Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
by
G. Venkateshwar Reddy
Technology, Delhi, for the award of the degree of Master of Science (Research) in
Electrical Engineering Department is a bona fide work carried out by him under my
supervision and guidance. The research reports and the results presented in this thesis
have not been submitted in parts or in full to any other University or Institute for the
iii
ACKNOWLEDGEMENTS
for his invaluable guidance and advice during every stage of this endeavour. I am greatly
indebted to him for his continuing encouragement and support without which, it would
not have been possible for me to complete this undertaking successfully. His insightful
Special thanks are due to Prof. D. Nagchoudhuri for his valuable suggestions and
My sincere thanks and acknowledgements are due to all my family members who
G. Venkateshwar Reddy
v
ABSTRACT
vii
clearly suggest the superiority of the proposed structure over the conventional DG SOI
MOSFET.
Another widely used approach to alleviate the short channel performance is
through channel engineering. Asymmetric single halo MOSFET structures have been
introduced for bulk as well as for SOI MOSFETs to adjust the threshold voltage and
improve the device SCE. Simulation studies have been done on channel engineering
concept when applied to a DG structure. Our results reveal that the channel engineering
concept indeed provides an improved performance over the conventional DG structure.
viii
TABLE OF CONTENTS
CHAPTER I.................................................................................................................................................1
INTRODUCTION...........................................................................................................................................1
1.1. MOTIVATION FOR PRESENT RESEARCH ........................................................................................1
1.2. NATURE OF THE PROBLEM ............................................................................................................5
1.3. RECENT RESEARCH RELEVANT TO THE PROBLEM .........................................................................5
1.4. RESEARCH PROBLEM STATEMENT ................................................................................................7
1.5. THESIS ORGANIZATION..................................................................................................................7
CHAPTER II ...............................................................................................................................................9
ANALYTICAL MODELING OF SURFACE POTENTIAL AND THRESHOLD VOLTAGE IN A
PARTIALLY DEPLETED (PD) DMG-SOI MOSFET ..................................................................................9
2.1. INTRODUCTION..............................................................................................................................9
2.2. DMG SOI STRUCTURE AND ITS PARAMETERS ............................................................................10
2.3. MODEL FORMULATION........................................................................................................11
2.3.1 Two-Dimensional Potential Analysis .........................................................................................11
2.3.2 Threshold Voltage Model ...........................................................................................................19
2.4. RESULTS AND DISCUSSION ..........................................................................................................20
2.4.1 Surface Potential and Electric Field ..........................................................................................20
2.4.2 Minimum Surface Potential and Threshold Voltage...................................................................22
2.4.3 Substrate Bias dependence .........................................................................................................24
2.4.4 Gate Material Engineering.........................................................................................................27
2.5. SUMMARY ...................................................................................................................................28
CHAPTER III............................................................................................................................................29
A NEW DUAL MATERIAL DOUBLE GATE NANOSCALE SOI MOSFET − TWO DIMENSIONAL
ANALYTICAL MODELING AND SIMULATION....................................................................................29
3.1. INTRODUCTION............................................................................................................................29
3.2. DMDG SOI STRUCTURE AND ITS PARAMETERS .........................................................................31
3.3. MODEL FORMULATION .............................................................................................................32
3.3.1 Surface Potential Model .............................................................................................................32
3.3.2 Threshold Voltage Model ...........................................................................................................38
3.3.3 IV Model.....................................................................................................................................41
a) The impact ionization and parasitic BJT effects.........................................................................42
