ABSTRACT
This chapter briefly presents a comparison among the conventional, symmetric dual-k (SymD-k), and asymmetric dual-k (AsymD-k) tri-gate architectures based on device/static random-access memory (SRAM) parameters such as Vth, ION, IOFF, and static noise margins (SNMs) under random and parametric variations. There are various statistical variability sources for a FinFET-based SRAM that affect the stability, and these issues are required to be addressed. Random discrete dopant fluctuations (RDFs), gate oxide roughness variations (TOX), and metal-grain-dependent work-function variations (WFV) increase the spread in threshold voltage; and thus, the on- and off-currents as the device is scaled down in the nanoscale regime. The sensitivity analysis provides a relative significance of each device parameter on the performance metrics. As transistor scaling continues, self-averaging of device properties for individual devices becomes less effective; and therefore, the statistical variability of device properties becomes more prominent. The Impedance Field Method (IFM) in Technology Computer-Aided Design (TCAD) provides fast, convenient, and accurate alternative for statistical variability analysis.
