nanomaterials

Article
Design and Analysis of Gallium Arsenide-Based Nanowire
Using Coupled Non-Equilibrium Green Function for RF
Hybrid Applications
Pattunnarajam Paramasivam 1 , Naveenbalaji Gowthaman 2, * and Viranjay M. Srivastava 2

1 Electronics and Communication Engineering, Prince Shri Venkateshwara Padmavathy Engineering College,
Chennai 600127, India
2 Department of Electronic Engineering, Howard College, University of KwaZulu-Natal,
Durban 4041, South Africa
* Correspondence: [email protected]; Tel.: +27-642-012-601

Abstract: This research work uses sp3 d5 s* tight-binding models to design and analyze the structural
properties of group IV and III-V oriented, rectangular Silicon (Si) and Gallium Arsenide (GaAs)
Nanowires (NWs). The electrical characteristics of the NWs, which are shielded with Lanthanum
Oxide (La2 O3 ) material and the orientation with z [001] using the Non-Equilibrium Green Function
(NEGF) method, have been analyzed. The electrical characteristics and the parameters for the multi-
gate nanowires have been realized. A nanowire comprises a heavily doped n+ donor source and
drains doping and n-donor doping at the channel. The specified nanowire has a gate length and
channel length of 15 nm each, a source-drain device length LSD = 35 nm, with La2 O3 as 1 nm (gate
dielectric oxide) each on the top and bottom of the core material (Si/GaAs). The Gate-All-Around
(GAA) Si NW is superior with a high (ION /IOFF ratio) of 1.06 × 109 , and a low leakage current, or
OFF current (IOFF ), of 3.84 × 10−14 A. The measured values of the mid-channel conduction band
energy (Ec ) and charge carrier density (ρ) at VG = VD = 0.5 V are −0.309 eV and 6.24 × 1023 C/cm3 ,
respectively. The nanowires with hydrostatic strain have been determined by electrostatic integrity
and increased mobility, making them a leading solution for upcoming technological nodes. The
transverse dimensions of the rectangular nanowires with similar energy levels are realized and
Citation: Paramasivam, P.;
comparisons between Si and GaAs NWs have been performed.
Gowthaman, N.; Srivastava, V.M.
Design and Analysis of Gallium
Keywords: nanowire; tight binding models; NEGF; GAA; hydrostatic strain; microelectronics;
Arsenide-Based Nanowire Using
nanotechnology; VLSI
Coupled Non-Equilibrium Green
Function for RF Hybrid Applications.
Nanomaterials 2023, 13, 959. https://
doi.org/10.3390/nano13060959
1. Introduction
Academic Editor: Jakob B. Wagner
Over the last two decades in the semiconductor industry, the advanced structure of
Received: 13 February 2023 Metal Oxide Semiconductor Field Effect Transistors (MOSFETs), from planar to multi-gate
Revised: 1 March 2023 design, has been proposed to achieve great electrostatic control over the channel. Various
Accepted: 5 March 2023 multi-gate structural designs named Double gate, Tri-gate, Pi, Omega, Top on-one side,
Published: 7 March 2023 and Gate-all-around (GAA) devices with nanotechnology approaches have been used for
forthcoming applications. The GAA device architecture has high resistance properties. It
exhibits static control on the gate over the conduction of the channel, which plays a major
Copyright: © 2023 by the authors.
role in avoiding short-channel effects [1,2]. Natori et al. [3] have proposed GAA NW to resist
Licensee MDPI, Basel, Switzerland.
(SCEs) by improving gate length and channel length scaling (LG < 5 nm and Lch < 15 nm).
This article is an open access article
The primary benefit of GAA devices is that they have a higher ION /IOFF ratio [4]. The
distributed under the terms and reduction in OFF current (IOFF ) produces a high (ION /IOFF ) ratio. Batakala et al. [5] have
conditions of the Creative Commons demonstrated the comparison between Si and GaAs GAA MOSFETs. The gain and current-
Attribution (CC BY) license (https:// driving capacity of the GaAs material were efficient and, thus, based on this channel
creativecommons.org/licenses/by/ material, was selected following the application field. The reduction in leakage current had
4.0/). been achieved by considering major features, such as smaller threshold voltage (Vth ~0.3 V),

Nanomaterials 2023, 13, 959. https://doi.org/10.3390/nano13060959 https://www.mdpi.com/journal/nanomaterials

Nanomaterials 2023, 13, 959 2 of 20

channel length (Lch ≥ 10 nm), shorter gate length, high dielectric constant oxide material,
and the dopant concentration of the source, channel, and the drain, respectively [6].
Several studies have analyzed NWs with multi-gate arrangements. Wang et al. [7]
reduced the computational complexity in SiNW by the scattering effects, and an approx-
imation of the effective mass was carried out using Buttiker probes. Arun et al. [8] have
proposed SiNW and exhibit the efficiency of the electrical characteristics by oxide thickness
(tox ) variation, dopant concentration, and thickness of the silicon (tsi ) in the nanometer
regime, respectively. Coquand et al. [9] have presented a study of channel performance
and electrostatic control for Tri-gate nanowires by reducing gate length and channel width
dimensions to the optimum limit. The optimization in the dimensions results in a low
Subthreshold Slope (SS) and Drain-Induced Barrier lowering (DIBL). The effects of electron
confinements on thin gate cross-sections surrounded by SiO2 surface roughness have been
discussed by Ramayya et al. [10]. The electron mobility was monotonously reduced due to
the surface roughness scattering effect, which was dominant.
Cresti et al. [11] have addressed the surface roughness for an electron mobility reduc-
tion in GAA and DG nanowires. The surface roughness in the device had been chosen
by transfer characteristic parameters, electron density, and low-field electron mobility.
Jin et al. [12] have discussed the carrier scattering and backscattering mechanisms. The
backscattering decreases the flow of current, electron density, and carrier velocity when
the channel’s position varies. The carrier velocity works in the same operation in both
carrier scattering and backscattering. The carrier velocity in carrier scattering may increase
perhaps increase more in higher order than the backscattering mechanisms.
Numerous studies have been done on the transport characteristics, charge density
distributions, and doping concentrations of SiNWs, which have been discussed both em-
pirically and theoretically [13–16]. The carrier densities for various cross-sections (square,
circular, elliptical, and rectangular) using Poisson Schrodinger (PS) and Drift Diffusion
(DD) simulations had been investigated with variations in the gate length and channel
position being discussed [17]. The source/drain connections were kept wider than the
channel to decrease access resistance. The framework for single and double-gate nano
MOSFETs using the NEGF coupled mode technique for NWs with six variants had been
provided by Svizhenko et al. [18]. The coupled mode effects for nonuniformity of the
transverse potential profile greatly impacted multi-gate nanowires. The transport calcu-
lations and the electrical properties, such as the transmission coefficient for Si and GaAs
with different orientations using NEGF mechanisms, had been discussed elaborately by
Luisier et al. [19–21].
Using Density Functional Theory (DFT)-based techniques, the electronic characteristics
of mechanical and crystalline properties of nanomaterials whose accuracies and efficiencies
were reported when it is subjected to the temperature [22,23]. Kumarasinghe et al. [24] have
investigated the electronic properties of pure and doped Si nanowires with dimensions
up to 10 nm using large-scale Density Functional Theory (DFT) modeling. Three steps are
used in modeling: (1) relaxation of the NW unit cell using the DFT method and extraction
of the Hamiltonian and overlap matrices, (2) mode space transformation of the imported
Hamiltonian matrices, and (3) non-orthogonal NEGF transport calculations using the mode
space Hamiltonians, in a way that is self-consistent with the Poisson’s equation.
The Tight Binding (TB) methodology had been used by many researchers to address
all the nanowire features that ultimately depend on the electronic structures. All varieties
of nanowires and nanotubes (semiconducting, metallic, oxide, and others) were carried
over in this method. The majority of research focuses on calculating the sub-band electronic
structure of technological semiconductors and its relationship to factors such as shape,
composition, and orientation, respectively [25].
Morioka et al. [26] have presented the electronic band structures of rectangular Si NWs
using sp3 d5 s* tight-binding models. This method considers one excited s* orbital, p orbitals
{px , py , and pz }, and d orbitals {dyz , dzx , dxy , d3z 2 -r 2 , and dx 2 -y 2 }. The x, y, and z coordinate
axes are set at [100], [010], and [001], respectively. The part of each atomic orbital typically
 Nanomaterials 2023, 13, x FOR PEER REVIEW 3 of 21

