Finite element modeling of BGA packages for life prediction

The life-prediction analysis of an electronic package requires a sequence of critical assumptions concerning the finite element models, such as the slice model and a global model with or without a submodel. Although specifics of such analyses are available in the literature, a comparison among themselves against the same electronic package with measured life cycles does not exist. This study addresses the questions arising during the life-prediction analysis by considering two particular packages with different geometric and material compositions. Solder joint reliability analyses for each of the packages are performed by constructing the following commonly accepted finite element models: (1) nonlinear slice model, (2) nonlinear global model with linear super elements, (3) linear global model with nonlinear submodel, (4) nonlinear global model with a nonlinear submodel, and (5) nonlinear global model. In the global analysis with a submodel, the displacement fields obtained from the global analysis are extrapolated for different temperatures and applied as boundary conditions in the submodel of the solder joint with a refined mesh. In the nonlinear analysis, four thermal cycles are sufficient to achieve a stable hysteresis loop. The volume-weighted plastic work density and the characteristic life are calculated in all cases. The life predictions based on widely accepted Darveaux empirical constants are compared to experimental measurements. This study helps to identify the effects of certain modeling assumptions on characteristic life predictions.