b) The channel length modulation, velocity overshoot and DIBL effects .......................................45
c) Total drain current .....................................................................................................................46
ix
3.4. RESULTS AND DISCUSSION ..........................................................................................................47
3.4.1 Surface Potential and Electric Field ..........................................................................................48
3.4.2 Threshold Voltage and Drain Induced Barrier Lowering ..........................................................52
3.4.3 IV Characteristics.......................................................................................................................54
3.5. SUMMARY ...................................................................................................................................57
CHAPTER IV ............................................................................................................................................59
INVESTIGATION OF THE NOVEL ATTRIBUTES OF A SINGLE HALO DOUBLE GATE SOI
MOSFET: 2D SIMULATION STUDY ........................................................................................................59
4.1. INTRODUCTION............................................................................................................................59
4.2. DG-SH STRUCTURE AND ITS PARAMETERS .................................................................................61
4.3. RESULTS AND DISCUSSION ..........................................................................................................61
4.3.1 Surface Potential ........................................................................................................................62
4.3.2 Threshold Voltage and DIBL......................................................................................63
4.3.3 Subthreshold Slope and On/Off Currents...................................................................................66
4.3.4 IV Characteristics.......................................................................................................................68
4.4. SUMMARY ...................................................................................................................................71
CHAPTER V..............................................................................................................................................73
CONCLUSIONS ...........................................................................................................................................73
APPENDICES...............................................................................................................................................77
REFERENCES..............................................................................................................................................85
LIST OF PUBLICATIONS...................................................................................................................................89
x
LIST OF TABLES
Table Page
47
Table 3.1 Device parameters used in the model and the simulation of the
DMDG and the DG SOI MOSFETs.
61
Table 4.1 Device parameters used for the simulation of the DG-SH and the DG
SOI MOSFETs.
xi
LIST OF ILLUSTRATIONS
Figure Page
2.3 Electric field along the channel towards the drain end obtained from 22
the analytical model and MEDICI simulation in DMG-SOI and SMG-
SOI MOSFET’s with a channel length L = 0.15µm and a drain bias
VDS = 1.75V.
2.5 Threshold voltage variation with channel length for DMG SOI device 24
compared between MEDICI simulations and model prediction with L1
constant at 50µm.
2.6 Threshold voltage variation versus channel length at a fixed back gate 25
bias of φB = 1V with L1 constant at 50µm.
2.7(a) Threshold voltage variation versus back gate bias φB with a channel 26
length L=0.2µm (L1=0.05µm, L2=0.15µm).
2.7(b) Threshold voltage variation versus back gate bias φB with a channel 26
length L=0.2µm (L1 = 0.05µm, L2 = 0.05µm).
2.8 Threshold voltage variation with gate work function difference with 27
φM2 fixed at 4.17eV for the DMG-SOI MOSFET with channel length
L (=L1+L2) of 0.25µm.
xiii
3.1 Cross-sectional view of (a) DG-SOI MOSFET (b) DMDG-SOI 31
MOSFET.
3.3 Surface potential profiles at front gate and back gate for DMDG and 50
DG-SOI MOSFETs for a channel length L = 0.1 µm (L1 = L2 = 0.05
µm).
3.4 Back gate surface potential profiles for different film thicknesses for 51
DMDG SOI MOSFETs for a channel length L = 0.1 µm (L1 = L2 =
0.05 µm).
3.5 Electric-field variation at the drain end along the channel at the Si- 51
SiO2 interface of DMDG and DG SOI MOSFETs for a channel length
L = 0.1µm (L1 =L2 = 0.05µm).
3.9 Variation of gm with different channel lengths, (L1 = L2) for DMDG 55
and DG SOI MOSFETs.
3.10 Variation of gd with different channel lengths, (L1 = L2) for DMDG 56
and DG SOI MOSFETs.
3.11 Variation of voltage gain with different channel lengths, (L1 = L2) for 56
DMDG and DG SOI MOSFETs.
xiv
with a film thickness of 20nm.
4.7(a) Variation of Ioff and Ion with channel length for DG-SH and DG SOI 67
MOSFET for a film thickness of 20nm.
4.7(b) Ratio of Ion and Ioff with channel length for DG-SH and DG SOI 67
MOSFET for a film thickness of 20nm at VDS= 0.75V.
4.9 Variation of gm, gd with different channel lengths for DG-SH and DG 70
SOI MOSFETs.
