A Simple Prediction Method for Chip-Level Electromigration Lifetime Using Generalized Gamma Distribution

A simple prediction method for chip-level electromigration lifetime based on the segment lifetime is proposed herein. The competing risk model, which consists of the lognormal distribution of the segment lifetime and a trend of the power law of current density distribution, is approximated by the generalized gamma distribution (GENG). The second shape parameter of the GENG, $\kappa$, can indicate whether the chip-level distribution is close to the lognormal or Weibull distribution. The proposed method, derived through Monte Carlo simulations, can provide an easy estimation of the chip-level electromigration using a competing risk level and parameters of the lognormal distribution.