A thermal fatigue life prediction model of J-lead solder joint assembly has been developed. This model is evolved from an empirically derived formula based on modified Manson-Coffin fatigue life Prediction theory. To estimate solder joint fatigue life, nonlinear finite element analysis (FEA) was conducted using the ABAQUS™ computer code. The analysis results show that cracks are initiated and propagated from both the heel and the toe of the solder joint toward the center portion of the joint. This condition results in the solder joint fatigue life degradation and is included in the model development. The fatigue life prediction model is then calibrated to life cycling test results, which were provided by Jet Propulsion Laboratory (JPL/NASA). The developed life prediction model, combined with the nonelastic strains derived from FEA and Miner’s cumulative damage law, was used to predict the cumulative damage index of the solder joint under NASA’s thermal cycling environment (between −55°C and 100°C). The analysis results indicate that this solder joint has a 50 percent failure probability when the solder joint is exposed up to 5206 thermal cycles. To shorten the test time, a modified thermal cycling profile was proposed. This profile is the same as the NASA thermal cycling environment except using the high end of the dwell temperature at 125°C. The analysis results show that a 50 percent failure probability of the solder joint would occur after the solder joint is exposed to 3500 cycles of the NASA thermal environment and followed by 1063 cycles of the modified thermal profile. In conclusion, the developed life prediction model is recommended to serve as an effective tool to integrate the process of design selection, quality inspection, and qualification testing in a concurrent engineering process. It is also recommended to conduct a micro-section in the solder joint to verify the solder crack paths and further validate the life prediction model. When additional thermal cycles have been added into the test specimens, recalibrating this model by test is also recommended.