4.10 Variation of voltage gain with different channel lengths DG-SH and 70
DG SOI MOSFETs.
xv
CHAPTER I
INTRODUCTION
surface layer (less than 0.5 µm of thickness) and are isolated from the silicon body with a
depletion layer of a P-N junction. The leakage current of this P-N junction exponentially
increases with temperature, and is responsible for several serious reliability problems.
Excessive leakage currents and high power dissipation limits operation of the
neighboring insulating tubs can cause latch-up failures and significantly degrade circuit
performance.
components and in conjunction with the lateral isolation, reduces various parasitic circuit
capacitances and thus, eliminates the possibility of latch-up failures. SOI technology
offers superior CMOS devices with higher speed, high density, and reduced second order
effects as compared to the bulk silicon technology for low-voltage and low-power VLSI
circuit applications.
Depending on the thickness of the silicon layer, MOSFETs will operate in fully
depleted (FD) or partially depleted (PD) regimes. When the channel depletion region
extends through the entire thickness of the silicon layer, the transistor operates in a FD
mode. PD transistors are built on relatively thick silicon layers with the depletion depths
1
100 – 200 nm Undepleted Depletion layer
Gate oxide thick Si Si layer ~ 50 nm extending into
n+ poly p+ poly Gate oxide thick Si handle wafer
n+ p n+ p+ n p+ n+ poly p+ poly
n+ p n+ p+ n p+
100 – 400 nm 100 – 400 nm
Fig. 1.1: Cross-sectional view of the PD (left) and FD (right) SOI CMOS devices [1].
of the fully powered MOS channel shallower than the thickness of the silicon layer. Fig.
The FD devices have several advantages compared to the PD devices: free from
kink effect, enhanced subthreshold swing, highest gains in circuit speed, reduced power
requirements and highest level of soft-error immunity. Several drawbacks of the FD SOI
design and process come along with their benefits: Although FD MOSFETS are naturally
free from the kink effect, the interface coupling effect affects their operation [2-3]. The
interface coupling is inherent to fully depleted SOI devices, where all parameters
insidiously affected by the opposite gate voltage (at the buried oxide). The threshold
voltage fluctuation due to SOI thickness variation is one of the most serious problems in
FD SOI MOSFETs. In comparison, partially depleted SOI devices [4-5] are built on a
thicker silicon layer and are simpler to manufacture. Most design features for developing
PD devices can be imported from the bulk silicon devices and used in the SOI
environment with only modest changes. This makes circuit redesign for the PD devices
Generally, CMOS device design has been optimized for digital applications even
2
for aggressively scaled channel length devices. However the same rules may produce a
poor analog performance due to short channel effects (SCE) [6]. Thus it becomes
necessary to optimize the existing CMOS logic technologies, so that they are compatible
with the conventional CMOS process, and at the same time lead to improved
performance in mixed-mode systems. During the past decade, excellent high-speed and
performance have been achieved through improved design, use of high quality material
and shrinking device dimensions [7-8]. However, with the reduction of channel length,
of the technology. The predominating short-channel effects are a lack of pinch-off and a
shift in threshold voltage with decreasing channel length as well as drain induced barrier
lowering (DIBL) and hot-carrier effect at increasing drain voltage. In contrast to the bulk
device, front gate of the SOI device has better control over its active device region in the
thin-film and hence charge sharing effects from source/drain regions are reduced.