Morioka et al. [26] have presented the electronic band structures of rectangular Si NWs
Nanomaterials 2023, 13, 959 using sp3d5s* tight-binding models. This method considers one excited s* orbital, p3 orbit- of 20
als {px, py, and pz}, and d orbitals {dyz, dzx, dxy, d3z -r , and dx -y }. The x, y, and z coordinate
2 2 2 2

axes are set at [100], [010], and [001], respectively. The part of each atomic orbital typically
depends onon
depends thethe nanowire's
nanowire’s width.
width. TheThe symmetrical
symmetrical characterofof
character thethe d001orbital
d001 orbitalhas
hasthe the
same
same width
width dependence
dependence configuration
configuration of pof001
p,001but
, but
thethe magnitude
magnitude of of orbital
orbital d001was
d001 wasfound
found
to to
bebe lesser,
lesser, about
about nanowire
nanowire based
based fromfrom p001
p001 . .
This
This workwork concentrates
concentrates onon designing
designing a novel
a novel nanowire-basedononGaAs
nanowire-based GaAsmaterial
materialtoto
useuse
in in
RFRF hybrid
hybrid applications.
applications. To To improve
improve thethe IONION
and and IOFFratio,
IOFF ratio,various
variousmethodologies
methodologies
have
have been
been utilized.The
utilized. Theusage
usage of high ƙ dielectric
of high- dielectricmaterial,
material,such suchasasLaLa2O23O
, has shown
3 , has shown var-
ious enhancements
various enhancements to to create an optimal design to be used in RF application
optimal design to be used in RF application setups. This setups. This
paper
paper hashas been
been organized
organized as follows.
as follows. TheThe basics
basics of NEGF
of NEGF modeling
modeling havehave
beenbeen discussed
discussed in
in Section
Section 2. The2.proposed
The proposed
nanowirenanowire
structure structure
with six with
variantssix has
variants has been in
been discussed discussed
Section 3.in
TheSection 3. The mathematical
mathematical modeling ofoxide
modeling of lanthanum lanthanum
with the oxide with the self-consistent
self-consistent methodologymeth- and
theodology
divisionand of the
the division of theinto
3D problem 3D problem
1D transport into 1Dandtransport
2D Schrodingerand 2D Schrodinger
equations were equa-
tions were in
investigated investigated
Section 4. inSection
Section5 4. Section 5the
discusses discusses the lt andofanalysis
lt and analysis the work. of the work.
Finally,
Section 6 concludes
Finally, the work and
Section 6 concludes the recommends
work and recommends the future theaspects.
future aspects.

2. Basics
2. Basics ofofNEGF
NEGF Modeling
Modeling
The Theelectronic
electronicproperties of hydrogen-passivated
properties of hydrogen-passivated compound
compoundsemiconductor nanowires
semiconductor nan-
grown in different crystallographic orientations, specifically the band structures,
owires grown in different crystallographic orientations, specifically the band structures, band gaps,
and effective
band gaps,electron masses,
and effective were discussed
electron masses, were previously [27–30].
discussed Horiguchi
previously et al. [31]
[27–30]. have
Horiguchi
discussed the Silicon nanowire bandgap dependency on the wire width
et al. [31] have discussed the Silicon nanowire bandgap dependency on the wire width using effective mass
theory
usingcalculations
effective mass and theory
using the boundaryand
calculations conditions
using theenvelope
boundary between wire confinement
conditions envelope be-
potential
tween wire and the barrier height
confinement confinement
potential and the potential’s
barrier heightfiniteness. Several
confinement authors have
potential’s finite-
presented nanoscale modeling using green’s function, quantum transport
ness. Several authors have presented nanoscale modeling using green’s function, quan- modeling, density
matrix
tum calculation, and analyzing
transport modeling, density electronic devices in equilibrium
matrix calculation, and analyzing conditions
electronic [32–35].
devices in
Seone et al. [36] have
equilibrium conditions [32–35]. proposed the Gate-All-Around (GAA) Si nanowire MOSFET and
the impact of current variability on the channel’s surface roughness
Seone et al. [36] have proposed the Gate-All-Around (GAA) Si nanowire MOSFET was analyzed using
3-D real-space
and the impactnon-equilibrium Green’s function.
of current variability Mazumder
on the channel’s et al. [37]
surface have proposed
roughness GAA
was analyzed
GaAs TFET, which works under the tunneling phenomenon.
using 3-D real-space non-equilibrium Green’s function. Mazumder et al.ON The maximum I /I haveratio
[37] OFF pro-
of posed
TFET is achieved by adjusting the few electric gate insulator
GAA GaAs TFET, which works under the tunneling phenomenon. The maximum and GAA TFET channel
architecture,
ION/IOFF ratio which
of TFETwereisbeen investigated
achieved by adjusting to provide
the few theelectric
best band-to-band
gate insulatortunneling
and GAA
and potential amplification. Montazeri et al. [38] have
TFET channel architecture, which were been investigated to provide demonstrated the band
the structure for
best band-to-
III–V
band tunneling and potential amplification. Montazeri et al. [38] have demonstrated of
compound semiconductor nanowires using k.p theory calculations. The calculation the
theband
strainstructure
is used tofor determine the particular
III–V compound nanowire structure,
semiconductor nanowires and it had
using k.pbeen employed
theory calcula-
using
tions.theTheelastic theory.of The
calculation resulting
the strain is used calculated strainthe
to determine was called hydrostatic
particular strain,
nanowire structure,
which depends on the proportions of structural dimensions
and it had been employed using the elastic theory. The resulting calculated strainand is independent of the
was
total size. Ren et al. [39] have modeled nanoscale MOSFETs and estimated the scattering
called hydrostatic strain, which depends on the proportions of structural dimensions and
and backscattering coefficients using the scattering theory. The critical length and carrier
is independent of the total size. Ren et al. [39] have modeled nanoscale MOSFETs and
velocity at the source’s end and the channel’s start were identified using transport models.
estimated the scattering and backscattering coefficients using the scattering theory. The
The ballistic, dual-gate nano transistors used for digital applications with a proper choice
critical length and carrier velocity at the source’s end and the channel’s start were identi-
of the gate oxide thickness and scaling limit down to 10 nm were discussed [40,41]. Several
fied using transport models. The ballistic, dual-gate nano transistors used for digital ap-
studies have incorporated the operation of the nanowire in a ballistic regime using analytic
plications with a proper choice of the gate oxide thickness and scaling limit down to 10
models [42–45] and numerical simulations [46,47]. In the simulation study, the density
nm were discussed [40,41]. Several studies have incorporated the operation of the nan-
of states, the electron density, and the conduction band energy (Ec ) variations along the
owire in a ballistic regime using analytic models [42–45] and numerical simulations
position of the channel were investigated [48].
[46,47]. In the simulation study, the density of states, the electron density, and the conduc-
tion
3. Design band ofenergy
Proposed (Ec)Novel
variations
GaAs along the position of the channel were investigated [48].
Nanowire
The generic structure of the proposed nanowire has been designed with a rectangular
3. Design of Proposed Novel GaAs Nanowire
cross-section with dimensions of 35 nm × 4.5 nm. The source and drain of the Si and
The generic
GaAs-based structure
nanowire of the
material proposed
have nanowire
a continuous n+has beenimpurity
donor designed concentration
with a rectangular
of
(2 cross-section
× 1020 cm−3with) anddimensions of 35 nm
n donor doping of×(1
4.5×nm. The
1020 cmsource
−3 ) at and
the drain of theThe
channel. Si and GaAs-
channel
based nanowire
direction materialishave
in this situation a continuous
longitudinal to then <001>
+ donor impurity concentration of (2 × 1020
z-axis, ‘x’ determines the channel
width, and ‘y’ determines the current flow into the nanowires, as shown in Figure 1a. The
electron movement in the longitudinal z direction is based on Kinetic Energy (Ez ) and is
called Transmission Probability T (Ez ). The proposed nanowire dimensions are listed in
Table 1.
 Drain length (LD) 10 nm
S/D n + Donor doping (cm−3) 2 × 1020
n channel doping (cm−3) 1 × 1020
Shape Rectangular
Transport 100
Nanomaterials 2023, 13, 959 4 of 20
Confinement and Y direction 010

Figure 1. Basic nanowire (a) The generic structure where gate length LG equals channel length Lch
Figure 1. Basic nanowire (a) The generic structure where gate length LG equals channel length Lch
(b) Schematic of the potential energy curve at drain bias VDS in the z-direction.
(b) Schematic of the potential energy curve at drain bias VDS in the z-direction.
The Landauer formula, as in reference [3], yields the following description of the
Table 1. Physical dimensions.
drain current:
∞
Device Parameters Proposed Nanowire
e
DS (upper)
ITox =  z z  FL
dE T ( E ) ×  f ( E , E ) − f ( E FR , E )  1 (1)
π h nvnm
,n 0
Tox (lower) nm 1
Channel Length (Lch ) 15 nm
Channel Width , Ech))= [1+ exp((E − EF ) / kBT )]−1 2.5 nm
f (E (W (2)
Channel HeightF(Hch ) 2.5 nm
Source Length (LS ) 10 nm
Drain length (LD ) E = En ,n ( zmax ) + Ez 10 nm (3)
v
S/D n + Donor doping (cm−3 ) 2 × 1020
n channel doping (cm ) − 3 1 × 1020
Shape Rectangular
Transport 100
Confinement and Y direction 010

The Landauer formula, as in reference [3], yields the following description of the
drain current:
Z∞
e
πh n∑
IDS = dEz T ( Ez ) × [ f ( EFL , E) − f ( EFR, E)] (1)
v ,n
0

f ( EF , E) = [1 + exp(( E − EF )/k B T )]−1 (2)

E = Env ,n (zmax ) + Ez (3)
where ‘n’ is the quantum number that matches the confinement in the wire cross-section;
EFL , and EFR (=EFL − eVDS ), where EFL and EFR are the fermi energy levels at the source
and drain. Equation (2) is the Fermi-Dirac Distribution. Env , n in Equation (3) fits into the
particular valley nv , where nv = 1, 2, 3, represents the energy confinement level at the top
barrier on the channel as (Etop = EFL + kB T). In Figure 1b, the potential energy distribution
is along the z-axis, where the maximum energy at the uppermost oxide interface from the
channel is represented as Emax . The Landauer equation can be simplified as follows:

