Electromigration (EM) is a major reliability concern for interconnects in advanced technology nodes. Most of the existing EM models are either empirical or calibrated based on finite element analysis. Most of them consider only EM failures in the line without considering the via. Furthermore, the existing EM models do not model variations in the EM induced failure times, as typically observed in measurements. In this work, we develop a variation-aware EM analysis framework to model the bimodal failure distribution with early failures in via along with late failures in line. This EM model can be used for material and dimension exploration while being able to model and predict the variations in the bimodal EM failure distribution at various operating conditions.
