Virtual prototyping and qualification of board level assembly

The interest in solder joint reliability has increased by the introduction of BGA type of packages. Many end-customers require lifetime prediction data with respect to board level reliability. As it is time consuming to determine the endurance performance of each package separately, we have to look for means to reduce test time. A possible method is to actually determine the lifetime of a few products within one product family, and use the obtained results to verify and improve advanced simulation-based prediction models. The improved models are subsequently used to predict the reliability performance of the entire product family in question. A new and innovative approach to accurately predict the board level performance of micro-electronic packages is proposed. This approach uses contact bodies, time- and temperature-dependent material properties, precisely modeling the shape of critical solder joints, and beam elements for other joints. Traditionally, global-local and sub structuring techniques are used, which do not allow for: the correct shape of the solder ball; time- and temperature-dependent material properties; and the processing/manufacturing history to be taken into account. Comparison with a series of thermal board level reliability tests prove the effectiveness of the new approach, compared to the more traditional one. Where the traditional approach under predicts measured mean time to failures on average by 25-40%, the new approach can obtain prediction accuracy within 5%. Our future goal is to develop a board level assembly design and qualification tool. This tool is based on the RSM's generated from the reliability prediction models. For certain given board design, packages and assembly processes, this tool can provide the designers with solder life time, board warpage and cracking predictions. In case that the probabilistic data of the design parameters are known, the histogram and probability of failures can also be generated.