Effect of Voids on Thermo-Mechanical Reliability of QFN Solder Joints

QFN packages have gained much popularity in the consumer electronics because of the thin profile, low inductance and low weight. This has made QFN packages much desirable in the harsh environment applications like oil and gas industry, under the hood automotive electronics as well. Though the effect of macro-voids on thermo-mechanical BGA reliability is well published upon, little literature can be found on effect of macro-voids of QFN solder joint reliability subjected to HTSL thermal cycling loads. Also the dimensions of the solder joint of a QFN package are very different comparison to its counterpart in BGA packages. This inherently brings the macro-voids closer to the crack path on the package side. It is widely reported that, for HTSL/HTOL loading conditions, macro-voids in solder balls depending on their size and location near the package side or the PCB side may cause an accumulation of shear stress in comparison to a no-void solder ball. When the void may be in the path of the package side crack path, the effective length of the solder joint reduces resulting in early failure. Lall et al have published earlier on micro-CT data based FE modeling of QFN solder interconnects with voids for one thermal cycle, considering linear material model for the solder [Lall 2019]. Researchers have included artificially modeled voids to study the effects of voids on thermal cyclic fatigue reliability of lead free solder interconnects of BGA packages. [Terasaki 2005], [Schwerz 2011] [Yubing 2005]. In this paper, a detail study has was undertaken using finite element modeling of a 32 pin QFN package solder joints with artificially introduced voids. The FE model was done on QFN solder joints with artificially introduced voids of different sizes and locations. A thermal cyclic load of 150
⁰
C to -40
⁰
C as per AECQ-100 was used for the study. Effect of void size, void location and number of voids on non-linear plastic work per volume after two thermal cycles was tracked at different cross-sectional locations and effect of each factor was individually quantified using a full factorial DOE approach.