A Flexible and Inherently Self-Consistent Methodology for MOL/BEOL/MIMCAP TDDB Applications with Excessive Variability-Induced Degradation

In this work, we develop a flexible and inherently self-consistent grand methodology to resolve the excessive variability-induced degradation encountered in measurements of time-dependent dielectric breakdown (TDDB). This methodology is based on the underlying percolation principle in terms of thickness-dependence of characteristic breakdown time (T63) and Weibull slope (β). Starting from an ideal β thickness-vs-dependence, this methodology involves the use of an iteration procedure for the adjustment of the β -vs-thickness relation for the optimal coefficients to obtain the simultaneous agreement with multiple sets of experimental data. Moreover, we show different governing statistical distributions, (normal or Weibull), and their bimodality for dielectric thickness variation, can also play a great role in affecting final TBD failure distributions. Due to the rigor of our methodology, it can be used for both reliability assessment and lifetime projection.