However, the thin-film thickness has to reduce to the order of 10nm to significantly
and causes large device external resistance due to shallow source/drain extension (SDE)
depths. MOSFET device design has been engineered by different approaches for the
channel engineering and gate work function engineering have been implemented for the
alleviation of these disadvantages. Gate engineering and channel engineering are the two
3
enhancement and suppression of short-channel effects due to the introduction of a step
function in the channel potential. In a DMG-MOSFET, the work function of metal gate 1
(M1) is greater than metal gate 2 (M2) i.e., φM1 > φM2 for an n-channel MOSFET and
vice-versa for a p-channel MOSFET. For an n-channel DMG-MOSFET, VT1 > VT2, which
has the inherent advantage of improving the gate transport efficiency by modifying the
electric field pattern and the surface potential profile along the channel. The step potential
profile, due to different work functions of two metal gates, ensures reduction in the short-
channel effects and screening of the channel region under M1 from drain potential
variations. Beyond saturation, M2 absorbs any additional drain-source (D/S) voltage and
hence the M1 region is screened from the drain potential variations. This work is
therefore used to study for the first time the potential benefits offered by the DMG gate in
modeling and numerical simulation. The model provides an efficient tool for further
Double Gate (DG) MOSFETs using lightly doped ultra thin layers seem to be a
very promising option for ultimate scaling of CMOS technology [11]. Excellent short-
channel effect immunity, high transconductance and ideal subthreshold factor have been
reported by many theoretical and experimental studies on this device. To incorporate the
MOSFET with the exception that the front gate of the DMDG structure consists of two
simulation and with the analytical model, the reduced short channel effects exhibited by
4
DMDG structure below 100nm, while simultaneously achieving a higher
The proposed structure exhibits the desired features of both DMG and DG structures.
is through channel engineering. Asymmetric single halo MOSFET structures have been
introduced for bulk as well as for SOI MOSFETs to adjust the threshold voltage and
improve the device SCE. Simulation studies have been done on channel engineering
The present work involves three distinct aspects, viz. (a) Two-dimensional
modeling of surface potential and threshold voltage of a PD SOI MOSFET with DMG,
(b) Investigation of the novel features of the proposed DMDG SOI MOSFET along with
a complete analytical model and (c) Numerical simulation studies using MEDICI [12] to
investigate the novel features offered by the DG single halo (DG-SH) doped SOI
MOSFET.
The concept of a Dual Material Gate is similar to that of a Split-Gate Field Effect
Transistor (SG FET) proposed by Shur [13]. However, SG FET suffers from the fringing
capacitance between the metal gates which increases as the separation between the metal
gates reduce. In 1999, Long et. al. [9] proposed a new gate structure called the dual
enhancement and suppressed SCEs. Zhou [14] suggested a way in which the Hetero-
5
Material Gate (HMG) MOSFET can be fabricated by inserting one additional mask in the
bulk CMOS processing technology and demonstrated the novel characteristics of this
new type of MOSFET by simulation studies. However, with PD SOI rapidly emerging as
the technology for next-generation VLSI, the effect of DMG in submicron MOS
model for surface potential and threshold voltage to aid in understanding the potential
DG MOSFETs employ the same type of gate material for both the gates (either p+ poly
or n+ poly) while asymmetrical DG MOSFETs employ both, p+ poly for the front gate
and n+ poly for the back gate. It would seem asymmetrical gates would undermine the
current drive because the resulting device has only one predominant channel. However, it
has been shown in [15] that gate–gate coupling in the asymmetrical DG MOSFET is
the former device for more reasons than just the threshold-voltage control. For the first
material gate concept is applied. This has been complemented by an analytical model,
which includes surface potential modeling, threshold voltage modeling and the drain
current modeling.
As pointed out previously, short channel performance can also be improved with
the help of channel engineering, extensive studies have been done on the local high
doping concentration in the channel near source/drain junctions via lateral channel
6
engineering, e.g., halo [16] or pocket implants [17]. Single halo MOSFET structures
have been introduced for bulk [18] as well as for SOI MOSFETs [19] to adjust the
threshold voltage and improve the device SCE. Halo implantation devices show excellent
output characteristics with low DIBL, no kink, higher drive currents, flatter saturation
MOSFET. However, no such attempt has been reported on DG MOSFET. For the first
time we have investigated the performance of DG structure with halo implantation using
The work accomplished in this dissertation has been carried out in terms of the
i) Physics based 2-D analytical model for the surface potential distribution and
ii) A novel design, the Dual Material Double Gate (DMDG) MOSFET is
simulation results.
compare the benefits of the DG single halo (DG-SH) doped structure over a
The dissertation is divided into five chapters and its outline is described as given
7
below:
Chapter I: Introduction.