1 + exp(( EFL − Env,n )/k B T )

 
ek B T
πh n∑
IDS = ln (4)
v ,n
1 + exp(( EFR − Env,n )/k B T )
 particular valley nv, where nv = 1, 2, 3, represents the energy confinement level at the top
barrier on the channel as (Etop =EFL + kBT). In Figure 1b, the potential energy distribution is
along the z-axis, where the maximum energy at the uppermost oxide interface from the
channel is represented as Emax. The Landauer equation can be simplified as follows:

ek BT  1 + exp(( E FL − E nv , n ) / k B T ) 
 ln  1 + exp(( E
Nanomaterials 2023, 13, 959 5 of 20
I DS =  (4)
πh nv , n  FR − E nv , n ) / k B T ) 

Equation
Equation(4)(4)determines
determines the the current
current IIDS where confinement energy levels at the oxide
DS where confinement energy levels at the oxide
interface top barrier z max exist. Multiple gates or very thin film structures were necessary
interface top barrier zmax exist. Multiple gates or very thin film structures were necessary
to
tocontrol
controlSCEs
SCEsinin III–V
III–V technologies,
technologies,asas suggested
suggestedpreviously
previously[3].
[3].There
Thereare
aresix
sixvariants
variants
shown
shown in Figure 2. Each variant differs in the number of gates and their arrangementwith
in Figure 2. Each variant differs in the number of gates and their arrangement with
natural
naturallength,
length,asasshown
shownin inTable
Table 2.
2.

Variantsof
Figure2.2.Variants
Figure ofnanowires
nanowires (a)
(a) DG
DG (b)
(b) GAA
GAA (c)
(c) OMEGA
OMEGA (d)
(d) PI
PI (e)
(e) TOP
TOP (f)
(f) TRI.
TRI.

2. Gate
TableThe variant
natural arrangement.
length λn can be calculated by:
Gate Variant No. of Gates ñ Natural Length λn
ε  ε t 
Double Gate λn =  si 1 + ox2 si  tsi tox 1.149 (5)
GAA nε ox  4ε4sitox  0.8129
Omega Gate 3.4 0.8817
where ‘ñ’ is Pi Gate to as the effective number
referred 3.14 of gates. The idea was to
0.9175
design devices
Top Gate
with both doped and undoped channels that1 use mid-gap gate material1.625
and yields the
Tri-gate 3 0.939
highest gate efficiencies for sub-10 nm technology [27,28]. The device’s short-channel be-
havior has been enhanced by raising the equivalent gate number ‘ñ’ and by maintaining
The
the size ofnatural
the gatelength n can be calculated
lengthλ(approximately) fiveby:
to ten times greater than that of the natural
length λn. s  
ε si ε ox tsi
λ n = 1 + t tox (5)
Table 2. Gate variant arrangement. n
eε ox 4ε si tox si
Gate Variant No. of Gates ñ Natural Length λn
where ‘ñ’ is referred to as the effective number of gates. The idea was to design devices with
Double
both doped andGate
undoped channels that use 2 mid-gap gate material and 1.149 yields the highest
GAA 4
gate efficiencies for sub-10 nm technology [27,28]. The device’s short-channel 0.8129behavior has
Omega Gate
been enhanced by raising the equivalent gate 3.4 number ‘ñ’ and by maintaining 0.8817the size of the
Pi (approximately)
gate length Gate 3.14 greater than that of the natural
five to ten times 0.9175length λn .
Top Gate
Scaling is possible with GAA devices1because they are built with the 1.625
gate in contact
with the Tri-gate
channel on all sides. The main benefit 3 of GAA devices is that they 0.939have a higher
ION /IOFF ratio. Owing to the asymmetric characteristics of the electrostatic control, the
tri-gate arrangement
Scaling is possibleresults in a lower
with GAA gate-controlled
devices because they arecharge
builtand is the
with 25%gate
smaller when
in contact
compared
with to the on
the channel GAA SiNWThe
all sides. for main
the specific
benefitW/H
of GAAratio because
devices therethey
is that are have
moreachannel
higher
Isides
ON/IOFFplaced
ratio. towards
Owing tothe thegate contact. characteristics of the electrostatic control, the tri-
asymmetric
When the gate voltage VG = 0 V, the potential in three-dimensional form has been
distributed out over the length of the NW, as shown in Figure 3, which is represented in
the order of Double gate, GAA, Omega, Pi, Top, and Tri-gate respectively. Three different
effective masses (ml , mt , mt ), (mt , ml , mt ), and (mt , mt , ml ) have been considered for the x,
y, and z directions. The (ml ) and (mt ) are the longitudinal and transverse effective masses
whose value is equal to 0.98 m0 and 0.19 m0. The mass (m0 ) is called free electron mass.
 The sub-bands in the z longitudinal direction are more energetic than those in the valley
pairs along x and y whose transport mass (mt) is less than longitudinal mass (ml). To sim-
ulate Δ valley for the electrons near the Z [001] direction for Si, the effective masses (mx,
my, and mz) obtained for valleys 1, 2, and 3 are (0.19, 0.19, and 0.98), (0.38, 0.38, and 1.17),
and (0.57, 0.57, and 1.36), respectively. Similarly, the effective masses (mx, my, and mz) for
Nanomaterials 2023, 13, 959 6 of 20
GaAs NW obtained for valleys 1, 2, and 3 are (0.067, 0.067, and 0.542), (0.134, 0.134, and
0.069), and (0.201, 0.201, and 0.676), respectively.

Figure 3.
Figure 3. 3D
3D potential
potentialprofile
profileofofthe
thesilicon nanowire
silicon forfor
nanowire allall
variants at VatG =V0 V= (a)
variants 0 VDG
(a)(b)
DGGAA (c)
(b) GAA
G
OMEGA (d) PI (e) TOP (f) TRI.
(c) OMEGA (d) PI (e) TOP (f) TRI.

4. Mathematical
The sub-bandsModeling
in the zoflongitudinal
the Nanowire with Laare
direction 2O3
more energetic than those in the
valley The proposed
pairs along xmulti-gate
and y whose device is a 3-D-dimensional
transport mass (mt ) is lessnanowire with a source
than longitudinal massand(ml ).
drain
To doping
simulate ∆ concentration
valley for the electrons 20 cm the
of 2 × 10 near 3 . The
Z source and drain
[001] direction forare made
Si, the of silicon
effective or
masses
gallium
(m x , my ,arsenide
and mz ) that has been
obtained highly 1,
for valleys doped
2, and with n+(0.19,
3 are atoms.0.19,
Theand
device’s
0.98),effective mass
(0.38, 0.38, and
Hamiltonian
1.17), has0.57,
and (0.57, beenand
denoted
1.36),by the notation:
respectively. Similarly, the effective masses (mx , my , and
mz ) for GaAs NW obtained for valleys 1, 2, and 3 are (0.067, 0.067, and 0.542), (0.134, 0.134,
and 0.069), and (0.201, 0.201,H3and
D
ψ (0.676), = Eψ (x, y, z)
x, y, z)respectively. (6)

4. Mathematical Modeling of the Nanowire with La2 O3

The proposed multi-gate device is a 3-D-dimensional nanowire with a source and
drain doping concentration of 2 × 1020 cm3 . The source and drain are made of silicon or
gallium arsenide that has been highly doped with n+ atoms. The device’s effective mass
Hamiltonian has been denoted by the notation:

H 3D ψ( x, y, z) = Eψ( x, y, z) (6)
!
h2 ∂2 h2 ∂ h2 ∂
 
1 ∂ 1 ∂
H 3D = − ∗ 2− − + V ( x, y, z) (7)
2m x ∂x 2 ∂y m∗y ∂y 2 ∂z m∗z ∂z

where the conduction band edge profile is represented by V (x, y, z) [26], and m∗ x , m∗ y , and
m∗ z are the effective masses:

V ( x, y, z) = EC1/2 ( x, y) − φ( x, y, z) (8)

where EC1/2 (x, y) is the band gap of the nanowire core material (Si/GaAs), the point (x,
y) links to the dioxide region, and (x, y, and z) corresponds to space potential. Due to the
movement of electrons in the z-direction, the effective core mass in the transport direction
Nanomaterials 2023, 13, 959 7 of 20

is m*z , and the effective oxide masses are represented by m*x = m*x (x,y) and m*y = m*y (x,y),
respectively.
The wavefunction of the three-dimensional Hamiltonian (x, y, and z) in the longitudi-
nal z direction is given as:

ψ( x, y, z) = ∑ ϕb (z)ψb (x, y; z) (9)

m

The bth mode eigen function ψb (x, y; z) represented in two-dimensional (2D) Schro
dinger equation is given as:

H 2D ψ( x, y; z) = Eb ( x )ψb ( x, y; z) (10)

where !
h2 ∂ h2 ∂
 
1 ∂ 1 ∂
H 2D = − − + V ( x, y; z) (11)
2 ∂x m∗x ( x, y) ∂y 2 ∂y m∗y ( x, y) ∂y
!
h2 ∂ h2 ∂
 
