated aging test campaigns have been carried out to estimate the long term reliability of solder joints. The two solder interconnection levels have been tested separately with two different test structures The objectives of this study were to make a ranking of the different batches under test and to evaluate the impact of the main technological parameters of our partner on the thermal cycling resistance of solder joints in order to provide general guidelines for the packaging designers. As regards the die attach, thermal impedance measurements have allowed to evaluate the damage in the solder joints. As far as the substrate-to-baseplate solder joint is concerned, the ranking has been based on crack growth r te measurements on cross-sectioned cycled samples dud on the comparison of the cracked areas obtained by SAM (Scanning Acoustic Microscopy). Furthermore, SEM (Scanning Electronic Microscopy) and EDX (Energy Dispersive X-rays) analyses have allowed to study the failure mechanisms by observing solder microstructure evolution and crack propagation. This work has been funded by the European Community under the Brite EuRam contract BE95-2105 named RAPSDRA (Reliability of Advanced Power Semiconductors Devices for Railway Applications), from 1996 to 1999.