Some fundamental concepts related to SOI devices, emerging ideas: advantages &
A physics based 2-D model for the surface potential variation along the channel in the
addition, threshold voltage model of the DMG-PD MOSFET is also developed and is
A New Dual Material Double Gate (DMDG) MOSFET is proposed along with the 2-
D model for the surface potential. An analytical model for the threshold voltage and
This chapter presents the novel features offered by the asymmetrical double gate
single halo (DG-SH) doped structure to enhance the MOSFET performance through
Chapter V: Conclusions.
8
CHAPTER II
ANALYTICAL MODELING OF SURFACE POTENTIAL AND THRESHOLD
VOLTAGE IN A PARTIALLY DEPLETED (PD) DMG-SOI MOSFET
2.1 Introduction
At very short gate lengths, the CMOS device operation is asymmetrical even at a
very small drain bias due to the higher drain side electric field resulting in short-channel
reduce the drain side electric field and its consequent impact upon the channel. Dual-
engineering” with different workfunctions to introduce a potential step in the channel [9].
This leads to a suppression of short-channel effects and an enhanced source side electric
field resulting in increased carrier transport efficiency in the channel region. And with its
unique structure, DMG offers flexibility in choosing thin-film thickness, channel doping,
buried oxide thickness and permittivity in short channel SOI MOSFET design.
Furthermore, the DMG structure may also be employed in symmetric structures, i.e.,
adding a layer of material with different workfunction to both sides of the gate (like a
LDD spacer).
Till now, two general approaches have been used to model the surface potential
profile, the electric field pattern and their impact on the threshold voltage. One of these
approaches is the two-dimensional (2-D) numerical simulation [20]. The other approach
is to develop an analytical solution, using either the charge sharing approach [21–22] or
solving the Poisson’s equation in the depletion region [23-25]. One-dimensional analysis,
9
short channels and is suitable only for a long channel transistor where the “edge” effects
along the sides of the channel can be neglected. In such an analysis, it is assumed that
electric field lines are perpendicular or have a component along the y-direction only. If
the channel is short (i.e., L is not much larger than the sum of the source and drain
depletion widths), a significant part of the electric field will have components along both
the y and x directions, the latter being the direction along the channel’s length. Thus a
two-dimensional analysis is needed. So far, no analytical model has been reported for
explaining the impact of partially depleted Dual Material Gate (DMG) SOI MOSFET
structure over parameters such as carrier transport efficiency, hot electron injection and
based analysis of the partially depleted Dual Material Gate (DMG) silicon-on-insulator
MOSFET. The expressions for the surface potential and electric field under the two metal
gates are derived. The model results are verified by comparing them to simulated results
obtained from the 2-D device simulator MEDICI. The characteristics of the DMG-PD
SOI MOSFETs are examined and compared with those of corresponding single material
gate partially depleted (SMG-PD) SOI MOSFETs. The model is simple in its functional
implemented using the 2-D device simulator MEDICI [12] is shown in Fig. 2.1 with gate
metals M1 and M2 of lengths L1 and L2, respectively. The source/drain (S/D) regions are
10
Gate
L1 L2
Source Drain
M1 M2
tox
wd
n+ tsi n+
Buried oxide tb
x
p substrate
y
Substrate
Fig. 2.1: Cross-sectional view of an n-channel partially depleted DMG SOI MOSFET.
