A Comparison Between V-Ramp TDDB Techniques For Reliability Evaluation

The key methods used to convert ramped voltage test (RVT) data to long term reliability measurement data are reviewed. During the last years the discussion to replace time consuming constant voltage stress (CVS) tests with ramp testing was pushed forward, mainly due to time constraints during qualification of high-k and also low-k material as well as through results obtained for ultrathin oxides [1]. The aim of this work is to contribute to the discussion with focus on thicker oxides used for automotive applications. While a future fWLR standard currently under discussion should also clearly state applicable oxide thickness ranges, advantages and limitations of various approaches, the principle assumptions and formulas are compared and applied to different data sets with gate oxide thicknesses of 9, 10, 17 and 50 nm, respectively. The results are then analyzed, discussed and compared to work based on material with much thinner/thicker oxides.