2D 1 ∂ 1 ∂
H = − ∗
− + V ( x, y; z) (12)
2 ∂x m x ( x ) ∂x 2 ∂y m∗y (y) ∂y

Under boundary conditions, the wave functions at the margins of the two-dimensional
(2D) cross-section plane is known as uncoupled mode space method, which eliminates the
coupling among several modes (or sub-bands), and ϕb (z) satisfies as follows:

h2 ∂2
 
− + Eb ( z ) ϕb (z) = Eϕb (z) (13)
2m∗z ∂z2

The Schrodinger Equation (13) with open boundary conditions describes the 1-D
transport problem, and further, the NEGF technique [31] has been used to solve it. The
primary notation for the sub-band b using 1-D Green’s function (Gb ) is as follows:
−1
Gb = [ E − Hb1D − ∑ S.b − ∑ D.b] (14)

where
h2 ∂2
H 1D = − + Eb (z) (15)
2m∗z ∂z2
ΣS,b and ΣD,b are the S/D self-energies of sub-band b, respectively. The 1D charge
density nk 1D (z) in the bth sub-band is then obtained via:

1
Z
n1D
b (x) = dE f s GΓS.b Gb† + f D Gk Γ D.b Gb† (16)
2π∆ x

where ∆x is the lattice spacing, and ΓS,b and ΓD,b are defined by:
 
†
ΓS.b = i ∑S.b − ∑S,b , (17)
 
†
Γ D.b = i ∑D.b − ∑D,b , (18)

The Fermi Distribution functions and the Fermi Energies at the source and drain are
given as follows:
1
f SD ( E) = (19)
( E− EFSD )/k B T
1+e
The 3-D quantum charge density has been employed in Poisson’s equation after
one-dimensional (1D) charge densities of each sub-band are resolve as follows:
2
n3D ( x, y, z) = ∑ n1D
b ( x )| ψb ( x, y; z | (20)
k
Nanomaterials 2023, 13, 959 8 of 20

q 
ND ( x, y, z) − n3D ( x, y, z)
∇2 φ( x, y, z) = − (21)
∈
Equation (21) determines the potential and doping profile (ND ) of (x, y, and z). The
current in Equation (22) is calculated using the Landauer–Buttiker formula, once self-
consistency and charge distributions are attained:

q
Z

h∑
ID = 2 dETb ( E)( f s ( E) − f D ( E)), (22)
b

where the Transmission Probability Tb (E) for sub-band ‘b’ is given by:

Tb ( E) = Tr (ΓS.b Gb Γ D.b Gb† ) (23)

4.1. Numerical Approaches

The two-dimensional (2-D) Schrodinger equations and the 1-D NEGF equation numer-
ical solutions have been presented. The mass discontinuity across the lanthanum oxide
(La2 O3 ) contact has been included in the 2-D Schrodinger equation using the following
k-space approach.

4.1.1. K-Space Solutions of Two-Dimensional Schrodinger Equations

Let’s first rewrite Equation (25) as follows:

ψ( x, y) = ψb ( x, y; z) (24)

where Ak ’s are expansion coefficients and |Ki is a basic set. The eigenvalue problem is
solved by substituting Equation (24) in Equation (10) and multiplying hL\ by the equation
sides:
∑ HLK2D
AK = Eb A L (25)
K

where HLK 2D A hL\H2D |Ki. In the standard k-space solution [19]:

K=

r s
2 2
|K i = sin(k i x ) sin(k j y) (26)
Lx Ly

Here, Lx and Ly are the cross-side section’s lengths in the x and y directions, respec-
tively.
πi
ki = (i = 1, . . . Nx ) (27)
Lx
πj
kj = ( j = 1, . . . Ny ) (28)
Ly
The corresponding grid numbers in the x and y directions are Nx and Ny . It must be
noted that the K index is derived with the indices i and j by the formula K = (Nx (i − 1) + j)
in Equation (26).
A rectangular cross-section with core/oxide interfaces at (x1 and x2 ) and (y1 and y2 ),
respectively (see Figure 4). Equation (29) defines the effective asymmetrical masses at the
core/oxide interfaces for the Hamiltonian using HLK 2D and it is given as follows:

2D 0 (x) (y)
HLK = HLK + HLK + HLK (29)

where

ZL x ZLy
!
0 4 h2 k2i h2 k2i
HLK = dxdy sin(k u x ) sin(k v y) + + V ( x, y) × sin(k i x ) sin(k j y) (30)
L x Ly 2m∗x ( x, y) 2m∗y ( x, y)
0 0
 Nanomaterials 2023, 13, 959 9 of 20

 Zy2
2h2 k i

(x) 1 1
HLK = (sin(k u x1 ) cos(k i x1 ) − (sin(k u x2 ) cos(k i x2 )) − ∗ × dy sin(k j y) sin(k v y) (31)
L x Ly m∗core,x mox
y1

Zx2
!
(y) 2h2 k j 1 1
HLK = (sin(k u y1 ) cos(k j y1 ) − (sin(k v y2 ) cos(k j y2 )) ∗
− ∗ × dx sin(k i x ) sin(k u x ) (32)
L x Ly mcore,y mox
x1

where m∗core,x
Nanomaterials 2023, 13, x FOR PEER REVIEW and m∗core,y
are the effective core masses in the x and y directions. The10u of
and21v
are the indices that are mapped with L in a parallel fashion to the index K, respectively.

Figure 4. Cross section of a nanowire.

Figure 4. Cross section of a nanowire.
4.1.2. Product Space Solutions of 2-D Schrodinger Equations
4.1.2. Product Space Solutions of 2-D Schrodinger Equations
From Figure 4, the rectangular cross-section with effective mass is represented as fol-
From Figure 4, the rectangular cross-section
lows:  ∗ with effective mass is represented as fol-
lows: ∗ m x ( x ) if y1 ≤ y ≤ y2
m x ( x, y) = (33)
m∗ox if y < y1 or y > y2
*
 m x ( x ) if y 1 ≤ y ≤ y 2
where m *x ( x , y ) =  *  (33)
m mif
∗ y <
core,x if y orxy≤>xy2 2
m∗x ( x ) =ox
x1 ≤
∗ (34)
mox if x < x1 or x > x2
where
For a good approximation, it is written as follows:
*
∗mcore
*
, x if x1 ≤ x ≤ x2
m ( x) =mx ( x,
x *
y) = m∗x ( x ) for 0 ≤ y ≤ Ly (35)
(34)
mox if x < x1 or x > x 2
The oxide region on either side of the core has been considered as (y1 < y < y2 ) and the
For a good
amplitude approximation,
in the top and bottom it is
of written
the oxideas regions
follows:are considered as (y < y1 and y > y2 ),
respectively. The band gap of the oxide materials is substantially wider than that of the
core material. Equationsm*x(36)
( x, and mx* (define
y) =(37) 0 ≤effective
x) forthe y ≤ Ly masses in the y direction: (35)
The oxide region on eitherm∗yside
( x, y)of = ∗
themcore has been considered as (y1 < y < y2) and the
y ( y ) for 0 ≤ x ≤ L x (36)
amplitude in the top and bottom of the oxide regions are considered as (y < y1 and y > y2),
where
respectively. The band gap of the oxide  ∗materials is substantially wider than that of the
core material. Equations (36)
my (and
∗
y) =
mcore,xthe
(37) define 1 ≤y≤y
if yeffective masses
2 in the y direction: (37)
m∗ox if y < y1 or y > y2
m*y ( x, y) = m*y ( y) for 0 ≤ x ≤ Lx (36)

where

*
mcore
*
, x if y1 ≤ y ≤ y2
m ( y) =  *
y (37)
mox if y < y1 or y > y2
Nanomaterials 2023, 13, 959 10 of 20

The adjacent side of the oxide effective mass regions is inappropriate for the above
reasons. Therefore, it is written as under good approximation:
!
h2 ∂ h2 ∂
 
2D 1 ∂ 1 ∂
H = − − + V ( x, y) (38)
2 ∂x m∗x ( x ) ∂x 2 ∂y m∗y (y) ∂y

The following equation is the product-space solution:

| K i = χ i ( x ) ξ j ( y ), (39)

Equation (40) determines the 1D Schrodinger equation in the x direction, where χi is

the ith Eigen function:

h2 ∂
   
1 ∂
− − V ( x ) χi ( x ) = ∈i χi ( x ) (40)
2 ∂x m∗x ( x ) ∂x

where
Zy2
1
V (x) = dz V ( x, y) (41)
y2 − y1
y1

ζ j (y) is the jth eigen function and V(x,y) is the confinement potential for the following
1-D Schrodinger equation in the y-direction:
( ! )
h2 ∂ 1 ∂
− − V ( y ) ξ i ( y ) = ∈i ζ j ( y ) (42)
2 ∂y m∗y (y) ∂y

where
Zy2
1
V (x) = dz V ( x, y) (43)
y2 − y1
y1

Substituting Equation (39) in Equation (38) and obtaining Equations (40)–(43).

