Abstract
Dielectric breakdown in high-κ gate stacks has been intensely studied over the past few years from an electrical, physical and statistical perspective. The mechanisms of stress induced leakage current (SILC) and subsequent time dependent dielectric breakdown (TDDB) as well as post-breakdown mechanisms in high-κ dielectrics have been relatively well understood now. Taking cue from the repeatable switching phenomenon in resistive random access memory (RRAM) devices, recent reports show that the percolative damage caused by TDDB and SBD is also partially recoverable. While electrical studies confirming SBD recovery are prevalent, the effect of this recovery on the statistics of the TDDB phenomenon has never been investigated. In particular, it would be useful to have a quantitative model to predict the lifetime enhancement for different degrees of recovery and estimate the role of recovery on the shape, convexity and Weibull slope of the breakdown distributions. Such a model can also be used to estimate the overall lifetime of the device for multiple recovery cycles and the role of gradually lowering efficiency in defect passivation as the device ages. We make use of the standard percolation cell framework here to examine the stochastics of recovery and subsequent breakdown events in pure high-κ dielectrics ignoring the presence of an intrinsic interfacial layer.