Comparative Sensitivity Analysis for μBGA and QFN Reliability

The estimation of product reliability during the design is one of the key questions in microelectronics. The assembly lifetime is a function of such parameters as geometry, material properties and loads. All these influences exhibit systematic and stochastic variations. The effect of variability can be analysed by a probabilistic FE-simulation and statistical methods. This paper presents an approach for the prediction of thermal fatigue life of two CSP types, the μBGA and the QFN. Besides geometry parameters, also material properties and cycle temperatures are used as variable inputs. Based on a preliminary study the input parameters were defined as normal distributed. Sensitivities of the lifetime to the design parameters were computed and ranked after FE-simulations for both μBGA and QFN packages parameters had been performed. The fatigue life prediction of solder joints used in this work is based on a Coffin-Manson model and it was performed using the stress-strain data extracted from FE-simulations. As there exists a dependency on the solder deformation behaviour, the correct choice of the deformation model of lead-free solder alloys is an important aspect of this work. Summarising, in this work a probabilistic simulation method was developed to compute realistic failure distributions of two CSP types.