FEM stress analysis in BGA components subjected to JEDEC drop test applying high strain rate lead-free solder material models

The influence of the strain rate dependent yield properties of solder is often neglected in the FEM analyses of JEDEC drop tests. In this paper the effect of this material behavior was tested in drop test simulations of productive BGA components, in order to evaluate the resulting stress distribution and levels in the 2^nd level interconnects. Two lead free solder alloys SnAg1.3CuO.5 (SAC1305) and SnAg3.5 (SA35) were characterized in a high strain rate tensile tester in a range of 40s^-1 to 800s^-1. The specimens were produced in a casting process with a high stress dimension close to real solder joints, in order to create similar size effects. The stress-strain behavior was recorded with high resolution using strain gauge sensors. The evaluation of the yield stress dependency on strain rate was done by the measured stress data. Hereby the SACI305 solder revealed a high sensitivity of yield stress on the applied strain rate, while the SA35 solder marginally increased its yield stress. The influence of both strain rate dependent solder models on the interconnection stress distribution was tested against a simple bilinear and an elastic material model. The simple solder models cause excessive stress in the copper pads of substrate and PCB due to their high and neglected yield behavior, respectively. The accurate solder models significantly reduce the copper stress by generating locally higher plastic deformations and a wider distribution of plastic strain in the solder balls. The different strain rate sensitivity affects the distribution of plastic strain between solder and copper. High strain rate sensitive solders reduce the plastic strain in the solder with higher PCB deformations and increase the stress and strain in the copper compared to materials with a low sensitivity. The application of strain rate dependent solder material models is necessary for realistic stress interpretations in drop test conditions. The neglect of this mate- ial behavior leads to unrealistic stress distributions and thus, it leads to wrong failure assumptions and lifetime predictions.