A Unified Physics-Based Stochastic Model for EM-Induced Resistance Degradation in BEoL Interconnect Segments

Accurate prediction of electromigration (EM) induced mean time to failure (MTF) of on-chip power/ground grids with advanced EDA tools [1] requires thorough knowledge of several randomly distributed physical parameters such as copper grain size, atomic diffusivity, copper resistivity, critical stress. Keeping in mind the dependence of distributions of these parameters on interconnect segment geometries makes this demand even more challenging [2], [3]. To be able to address this challenge, an accurate unified stochastic model describing the evolution of resistance of segments with arbitrary geometries, loaded with arbitrary current densities at specified temperatures should be available. Random physical parameters and their distributions, which were calibrated for specific line geometries and loads based on electrical measurements, could be projected on a variety of different geometries and stressing scenarios. The aim of this paper is to present such a model and to demonstrate its predictive power by comparison with the wide specter of experimental data from the 12 nm GlobalFoundries FinFET technology node.