∂ξ j (y)
2
  2
 
∂χi ( x )
H 2D K = −ξ j (y) h2 ∂ 1
m∗x ( x )
− χi ( x ) h2 ∂ 1
m∗y (y)
+ V ( x, y)χi ( x )ξ j (y)
∂x ∂x ∂y ∂y (44)
= (∈i + ∈ j +V ( x, y) − V ( x ) − V (y)) K

By multiplying hL\ in Equation (44), we obtain:

2D
HLK = ∈ L δLK + L\(V ( x, y) − V ( x ) − V (y)) K (45)

where
∈ L = ∈i + ∈ j (46)
After resolving the Schrodinger equations, the eigenvalue problem and the product-
space solution have been found. Equations (40) and (42) illustrate one dimensional (1-D)
version of the k-space solution approach, which was introduced in the previous section
and has been employed in the modeling.

5. Analysis of the GaAs-Based Nanowire

There are six possible structures that have been considered in the simulated NWs.
Two distinct materials (Si and GaAs), the Double Gate (DG), Gate-All-Around (GAA),
Omega, Pi, Top, and Tri-gate variants have been discussed. Figure 2 depicts the rectangular
structure with the physical dimensions of all six variants. The crystallographic orientation z
<001> direction has coincided with the channel transport direction. The design parameters
of the nanowire, listed in Table 1, have been considered for modeling. The conduction
band margins of the NWs for different dielectrics had been addressed previously [49].
 lar structure with the physical dimensions of all six variants. The crystallograph
tation z <001> direction has coincided with the channel transport direction. The
parameters of the nanowire, listed in Table 1, have been considered for model
Nanomaterials 2023, 13, 959 11 of 20
conduction band margins of the NWs for different dielectrics had been addresse
ously [49]. Higher gate dielectric constant materials have lower conduction ban
The SiO
Higher 2 has a greater conduction band edge than La2O3 when used as a gate d
gate dielectric constant materials have lower conduction band edges. The SiO2 has
aoxide.
greaterThus, lanthanum
conduction band edge oxide
than(La O33)when
La22O has been
used aschosen
a gate as a better
dielectric choice
oxide. Thus,for a gat
tric oxideoxide
lanthanum and it(Lais2 O
one of the
3 ) has beenbest reasons
chosen to provide
as a better conduction
choice for at lower
a gate dielectric oxideenergie
and
andit GaAs
is one of the best reasons
rectangular to provide
nanowire conductionhave
simulations at lower
beenenergies. The Si
designed and the sam
with
GaAs rectangular nanowire simulations have been designed with the same wire length
length (Lwire = 35 nm). Figure 5 shows the comparison between the first and last
(Lwire = 35 nm). Figure 5 shows the comparison between the first and last state energy. The
ergy.
first andThe first and
sixteenth energysixteenth energy levels
levels of conduction band of conduction
electrons band electrons
in a rectangular wire differin a rect
wire
by 22%differ by 22%
at the left atofthe
contact theleft contact
fermi of=the
level EFL −5 fermi
eV. level EFL = −5 eV.

Figure
Figure 5. 5.Conduction
Conduction bandband energy
energy using
using the the
NEGF NEGF (when
approach approach
VDS =(when VDS
0.6 V and Lch= =0.6
15 V and
nm). Lch = 1

Based on the full-band model (sp3 d5 s*) model, Figure 6 illustrates transmission co-
Based on the full-band model (sp3d5s*) model, Figure 6 illustrates transmissi
efficients for the 2.5 nm wire in the conduction band. The conduction band reaches high
ficients forwhen
transmission the 2.5 nm wireofin
the thickness thethe
wireconduction
get decreases,band. Thepreviously
as shown conduction band reach
[50,51].
transmission
The ballisticwhen
currentthe
hasthickness of theby
been calculated wire get decreases,
a comparison of the as shown previously
transmission and [
energy. The transmission steps depend on the channel and have high transmission regions
at an energy E = 2.6 eV. The energy differences are nearly parallel; the higher transmission
obtained for both Si and GaAs nanowires are 2.8892 eV and 3.5768 at 2.6 eV. Hence, the
GaAs is 1.23 times greater than Si NW. Maximum transmission can be achieved with
an increase in wire dimension. Higher transmission had been achieved using different
orientations with an increase in gate bias, as shown previously [20]. The transmission
spectrum has been fixed with zero gate bias (VG ) and a drain voltage (VD ) of 0.6 V. When
the gate voltage increases, higher transmission is achieved due to the lowering of the
barrier. To normalize the current density in ballistic conditions, the effective width (Weff ) is
assumed to be four times the channel width (Wch ), as shown previously [11,52].
Nanomaterials 2023, 13, x FOR PEER REVIEW
Nanomaterials 2023, 13, 959 12 of 20

Figure
Figure 6. 6. Comparisons
Comparisons of transmission
of transmission coefficients
coefficients between Si between Si and
and GaAs for GaAs
the 2.5 for the 2.5 nm
nm Trigate
Nanowire.
owire.
Figure 7 shows the normalized current density spectra (iz /iz , avg) calculated by
(T × (fThe
L − fRballistic
)), where Tcurrent has been
is the transmission andcalculated
(fL − fR ) are by a comparison
the left and right fermiof the tran
level
Nanomaterials 2023, 13, x FOR PEER REVIEW
contacts.
energy.TheThenormalized current steps
transmission densitydepend
distribution
onisthe
uniform in theand
channel GaAshave
NW, and14
highof
this21
transmi
uniformity occurs when the Wagner number (Wa > 5), as shown previously [53].
at an energy E = 2.6 eV. The energy differences are nearly parallel; the higher
obtained for both Si and GaAs nanowires are 2.8892 eV and 3.5768 at 2.6 eV
GaAs is 1.23 times greater than Si NW. Maximum transmission can be achi
increase in wire dimension. Higher transmission had been achieved using di
tations with an increase in gate bias, as shown previously [20]. The transmiss
has been fixed with zero gate bias (VG) and a drain voltage (VD) of 0.6 V. W
voltage increases, higher transmission is achieved due to the lowering of th
normalize the current density in ballistic conditions, the effective width (We
to be four times the channel width (Wch), as shown previously [11,52].
Figure 7 shows the normalized current density spectra (iz/iz, avg) calcu
(fL − fR)), where T is the transmission and (fL − fR) are the left and right fermi l
The normalized current density distribution is uniform in the GaAs NW,
formity occurs when the Wagner number (Wa > 5), as shown previously [53

Figure7.7.Comparisons
Figure Comparisonsof
ofnormalized
normalizedcurrent
currentdensity
densityfor
forthe
theSiSiand
andGaAs
GaAsTrigate
Trigatevariants.
variants.

The one-dimensional electron density (N1D) along the channel has been plotted
against Si and GaAs NW. The comparison has been noticed specifically at the midchannel
‘z’. The NEGF calculations are made to compute electron density and the electrostatic po-
tential at the interface. At zero gate bias, there is no creation of a potential barrier and
electrons to penetrate the channel. The electron density (N1D ~ 1 × 1020) cm−3 has been ob-
 Figure 7. Comparisons of normalized current density for the Si and GaAs Tri

The one-dimensional electron density (N1D) along the channe

Nanomaterials 2023, 13, 959 against Si and GaAs NW. The comparison has been noticed13specificall of 20

‘z’. The NEGF calculations are made to compute electron density and
tential at the interface. At zero gate bias, there is no creation of a p
The one-dimensional electron density (N1D ) along the channel has been plotted against
Sielectrons
and GaAs NW. to penetrate
The comparison the
haschannel. The
been noticed electron
specifically at thedensity
midchannel(N‘z’.1DThe
~ 1 × 102
NEGF
tainedcalculations
in the OFFare made to compute
state whenelectron
VG = 0density
V at theand source
the electrostatic potential at mor
and decreases
the interface. At zero gate bias, there is no creation of a potential barrier and electrons to
The electron
penetrate density
the channel. increases
The electron at1Dthe
density (N ~1× midchannel dueobtained
1020 ) cm−3 has been to three in thereasons
ness,
OFF state(b)
whenHigher
VG = 0 VGate
at the bias
sourcevoltage (VGmore
and decreases > 0.3 V),midchannel.
at the or (c) when channel do
The electron
density increases at the midchannel due to three reasons: (a) Surface Roughness, (b) Higher
source-drain doping. Here, the middle of the channel has a low elec
Gate bias voltage (VG > 0.3 V), or (c) when channel doping is greater than source-drain
whichHere,
doping. maintains
the middle aof higher
the channelconcentration at the drain.
has a low electron concentration, The
which electron de
maintains
a higher concentration at the drain. The electron densities are uniform throughout the
throughout the height of the channel, and the GaAs have attained a h
height of the channel, and the GaAs have attained a higher electron concentration at the
centration
midchannel thanatSithe
NW,midchannel
as shown in Figurethan
8. Si NW, as shown in Figure 8.