uniform at 1×1018cm-3. Typical values of the front-gate oxide thickness, the buried-oxide
thickness and the body-film thickness are 2nm, 450nm and 100nm respectively. In the
Assuming that the impurity density in the channel region is uniform and
neglecting the influence of charge carriers on the electrostatics of the channel, the
potential distribution in the silicon thin-film, before the onset of strong inversion can be
written as
11
∂ 2φ ( x , y ) ∂ 2φ ( x , y ) qN A
+ = for 0 ≤ x ≤ L, 0 ≤ y ≤ wd (2.1)
∂x 2 ∂y 2 ε si
where NA is the body doping concentration , εsi is the silicon dielectric constant, L is the
2ε si (2φ F + φ B )
wd = (2.2)
qN A
where φB is the body electrostatic potential and φF = VT ln(NA/ni) is the Fermi potential
The potential profile in the vertical direction is assumed to be a third order polynomial
[26] i.e.,
φ ( x, y ) = φ s ( x ) + a1 ( x ) y + a2 ( x ) y 2 + a3 ( x ) y 3 (2.3)
where φs(x) is the surface potential and a1, a2 and a3 are arbitrary constants which are
functions of x only. In the DMG structure, we have two different materials with different
work functions φM1 and φM2. Therefore the flat-band voltages of the two gates would be
Eg
φ Si = χ Si + + φF (2.5)
2q
where Eg is the silicon bandgap and χsi is the electron affinity. Since we have two regions
in the DMG structure, the potential under the metal gates M1 and M2 can be written as
12
The Poisson’s equation is solved separately under the two gate regions using the
φ1 ( x, wd ) = φ2 ( x, wd ) = φ B (2.9)
∂φ1 ( x, y ) ∂φ2 ( x, y )
x = L1 = x = L1 (2.10)
∂x ∂x
4. Electric flux at the interface of the gate/oxide is continuous for both the metal gates
∂φ2 ( x, y ) ε ox φs 2 ( x) − VGS' 2
y =0 = (2.12)
∂y ε si tox
where εox is the dielectric constant of the oxide, tox is the gate oxide thickness and
where VGS is the gate-to-source bias voltage, VFB1 and VFB2 are the front-channel
∂φ1 ( x, y ) ∂φ2 ( x, y )
= =0 at y = wd (2.14)
∂y ∂y
13
6. Potential at the source end is
N N
Vbi = VT ln A 2 D (2.16)
ni
φ 2 ( L,0) = φ s 2 ( L ) = V bi + V DS (2.17)
The expression for the constants a11 ( x) , a12 ( x) , a13 ( x) , a21 ( x) , a22 ( x) and a23 ( x) can
From (2.6), (2.11) and (2.14) we can obtain the following relations for the region under
metal 1:
ε ox φs1 ( x) − VGS' 1
a11 ( x) = (2.19)
ε si tox
Similarly for the region under metal 2, we obtain the following expressions using (2.7),
14
ε ox φs 2 ( x) − VGS' 2
a21 ( x) = (2.22)
ε si tox
2ε ox φ s1 ( x) − VGS' 1 3
a12 ( x) = − − 2 (φ s1 ( x) − φ B ) (2.24)
ε si wd tox wd
2 ε
3 ( s1
a13 ( x) = φ ( x) − VGS' 1 ) + ox 2 (φ s1 ( x) − VG' S1 ) (2.25)
wd ε si wd
Thus substituting the values of a11 ( x ) , a12 ( x ) and a13 ( x ) in (2.6) and using φ1 ( x, y ) in
∂ 2φs1 ( x)
α + βφs1 ( x) = γ 1 (2.26)
∂x 2
3 y 2 2 y 3 ε ox y 2ε ox y 2 ε ox y 3
where α =1− 2 + 3 + − +
wd wd ε si t ox ε si t ox wd ε si t ox wd2
12 y 6 6ε ox y 4ε ox
β= − + −
wd3 wd2 ε si t ox wd2 ε si t ox wd
qN A 12 y 6 6ε ox y 4ε ox '
γ1 = + 3 − 2 φ B + − VGS 1
