UC Riverside

The aggressive downsizing of the transistor will continue for at

least another decade. The critical feature size (physical gate

length) of transistors will drop to 5 nm in 2020 (referred to as the

11 nm technology node). In the sub-10 nm range, a variety of

low-dimensional materials and structures are being considered to

increase device and circuit performance. Examples are semiconductor

nanowires (NWs), carbon nanotubes (CNTs), and single-atomic layers

of carbon called graphene.

In order to investigate the performance, understand the physics,

propose device design, and guide experiments of nanometer scale

complementary metal-oxide semiconductor (CMOS) devices with

one-dimensional (1-D) novel channel materials, such as III-V NWs, a

generalized quantum mechanical modeling and simulation approach is

undertaken in this dissertation. We have developed models and

simulation tools, derived theory to understand and investigate III-V

NW field-effect transistors (FETs) for next generation high-speed,

low-power logic applications. These alternative materials and

geometries are being investigated for two different types of

transistors, (a) standard FETs, and (b) band-to-band tunneling FETs

(TFETs).

In the first part of the dissertation, we have investigated the key

device metrics such as the quantum capacitance, the drive current,

the charge, the power-delay product, the energy-delay product, and

switching frequency of NW FETs based on InSb, InAs, and InP

materials. We have identified two operational regimes for these

nanoscale devices, namely, the quantum capacitance limit (QCL) and

the classical capacitance limit (CCL). It is shown that n-type NW

FETs upto <=50 nm in core diameter operate in the QCL, and the

corresponding p-type devices operate in the CCL. Drive currents at

a fixed gate overdrive for the n- and p-type devices are found

to be well-matched. Significant performance improvement in terms of

device metrics are predicted for devices operating in the QCL.

In the second part of the dissertation, we have investigated III-V

NW and CNT TFETs. A generalized approach to quickly determine the

drive current as a function of materials, diameter, and electric

field is developed. It is found that a CNT with the same bandgap as

a NW can provide 10x drive current. We have developed a

general non-equilibrium Green's function (NEGF) based approach

within recursive Green's function (RGF) algorithm to investigate the

effects of `band-tails' on the subthreshold characteristics of

TFETs. Band-tails can result from heavy doping, impurities, and

phonons. We show that band-tails resulting from necessary heavy

doping of the source are not a show-stopper for TFETs.