Effects of interface roughness scattering on radio frequency performance of silicon nanowire transistors

Abstract

The effects of an atomistic interface roughness in n-type silicon nanowire transistors (SiNWT) on the radio frequency performance are analyzed. Interface roughness scattering (IRS) is statistically investigated through a three dimensional full-band quantum transport simulation based on the sp 3 d 5 s * tight-binding model. As the diameter of the SiNWT is scaled down below 3 nm, IRS causes a significant reduction of the cut-off frequency. The fluctuations of the conduction band edge due to the rough surface lead to a reflection of electrons through mode-mismatch. This effect reduces the velocity of electrons and hence the transconductance considerably causing a cut-off frequency reduction.Keywords: interface roughness scattering, silicon nanowire transistor, RF, cut-off frequency Since the lengths of silicon (Si) metal-oxidesemiconductor field effect transistor (MOSFET) have been scaled down to the sub-100 nm regime, the cutoff frequency has increased significantly to reach hundreds of gigahertz (GHz) 1-3 . Even though the cut-off frequency is not the only important parameter in radio frequency (RF) MOSFETs, a high cut-off frequency certainly represents a good criterion for Si MOSFETs to catch up with III-V transistors if other shortcomings are overcome. Power losses due to a long skin depth of the Si substrate, a poor noise figure and a high gate resistance 4 are the examples of such obstacles. Recently there have been tremendous efforts to improve the RF performance of the Si MOSFET and it is becoming competitive to III-V high electron mobility transistor (HEMT)/heterojunction bipolar transistor (HBT) or silicon germanium (SiGe) HBT 2,3,5 . Silicon-on-insulator (SOI) multi-gate (MG) structures also have been found to be capable of achieving the cutoff frequency predicted by the international technology roadmap for semiconductors (ITRS) 6 for RF applications while reducing substrate losses and noise figures 7 . Gateall-around (GAA) siicon nanowire transistors (SiNWTs) have attracted attention since it was found that their cutoff frequency can be much larger than that of planar Si MOSFET 8 . Traditionally, interface roughness scattering (IRS) has been considered as one of the most important scattering mechanisms. At a high effective electric field, IRS dominates the universal mobility trend 10 . In SiNWTs, IRS is still an important scattering mechanism reducing the oncurrent and the mobility significantly from the ballistic values 11 .a) Electronic mail: kim568@purdue.edu.This paper focuses on the effects of interface roughness scattering on the RF performance of SiNWTs, especially on the cut-off frequency (f T ). For that purpose, a three dimensional full-band quantum transport simulator based on the spis used. As the maximum oscillation frequency (f max ) -another important figure of merit of the RF MOSFETs is directly related to the cut-off frequency 8 , the effects of IRS on the theoretical limit of the SiNWT's RF performance can be estimated through this study.The structure of the SiNWT studied in this pap...

Time-dependent simulation of particle and displacement currents in THz graphene transistors

“…where the current I(t) can be measured at the gate, source or drain. The value ⟨ ⟩ I is the time average value of current I(t) during a large period of time T: Then, the Fourier transform of equation (10) gives the PSD of the noise ( ) ω S :…”

“…There are many successful examples in the literature on how quasi-static approaches are still capable of getting THz information of electron devices. Among many others, we mentioned those based on time-independent solutions of the Non-Equilibrium Green's function framework owing to the Klimeck's group with the NEMO simulator [10] or to Fiori and Iannaccone's group with the NANOTCAD ViDES simulator [11]. In this regard, the ab initio (time-independent ground-state) density functional theory (DFT) has also been successfully used in the literatures for such graphene THz predictions [12,13].…”