Figure
Figure8. Electron density (N
8. Electron 1D ) along(N
density Z-axis.
the1D) along the Z-axis.
The conduction band energy depends upon the function of both y and z, which is a
function of width and length, whereas the sub-band energy minima depends on length [33].
Using the relationship with the carrier velocity, it is concluded that the frequency of electron
transmission and channel length are inversely proportional with each other. Thus, the
saturation current (ION ) increases when the channel length gets reduced. The carriers can
travel more easily through shorter gate lengths and channel lengths in comparison to a
longer channel, as shown previously [45]. The conduction band edge profiles for the GAA
variant by fixing VD = 0.5 V and VG have been varied between 0 and 1, as shown in Figure 9.
The device gets off at low gate voltages. When gate voltage increases, the potential barrier
gets lower, and the energy attained by the electrons will move faster from source to drain
and gets lowered with an increase in drain bias.
The conduction band energy decreases at the midchannel when gate voltage VG
increases from 0 to 1 V. Each band energy differs with a voltage of 0.1 V. Due to a higher
impurity concentration than that of the channel, a sudden peak charge density (ρ) of
6.75 × 1026 Coul.m−3 was produced at the source and drain when VG = 0 V. With the
increase in gate voltage VG = 0.5 V, the charge density (ρ) over the length of the channel
 function of width and length, whereas the sub-band energy minima depends on
[33]. Using the relationship with the carrier velocity, it is concluded that the freque
electron transmission and channel length are inversely proportional with each
Thus, the saturation current (ION) increases when the channel length gets reduce
Nanomaterials 2023, 13, 959 carriers can travel more easily through shorter gate lengths and channel 14 of lengths
20 in
parison to a longer channel, as shown previously [45]. The conduction band edge p
for the GAA variant by fixing VD = 0.5 V and VG have been varied between 0 an
decreases from 6.48 × 1026 Coul.m−3 to 6.24 × 1026 Coul.m−3 from the source to the
shown in Figure 9. The device gets off at low gate voltages. When gate voltage inc
midchannel and then increases when it reaches near the drain terminal, as shown in
the potential
Figure barrierdensity
10. The charge gets lower, and
increases at the energy attained
the midchannel by the
when gate electrons
voltage will move
increases
from0 source
from to drain
to 1 V with voltageand gets lowered
difference of 0.1 V. with an increase in drain bias.

Nanomaterials 2023, 13, x FOR PEER REVIEW

Figure9.9.Conduction
Figure band
Conduction band energy
energy of the of
GAAtheNW
GAA NW
along thealong the
channel channel
when when
VD = 0.5 V. VD = 0.5 V.

The conduction band energy decreases at the midchannel when gate voltage
creases from 0 to 1 V. Each band energy differs with a voltage of 0.1 V. Due to a
impurity concentration than that of the channel, a sudden peak charge density (ρ)
× 1026 Coul.m−3 was produced at the source and drain when VG = 0 V. With the incr
gate voltage VG = 0.5 V, the charge density (ρ) over the length of the channel dec
from 6.48 × 1026 Coul.m−3 to 6.24 × 1026 Coul.m−3 from the source to the midchannel an
increases when it reaches near the drain terminal, as shown in Figure 10. The charg
sity increases at the midchannel when gate voltage increases from 0 to 1 V with v
difference of 0.1 V.

Figure
Figure The charge
10.10.The density
charge of the GAA
density nanowire
of the GAAalong the channel
nanowire when
along VDchannel
the = 0.5 V. when VD = 0.5

Table 3 shows the comparison modeling results of Si and GaAs Trigate

the geometrical dimensions are identical for Si and GaAs NW, the accumul
trons in GaAs NW is 11% more when it varies with gate voltage when co
Nanomaterials 2023, 13, 959 15 of 20

Table 3 shows the comparison modeling results of Si and GaAs Trigate NW. Though
the geometrical dimensions are identical for Si and GaAs NW, the accumulation of electrons
in GaAs NW is 11% more when it varies with gate voltage when compared to Si NW. Hence,
it is evident from the results that the increase in electron density of GaAs NW results in
a decrease in current density, which shows that the current density depends upon the
property and nature of the material and is independent of electron density. The inversion
charge shifts away from the interface in the charge on the quantum modulation effect on
Si and GaAs NW, which has been demonstrated to have no impact on the Subthreshold
Slope (SS), as shown previously [54]. The log-scale (ID -VGS ) transfer curve with VGS at
the subthreshold region has been used to calculate the Subthreshold Slope (SS), which is
defined as −[d(log10 ID )/dVG ].

Table 3. Comparison results of Si and GaAs Tri-gate NW.

Parameter Si NW GaAs NW
ID max(A) 6.08 ×10−8 8.20 × 10−8
Transmission 2.889 3.576
Normalized Current Density 1 1.06 × 10−6
Electron Density, N1D (#/cm3 ) 1.76 × 1020 1.98 × 1020
ION (A) 5.6 × 10−13 8.21 × 10−7
IOFF (A) 3.75 × 10−23 2.62 × 10−15
ION /IOFF 1.49 × 1010 3.13 × 108
Subthreshold Slope SS (mV/decade) at VG = 0.1 V to 0.6 V 8 mV to 107 mV 8 mV to 99 mV

The transfer characteristic curve for Si NW is shown in Figure 11. The GAA device
has the highest ON current of 4.09 × 10−5 A. Our simulation results of 15 nm Si NW are
compared with GaAs NW. Figure 12 shows the transfer characteristics curve for Si and
GaAs Tri-gate NW. Here, the ION current of the GaAs Tri-gate nanowire is 10−7 A (VS = 0 V
and VD = 0.6 V), and with the silicon nanowire is 10−13 A (VS = 0 V and VD = 0.6 V). This
analysis shows that the gallium arsenide nanowire, due to its larger ION current, has more
advantages over other types of devices. The results of ION , IOFF, and ION /IOFF are identical
values in the simulation results of the Double Gate and Omega variants, as shown clearly
in Table 4 comparison results.
It was observed previously [55] that when the gate length increases to 35 nm, the
ION /IOFF ratio remains high in GAA NWs compared to all other gates. The Omega and
the Double gate NWs modeling results remain the same and high when compared to the
Ω -gate MoS2 FET [56,57], which are shown in Table 4. The GAA has a smaller leakage
current than any other gate and a higher conduction band energy of 8% and 37% (at 15 nm)
than the Tri-gate and Pi gate. Thus, the GAA Si NW also shows a good Subthreshold Slope
(SS) of 176 mV/dec, which is 39% greater than the Tri-gate NW. Therefore, the GAA device
has been chosen as a better electrostatic control device.
Our proposed method has been compared with a previous study [11], where channel
length (Lch = 15 nm) and oxide thickness (tox = 1 nm) are the same. Hence, it has been
observed that the increase in the gate length (LG ≥ 15 nm) and silicon dioxide material
(SiO2 ) material leads to an increase in leakage current. Thus, in our proposed method, the
OFF current (IOFF ) has been reduced by various parameters, such as (1) oxide material with
high dielectric constant, (2) gate length scaling, and (3) low threshold voltage.
Table 5 summarizes the Si NW GAA variant with existing SiNW for an S/D doping
concentration of 2 × 1020 cm−3 . According to modeling results, the GAA NW has achieved
a high ION /IOFF ratio (1.06 × 109 ) when the width-to-height (W/H) ratio dimensions
are equal to 1. For fixed gate and drain voltages, the small (W/H) ratio changes in the
geometrical dimensions result in a low ION /IOFF ratio, a high electron density, Subthreshold
Slope (SS), and Drain Induced Barrier Lowering (DIBL). The smooth and rough surface in
the channel also differs between ON and OFF currents. However, the ION /IOFF ratios are
the same in both cases, the smooth surface produces a high ON current (1 × 10−6 A) as in
Ref. [32].
Nanomaterials 2023, 13, x FOR PEER REVIEW
Nanomaterials 2023, 13, 959 16 of 20

Figure
Figure 11.11.Transfer Characteristics
Transfer Characteristics of Si NW. of Si NW.

Figure
Figure 12.12. Transfer
Transfer characteristics
characteristics comparisoncomparison
for Si and GaAsfor Si and
NWs. GaAs NWs.

Table 4. Comparison results of Si NW variants.

Parameter DG GAA OMEGA PI TOP

ION
ON (A) 4.04 × 10−5
−5 4.09 × 10−5
−5 4.04 × 10−5
−5 3.46 × 10−5
−5 1.99 × 10−−
−14
−14 −14
−14 −14
−14 −13
−13 −−
Nanomaterials 2023, 13, 959 17 of 20

Table 4. Comparison results of Si NW variants.

Parameter DG GAA OMEGA PI TOP TRI

ION (A) 4.04 × 10−5 4.09 × 10−5 4.04 × 10−5 3.46 × 10−5 1.99 × 10−5 3.38 × 10−7
IOFF (A) 4.35 × 10−14 3.84 × 10−14 4.35 × 10−14 3.62 × 10−13 5.03 × 10−8 5.64 × 10−13
ION /IOFF 0.92 × 109 1.06 × 109 0.92 × 109 0.95 × 108 0.95 × 103 0.6 × 106
Conduction band Ec in eV at 15 nm −0.310 −0.309 −0.310 −0.330 −0.424 −0.334
Charge density ρ (Coul.m−3 ) at 15 nm 6.42 × 1023 6.24 × 1023 6.42 × 1023 1.01 × 1024 4.6 × 1024 1.01 × 1024
Normalized current density 1.03 × 1029 1.08 × 1029 1.03 × 1029 1.06 × 1029 1.07 × 1029 1.08 × 1029
Subthreshold slope (mV) 106 176 106 107 115 108
No. of nodes 9 9 10 10 10 9
No. of iterations 21 21 21 16 16 15
NEGF simulation time in secs 0.277 0.159 0.278 0.250 0.173 0.543
Schrodinger simulation time in secs 0.323 0.333 0.365 0.369 0.348 0.879
Poisson simulation time in secs 0.666 0.582 0.690 0.758 0.711 4.061

Table 5. Comparison of various Si nanowires with conventional nanowires.

