Assessing the impact of uncertainty in physics-of-Failure analysis of microelectronics damage

The primary intent of this research is to identify and demonstrate a quantitative methodology for assessing uncertainty in a Physics-of-Failure (PoF) model of a vibrating environment. Solder joint fatigue-life models are used for analysis. Identifying an approach for quantifying the impact of uncertainty on input parameters enables the determination of more realistic confidence limits on PoF predictions. We can decompose the elements of any mathematical model into the physical parameters used in forming the model equations, such as the material properties or geometric dimensions of the model structure. These parameters result from simplifications, assumptions, and approximations. Variability in these physical parameters can be addressed using existing stochastic methods, which are designed to propagate probability distributions of the parameters through a fixed model structure to estimate the statistics of interest on the model response quantities. This paper focuses on uncertainty analysis, that is, how uncertainty in the input data affects the uncertainty in the output data. We assume the model is correct and only address uncertainties arising from variations in the input parameters. Output uncertainty, that is, uncertainty in the correction factor, is determined using Monte Carlo simulations.