Comparison between intrinsic and integrated reliability properties of low-k materials

Using a dedicated test vehicle (low-k planar capacitor) for studying the intrinsic properties of low-k materials and using standard single damascene 50 and 90 nm ½pitch test vehicles, differences in reliability behavior between intrinsic and integrated SiOCH porous low-k materials were investigated. The studied parameters were leakage current, breakdown field, distributional shape of failure times and TDDB lifetimes. Compared to the intrinsic material, the integrated properties significantly deteriorated and these differences were quantified. The low-k planar capacitor test vehicle was also used to study the intrinsic breakdown behavior at low fields. Using statistical simulations, we found that the proper choice of high field conditions is crucial to be able to discriminate between different lifetime models at low fields and recommendations are given on how to choose these high field conditions. Based on tests of several months, the E- and 1/E-model can be excluded (all test material had a standard TaNTa barrier between copper and dielectric). The √E-model and the power law are statistically not different, but a strong tendency towards the power law is observed. As a corollary, we compared the benefit of choosing less conservative lifetime models for extrapolations to lower fields to the loss that occurs when predicting median failure times to lower percentiles. We show that process optimizations leading to a lower spread in the failure times are more important than choosing less conservative lifetime models.