Nanowire Oxide (nm) Lch (nm) LG (nm) VGS , VDS (V) ION (A) IOFF (A) ION /IOFF
[4] HFO2 5 20 (0.6, 0.6) 4.5 × 10−5 0.1 × 10−6 4500
[6] SiO2 15 5 (1, 0.5) 1 × 10−5 8.8 × 10−16 0.11 × 1011
[8] SiO2 11 10 (1, 1) 0.87 × 10−3 3.4 × 10−12 0.25 × 109
[12] SiO2 15 15 (0.3, 0.5) 1.22 × 10−5 3 × 10−11 0.40 × 106
[36] SiO2 6 6 (0.6, 0.001) 1 × 10−6 1 × 10−13 1 × 107
[49] La2 O3 10 10 (1, 0.5) 5.5 × 103 0.84 × 10−8 6.54 × 1011
[51] SiO2 5 6 (0.6, 0.6) 9.8 × 10−7 3.4 × 10−13 2.88 × 106
This work La2 O3 15 2 (1, 0.6) 4.09 × 10−5 3.84 × 10−14 1.06 × 109
This work La2 O3 15 2 (0.6, 0.6) 2.11 × 10−6 3.84 × 10−14 0.55 × 108

6. Conclusions and Future Recommendations

The electrical characteristics of the NWs shielded with Lanthanum Oxide (La2 O3 ) ma-
terial and the orientation with [001] z using the Non-Equilibrium Green Function (NEGF)
method were analyzed. Using the NEGF technique, the performance of Silicon and Gallium
Arsenide Nanowires with multi-gate structural arrangements, the electrical characteristics,
and their parameters are computed. The comparison between all the nanowire variants was
simulated. The semi-empirical tight-binding technique (sp3 d5 s*) was used to determine
the transmission coefficient of Silicon and Gallium Arsenide nanowires for [001] orienta-
tions. The transverse dimensions of rectangular nanowires with similar energy levels have
been examined, and the comparisons between Silicon and Gallium Arsenide NWs were
investigated. The III–V compound semiconductor, such as GaAs NW, shows an attractive
simulation in a few parameter results, such as transmission and electron density, compared
to Silicon NW. Considering the issue of leakage current reduction, Silicon NWs are more
suitable than Gallium Arsenide NWs.
In future work, the comparison between the same wire (Si or GaAs) with different
orientations and the same orientations for different materials (Si and GaAs) should be
investigated. The problems solved by Gallium Arsenide have focused on III–V compounds
along with Silicon or IV–IV compounds, which could be used for applications such as
energy storage, flexible electronics, and biomedical devices. Additionally, the develop-
ment of new synthesis techniques may lead to the production of nanowires with novel
compositions and improved properties.

Author Contributions: Conceptualization, P.P. and N.G.; methodology, P.P.; software, N.G.; valida-
tion, P.P., N.G. and V.M.S.; formal analysis, P.P.; investigation, P.P.; resources, N.G.; data curation, P.P.;
writing—original draft preparation, P.P.; writing—review and editing, N.G. and V.M.S.; visualization,
P.P. and N.G. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.
Nanomaterials 2023, 13, 959 18 of 20

References
1. Colinge, J.P. FinFETs, and Other Multi-Gate Transistors; Springer US: New York, NY, USA, 2008.
2. Goldberger, J.; Hochbaum, A.I.; Fan, R.; Yang, P. Silicon Vertically Integrated Nanowire Field Effect Transistors. Nano Lett. 2006, 6,
973–977.
3. Natori, K. Ballistic/quasi-ballistic transport in nanoscale transistor. Appl. Surf. Sci. 2008, 254, 6194–6198. [CrossRef]
4. Numata, T.; Uno, S.; Hattori, J.; Milnikov, G.; Kamakura, Y.; Mori, N.; Nakazato, K. A Self-Consistent Compact Model of Ballistic
Nanowire MOSFET with Rectangular Cross Section. IEEE Trans. Electron Devices 2013, 60, 856–862. [CrossRef]
5. Batakala, J.; Dhar, R.S. Effect of channel material on the performance parameters of GAA MOSFET. J. Nano Electron. Phys. 2022,
14, 02003. [CrossRef]
6. Hosseini, R.; Fathipour, M.; Faez, R. A comparative study of NEGF and DDMS models in the GAA silicon nanowire transistor.
Int. J. Electron. 2012, 99, 1299–1307. [CrossRef]
7. Wang, J.; Polizzi, E.; Lundstrom, M. A Three-dimensional Quantum Simulation of Silicon Nanowire Transistors with the
Effective-Mass Approximation. J. Appl. Phys. 2004, 96, 2192–2203. [CrossRef]
8. Samuel, A.; Arumugam, N.; Shenbagavalli, A. Drain current characteristics of silicon nanowire field effect transistor.
J. Microelectron. 2016, 2, 284–287. [CrossRef]
9. Coquand, R.; Barraud, S.; Cass, M.; Leroux, P.; Vizioz, C.; Comboroure, C.; Perreau, P.; Ernst, E.; Samson, M.-P.; Maffini-Alvaro, V.;
et al. Scaling of high-/metal-gate TriGate SOI nanowire transistors down to 10 nm width. Solid-State Electron. 2013, 88, 32–36.
[CrossRef]
10. Ramayya, E.B.; Vasileska, D.; Goodnick, S.M.; Knezevic, I. Electron transport in silicon nanowires: The role of acoustic phonon
confinement and surface roughness scattering. J. Appl. Phys. 2008, 104, 063711. [CrossRef]
11. Cresti, A.; Pala, M.G.; Poli, S.; Mouis, M.; Ghibaudo, G. A Comparative Study of Surface-Roughness-Induced Variability in Silicon
Nanowire and Double-Gate FETs. IEEE Trans. Electron Devices 2011, 58, 2274–2281. [CrossRef]
12. Jin, S.; Tang, T.; Fischetti, M.V. Anatomy of Carrier Backscattering in Silicon Nanowire Transistors. In Proceedings of the 2009
13th International Workshop on Computational Electronics, Beijing, China, 27–29 May 2009. [CrossRef]
13. Dargar, S.K.; Srivastava, V.M. Performance analysis of High-k dielectric-based silicon nanowire gate-all-around tunneling FET.
Int. J. Electr. Electron. Eng. Telecommun. 2019, 8, 340–345. [CrossRef]
14. Srivastava, V.M.; Singh, G. MOSFET Technologies for Double-Pole Four-Throw Radio-Frequency Switch; Analog Circuits and Signal
Processing; Springer International Publishing: Cham, Switzerland, 2014. [CrossRef]
15. Martinez, A.; Seoane, N.; Brown, A.R.; Barker, J.R.; Asenov, A. 3-D Non-equilibrium Green’s Function Simulation of Nonperturba-
tive Scattering from Discrete Dopants in the Source and Drain of a Silicon Nanowire Transistor. IEEE Trans. Nanotechnol. 2009, 8,
603–610. [CrossRef]
16. Dubrovskii, V.G.; Hijazi, H.; Isik Goktas, N.; LaPierre, R.R. Be, Te and Si Doping of GaAs Nanowires: Theory and Experiment.
J. Phys. Chem. C 2020, 124, 17299–17307. [CrossRef]
17. Asenov, A.; Wang, Y.; Cheng, B.; Wang, X.; Asenov, P.; Al-Ameri, T.; Georgiev, V.P. Nanowire transistor solutions for 5 nm and
beyond. In Proceedings of the 2016 17th International Symposium on Quality Electronic Design (ISQED), Santa Clara, CA, USA,
15–16 March 2016. [CrossRef]
18. Svizhenko, A.; Anantram, M.P.; Govindan, T.R.; Biegel, B.; Venugopal, R. Two-dimensional quantum mechanical modeling of
nanotransistors. J. Appl. Phys. 2002, 91, 2343–2354. [CrossRef]
19. Luisier, M.; Schenk, A.; Fichtner, W. Quantum transport in two- and three-dimensional nanoscale transistors: Coupled mode
effects in the non-equilibrium Green’s function formalism. J. Appl. Phys. 2006, 100, 043713. [CrossRef]
20. Luisier, M.; Schenk, A.; Fichtner, W.; Klimeck, G. Atomistic simulation of nanowires in the sp3d5s* tight-binding formalism: From
boundary conditions to strain calculations. Phys. Rev. B 2006, 74, 205323. [CrossRef]
21. Luisier, M.; Schenk, A.; Fichtner, W.; Klimeck, G. Transport calculation of Semiconductor Nanowires Coupled to Quantum Well
Reservoirs. J. Comput. Electron. 2006, 6, 199–202. [CrossRef]
22. Gueorguiev, G.K.; Stafström, S.; Hultman, L. Nano-wire formation by self-assembly of silicon–metal cage-like molecules. Chem.
Phys. Lett. 2008, 458, 170–174. [CrossRef]
23. Kakanakova-Georgieva, A.; Ivanov, I.G.; Suwannaharn, N.; Hsu, C.-W.; Cora, I.; Pécz, B.; Gueorguiev, G.K. MOCVD of AlN on
epitaxial graphene at extreme temperatures. CrystEngComm 2020, 23, 385–390. [CrossRef]
24. Kumarasinghe, C.; Bowler, D.R. DFT study of undoped and As-doped Si nanowires approaching the bulk limit. J. Phys. Condens.
Matter 2019, 32, 035304. [CrossRef]
25. García-Cristóbal, A.; Molina-Sánchez, A. Tight-Binding Simulations of Nanowires. In Encyclopedia of Nanotechnology; Springer:
Dordrecht, The Netherlands, 2015; pp. 1–12. [CrossRef]
26. Morioka, N.; Yoshioka, H.; Suda, J.; Kimoto, T. Quantum-confinement effect on holes in silicon nanowires: Relationship between
wave function and band structure. J. Appl. Phys. 2011, 109, 064318. [CrossRef]
27. Paramasivam, P.; Gowthaman, N.; Srivastava, V.M. Design and analysis of InP/InAs/AlGaAs based Cylindrical Surrounding
Double-Gate (CSDG) MOSFETs with La2 O3 for 5-nm technology. IEEE Access 2021, 9, 159566–159576. [CrossRef]
28. Gowthaman, N.; Srivastava, V.M. Mathematical modeling of drain current estimation in a CSDG MOSFET, based on La2 O3 oxide
layer with fabrication—A nanomaterial approach. Nanomaterials 2022, 12, 3374. [CrossRef]
Nanomaterials 2023, 13, 959 19 of 20

