Probabilistic design approach for package design and solder joint reliability optimization for a lead free BGA package

The natural variations that exist in the material properties and manufacturing processes in the assembly of an IC device can impact the solder joint reliability. It would be desirable and more efficient if these variations were incorporated into the design so that when all variations are taken into account, the assembled device always operates away from the failure domain. It is well known that a deterministic optimum is not always a reliable solution as scatter on the inputs may propagate into scatter on the outputs and even small parameter changes may have substantial effects on the functional performance. The deterministic optimum has to be shifted to a reliable optimum. By taking into account realistic variability, it is ensured that constraints on performance targets are never violated. This requires an incorporation of a probabilistic design methodology in the FE simulations. In this paper, a complete reliability optimization case will be shown for a lead free BGA package mounted on a PCB. The parameters being optimized are the material properties of the molding compound and the dimensions of the package while the optimized responses are the warpage of the package at room temperature and the solder joint reliability. For the reliability index calculations, a state-of-the-art FORM (first order reliability method) algorithm is used. For the optimization, a new efficient global optimization algorithm is proposed, an iterative method using Kriging response surface models and smart selection of design points, that converges to the optimal solution in a minimum number of simulations.