I. Introduction
With the continuous increase of power dissipation in modern integrated chips, the need to realize devices with subthreshold swing (SS) less than 60 mV/decade becomes more and more urgent. The theoretical limit of SS with 60 mV/decade in conventional MOSFETs is set by Boltzmann distribution at room temperature. This limit could potentially be broken by a new type of devices that emerged recently by exploiting the concept of negative capacitance. This negative gate capacitance is able to amplify the surface potential at the channel of the NC-FETs relative to the applied gate voltage, making SS less than 60 mV/decade and enabling a significant reduction in power consumption. In fact, significant effort and great progress have been made in NC-FETs in the past few years, including basic theories [1 , 2] and experimental demonstrations in the device level [ 3 – 6 ]. However, there have been very limited works in evaluating the performance of NC-FET-based circuits. Tapas Dutta et al . evaluated 7-nm node NC-FinFET-based SRAM in 2017 [7] . They reported that, NC-SRAM shows advantage on standby leakage power, read delay, and static noise margin as compared to conventional SRAM. However, their work only focused on one supply voltage V DD (0.7 V), and the performance of NC-SRAM at other V DD values was not investigated. In addition, the effect of damping constant ( ξ FE ), which would strongly affect the dynamic behavior of the NC-FETs and NC-SRAM, was not considered in their work. Recently, ξ FE is widely used to model the gate delay of NC-FET caused by polarization switching of ferroelectric thin film [8] . Therefore, in order to get more comprehensive and accurate prediction to guide the design of future NC-FinFET-based circuits, evaluation of NC-SRAM at different V DD with the consideration of the effect of ξ FE is important.