Issues in Junction-to-Case Thermal Characterization of Power Packages with Large Surface Area

There are several ways to define the junction-to-case thermal resistance; however, it is rather challenging to characterize the heat-flow in a package by a single number in an accurate and reproducible way. For many power package families such as TO-type packages the thermal transient testing and the so-called dual interface method can give reliable results. The diverging point of structure functions from dual thermal transients gives a good picture of the material interfaces in such structures. However, the location and nature of the diverging point strongly depends on the shape and direction of the heat-spreading. If the package area is much larger than the dissipating chip the shape of the heat-flow changes when using different interfaces [1,2]. This causes structure functions corresponding to the two setups deviate much before reaching the case surface. In this paper the origin of this phenomenon is investigated. Measurement and simulation results are compared on different large IGBT modules with several modifications in their structure enabling a detailed analysis of the heat-flow path. A comparison is given between heating only a small fraction of a large module and heating all chips. Some samples went through thermal cycling reliability tests which resulted in cracks below the chips. The effect of the reduced die-attach area is visualized with the help of structure functions.