부산대학교 도서관

Investigations on device variability for three different emerging field-effect transistor (FET) technologies are performed to determine the statistical dependence or independence of line edge roughness (LER) and random dopant fluctuation (RDF) variability mechanisms. The device candidates include standard inversion-mode (IM) FinFETs, junctionless (JL) FinFETs, and tunnel FETs (TFETs) designed for sub-32-nm generations. Using technology computer-aided design simulations, extracted standard deviations in linear and saturation threshold voltages ($V_{T,{\rm lin}}$ and $(V_{T,{\rm sat}}$), ON-state current $(I_{\rm ON})$ , OFF-state current $(I_{\rm OFF})$, subthreshold swing (SS), and drain-induced barrier lowering (DIBL) are compared for the cases: 1) when LER and RDF are separately modeled during device simulations and assumed to combine in an uncorrelated fashion, and 2) when LER and RDF are simultaneously modeled in device simulations and no assumption is made about their interaction. After performing the comparisons for each FET technology, we find that LER and RDF cannot be considered independent for IM-FinFETs and TFETs, but can be for JL-FinFETs. The different outcomes are related to local versus distributed variability dependencies in each transistor type. Our conclusions reinforce the need for more comprehensive treatment of variability effects to provide accurate estimations of expected device variability in junction-based FETs.