29. Neophytou, N.; Kosina, H. Hole mobility increase in ultra-narrow si channels under strong (110) surface confinement. Appl. Phys.
Lett. 2011, 99, 092110. [CrossRef]
30. Alam, K.; Sajjad, R.N. Electronic Properties and Orientation-Dependent Performance of InAs Nanowire Transistors. IEEE Trans.
Electron Devices 2010, 57, 2880–2885. [CrossRef]
31. Horiguchi, S. Validity of effective mass theory for energy levels in Si quantum wires. Phys. B Condens. Matter 1996, 227, 336–338.
[CrossRef]
32. Datta, S. Nanoscale device modeling: The Greens function method. Superlatt. Microstruct. 2000, 28, 253–278. [CrossRef]
33. Ren, Z.; Venugopal, R.; Goasguen, S.; Datta, S.; Lundstrom, M.S. nanoMOS 2.5: A two-dimensional simulator for quantum
transport in double-gate MOSFETs. IEEE Trans. Electron Devices 2003, 50, 1914–1925. [CrossRef]
34. Martinez, A.; Barker, J.R. Quantum Transport in a Silicon Nanowire FET Transistor: Hot Electrons and Local Power Dissipation.
Materials 2020, 13, 3326. [CrossRef]
35. Abramo, A.; Cardin, A.; Selmi, L.; Sangiorgi, E. Two-dimensional quantum mechanical simulation of charge distribution in silicon
MOSFETs. IEEE Trans. Electron Dev. 2000, 27, 1858–1863. [CrossRef]
36. Seoane, N.; Martinez, A.; Brown, A.R.; Asenov, A. Study of surface roughness in extremely small Si nanowire MOSFETs using
fully-3D NEGFs. In Proceedings of the 2009 Spanish Conference on Electron Devices, Santiago de Compostela, Spain, 11–13
February 2009. [CrossRef]
37. Al Mamun Mazumdar, A.; Hosen, K.; Islam, M.S.; Park, J. Numerical investigations of nanowire gate-all-around negative
capacitance GaAs/InN tunnel FET. Digit. Object Identifier 2022, 10, 30323–30334. [CrossRef]
38. Montazeri, M.; Fickenscher, M.; Smith, L.M.; Jackson, H.E.; Yarrison-Rice, J.; Kang, J.H.; Gao, Q.; Tan, H.H.; Jagadish, C.; Guo,
Y.; et al. Direct Measure of Strain and Electronic Structure in GaAs/GaP Core−Shell Nanowires. Nano Lett. 2010, 10, 880–886.
[CrossRef] [PubMed]
39. Ren, Z.; Lundstrom, M. Simulation of nanoscale MOSFETs: A scattering theory interpretation. Superlattices Microstruct. 2000, 27,
177–189. [CrossRef]
40. Naveh, Y.; Likharev, K.K. Shrinking limits of silicon MOSFETs: Numerical study of 10 nm scale devices. Superlattices Microstruct.
2000, 27, 111–123. [CrossRef]
41. Cheng, H.; Yang, Z.; Zhang, C.; Xie, C.; Liu, T.; Wang, J.; Zhang, Z. A New Approach to Modeling Ultrashort Channel Ballistic
Nanowire GAA MOSFETs. Nanomaterials 2022, 12, 3401. [CrossRef] [PubMed]
42. Pikus, F.G.; Likharev, K.K. Nanoscale field-effect transistors: An ultimate size analysis. Appl. Phys. Lett. 1997, 71, 3661–3663.
[CrossRef]
43. Ren, Z.; Venugopal, R.; Datta, S.; Lundstrom, M.; Jovanovic, D.; Fossum, J. International Electron Devices Meeting, 2000, IEDM
Technical Digest. In Proceedings of the International Electron Devices Meeting. IEDM Technical Digest, Washington, DC, USA,
10 December 1997; p. 715.
44. Natori, K. Scaling limit of the MOS transistor—A ballistic MOSFET. IEICE Trans. Electron. 2001, 84, 1029–1036.
45. Rahman, A.; Jing Guo Datta, S.; Lundstrom, M.S. Theory of ballistic nanotransistors. IEEE Trans. Electron Devices 2003, 50,
1853–1864. [CrossRef]
46. Naveh, Y.; Likharev, K.K. Modeling of 10-nm-Scale ballistic MOSFETs. IEEE Electron Device Lett. 2000, 21, 242–244. [CrossRef]
47. Ren, Z.; Venugopal, R.; Datta, S.; Lundstrom, M.S.; Jovanovic, D.; Fossum, J.G. The ballistic nanotransistor: A simulation study. In
Proceedings of the International Electron Devices Meeting 2000. Technical Digest. IEDM (Cat. No.00CH37138), San Francisco,
CA, USA, 10–13 December 2000; pp. 715–718.
48. Ren, Z. Nanoscale MOSFETs: Physics, Simulation, and Design. Ph.D. Thesis, Purdue Univ., West Lafayette, IN, USA, 2001.
49. Mech, B.C.; Kumar, J. Effect of high-k dielectric on the performance of Si, InAs, and CNT FET. Micro Nano Lett. 2017, 12, 624–629.
[CrossRef]
50. Ren, Z.; Venugopal, R.; Datta, S.; Lundstrom, M.S. Examination of design and manufacturing issues in a 10 nm double gate
MOSFET using non-equilibrium green’s function simulation. In Proceedings of the International Electron Devices Meeting.
Technical Digest (Cat. No.01CH37224), Washington, DC, USA, 2–5 December 2001; pp. 5.4.1–5.4.4.
51. Khan, T.; Iztihad, H.M.; Sufian, A.; Alam MN, K.; Mollah, M.N.; Islam, M.R. Gate length scaling of Si nanowire FET: A NEGF
study. In Proceedings of the 2015 International Conference on Electrical Engineering and Information Communication Technology
(ICEEICT), Savar, Bangladesh, 21–23 May 2015. [CrossRef]
52. Gowthaman, N.; Srivastava, V.M. Mathematical modeling of electron density arrangement in CSDG MOSFET: A nano-material
approach. J. Mater. Sci. 2022, 57, 8381–8392. [CrossRef]
53. Moorthy, V.M.; Srivastava, V.M. Device modeling and optimization of nanomaterial-based planar heterojunction solar cell (by
varying the device dimensions and material parameters). Nanomaterials 2022, 12, 3031. [CrossRef] [PubMed]
54. Hiblot, G.; Rafhay, Q.; Boeuf, F.; Ghibaudo, G. Compact modeling of subthreshold swing in the double gate and nanowire
MOSFETs, for Si and GaAs channel materials. Solid-State Electron. 2015, 111, 188–195. [CrossRef]
55. Al-Ameri, T.; Wang, Y.; Georgiev, V.P.; Adamu-Lema, F.; Wang, X.; Asenov, A. Correlation between gate length, geometry, and
electrostatically driven performance in ultra-scaled silicon nanowire transistors. In Proceedings of the 2015 IEEE Nanotechnology
Materials and Devices Conference (NMDC), Anchorage, AK, USA, 13–16 September 2015. [CrossRef]
Nanomaterials 2023, 13, 959 20 of 20

56. Zhao, D.-H.; Tian, Z.-L.; Liu, H.; Gu, Z.-H.; Zhu, H.; Chen, L.; Zhang, D.W. Realizing Omega-Shaped Gate MoS2 Field-Effect
Transistor Based on SiO2 /MoS2 Core-Shell Heterostructure. ACS Appl. Mater. Interfaces 2020, 12, 14308–14314. [CrossRef]
57. Natori, K. Ballistic metal-oxide-semiconductor field effect transistor. J. Appl. Phys. 1994, 76, 4879–4890. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.