Comparison between gate oxide lifetime models with Rseries and trapping effect correction in the FN-regime

This work analyzes the effect of stress parameter (i.e. E-field) correction on accurate reliability projections of SiO2 based gate dielectrics. Considered are oxide thicknesses in the non-ballistic FN dominated range, only. Irrespective of which underlying dielectric degradation mechanism is actually responsible for failure, here, an empirical approach is chosen that focuses on the stress, test and reliability assessment metric. Empirical or physically justified lifetime models depend on the outcome of well-designed stress tests. However, when adjustable stress parameters such as electrical field or indirect parameters such as dissipated energy at the anode side are affected by the degradation process itself, the goodness-of-fit to measured data is under question. Charge trapping is such an effect - it interferes with applied bias. As a result, the effective stress field to the material or the delivered energy to the anode is altered while it is constant or clearly defined assumed. Changing the viewpoint from a macroscopic to a microscopic view of inner material effects can help to consider a more suitable metric for material assessment. Following this principle, common gate oxide lifetime models are being reviewed and compared by adjusting the applied bias to the effectively acting stress bias within the material stack. As a result, competing lifetime models change their goodness-of-fit rank order. The inferred process indirectly qualifies as a selection tool to pick a more suitable lifetime model for a processed gate material. In addition the results point to possible mistakes in time dependent dielectric breakdown (TDDB) data acquisition that can influence the view on today's three most competing lifetime models.