Abstract
Reliability is essential in electronic packaging development. Since low-cycle fatigue is a common WLP (Wafer-Level Packaging) solder joint failure mechanism, industries often apply the accelerated thermal cycle test to estimate the risk experimentally. Due to the cost and execution time of the accelerated thermal cycling test (ATCT) being high and long, so the finite element simulation is critically needed to replace the ATCT experiment. To numerically estimate the solder fatigue risk, equivalent plastic strain-based Coffin-Manson and energy density equations are often applied. However, only the energy density equation works well while time-dependent material nonlinearity is applied.In this research, we build five finite element models corresponding to five validated test vehicles. These test vehicles have a lifetime range from 313 to 1013 cycles. For reliability estimation, we use the energy density-based method. The modified energy density method is applied to establish the correlation between the strain energy and the lifetime. We also employed two creep models, Garofalo Hyperbolic Sine model and Anand model, to analyze five test vehicles, subsequently comparing the reliability predictions calculated using the modified energy method with the increment of strain energy density. After calculation, both the Garofalo and Anand models demonstrated the ability to provide accurate results, with errors measuring less than 20%. Notably, the Anand model outperformed the Garofalo model in terms of accuracy.