Package-to-Board Attach Reliability — Methodology and Case Study on OMPAC Package

Abstract

Over molded pad array carrier (OMPAC), also called plastic Ball Grid Array (pBGA) package family offers increased I/O density at the board level; further, package size and thickness are reduced in comparison to other competing package types leading to many product applications. The OMPAC package is investigated for solder pad interconnect reliability. Elastic and inelastic finite element method (FEM) simulations are employed with experimental temperature cycling data. Linear elastic analyses are performed for various package parameters that influence the stress distribution in the solder pad interconnect for temperature loading. The package parameters considered are (a) ratio of thickness of die to substrate (tdie/tsub.), (b) ratio of width of die to substrate (wdie/wsub.), and (c) ratio of height to diameter of the solder pad interconnect ((h/d)pad). The results are evaluated based on the maximum Tresca (maximum shear stress) failure criterion. The key parameter is (h/d)pad ratio. As this ratio increases, the maximum Tresca failure criterion decreases. The performance of the solder pads are evaluated based on the maximum total permanent strain (plastic + creep) range per temperature cycle. FEM simulation was performed up to eight consecutive temperature cycles. The permanent strain (plastic + creep) range were calculated for each temperature cycle and observed that after a few temperature cycles the strain ranges stabilize. The maximum stabilized total permanent strain range is in the inner solder pad and is equal to 0.005. This maximum total strain range in the solder pad per temperature cycle loading can be used to estimate the life time of the solder pads. The FEM simulation results are correlated with the experimental data. Both the simulation and the experimental results are indicate that the outer pads are not as critical as the inner or the middle pads. The experimental results indicate that there is slightly larger probability for the middle pads to fail before the inner pads. However, the simulation data indicate that the inner pads are more critical than the middle pads. We believe that this small discrepancy in simulation result is due to the actual temperature variation on the package not being included in the present analysis.