ε si wd wd ε si t ox wd ε si t ox wd
2
The above equation is a simple second order non-homogenous differential equation with
γ1
φ s1 ( x ) = A1 exp(ηx ) + B1 exp( −ηx ) + for 0 ≤ x ≤ L1 (2.27)
β
where A1, B1 are constants and η = − β / α . Now using the boundary condition (2.15),
15
we obtain
γ1
A1 + B1 + = Vbi (2.28)
β
2ε ox φs 2 ( x) − VGS' 2 3
a22 ( x) = − − 2 (φ s 2 ( x ) − φ B ) (2.29)
ε si wd tox wd
2 ε
3 ( s2
a23 ( x) = φ ( x) − VGS' 2 ) + ox 2 (φs 2 ( x) − VG' S 2 ) (2.30)
wd ε si wd
Thus substituting the values of a21 ( x ) , a22 ( x ) and a23 ( x ) in (2.7) and using φ2 ( x, y ) in
∂ 2φs 2 ( x )
α + βφs 2 ( x ) = γ 2 (2.31)
∂x 2
qN A 12 y 6 6ε ox y 4ε ox '
γ2 = + 3 − 2 φ B + − VGS 2
ε si wd wd ε si t ox wd ε si t ox wd
2
The above equation is a simple second order non-homogenous differential equation with
γ2
φ s 2 ( x) = A2 exp(ηx) + B2 exp( −ηx ) + for L1 ≤ x ≤ L (2.32)
β
where A2, B2 are constants and η = − β / α . Now using the boundary condition (2.16),
we obtain
γ2
A2 exp(η L) + B2 exp(−η L) + = Vbi + VDS (2.33)
β
16
Using boundary conditions (2.8) and (2.10), we get the following expressions
γ1 γ
A1 exp(η L1 ) + B1 exp(−η L1 ) + = A2 exp(η L1 ) + B2 exp(−η L1 ) + 2 = φ L1 (2.34)
β β
Solving (2.28), (2.33) and (2.34) for A1, B1, A2 and B2, we obtain
− (Vbi − σ 1 ) + (φ L1 − σ 1 )exp(ηL1 )
A1 =
exp( 2ηL1 ) − 1
B1 =
(Vbi − σ 1 )exp(2ηL1 ) − (φL1 − σ 1 )exp(ηL1 )
exp(2ηL1 ) − 1
A2 =
(Vbi + VDS − σ 2 )exp(ηL) − (φL1 − σ 2 )exp(ηL1 )
exp(2ηL) − exp(2ηL1 )
where σ1=γ1/β and σ2=γ2/β . Since the electrostatic potential and the electric field are
continuous at the interface of the region below the two gates as indicated by the boundary
condition 1 in (2.8), the value of φL1 can be obtained using (2.35) as:
The concept of drain induced barrier lowering (DIBL) can be illustrated by the
channel surface potential. Since the sub-threshold leakage current often occurs at the
behavior of the device can be monitored by the minimum surface potential. DIBL can be
demonstrated by plotting the surface potential minima as a function of the position along
the channel for different drain bias conditions. Due to the co-existence of the two
17
dissimilar gate metals, M1 and M2, having a finite workfunction difference, the position
of minimum surface potential, xmin, will be solely determined by the gate metal with
higher work function. For the NMOSFET (present case), the work function of M1 is
greater than that of M2 and for the PMOSFET, it is vice-versa. The minimum potential of
∂φS 1 ( x )
=0
∂x x = x
min
1 B1
xmin = ln (2.37)
2η A1
The minimum surface potential can be obtained by substituting for x = xmin in (2.27)
which is given as
γ1
φs1, min = A1 exp(ηxmin ) + B1 exp(−ηxmin ) + (2.38)
β
An expression for the electric field can be obtained by differentiating the surface
∂φ1 ( x, y )
E1 ( x ) = y =0 = A1η exp(η x) − B1η exp(−η x ) 0 ≤ x ≤ L1 (2.39)
∂x
∂φ2 ( x, y )
E2 ( x ) = y =0 = A2η exp(η x ) − B2η exp(−η x) L1 ≤ x ≤ L (2.40)
∂x
The above two equations are quite useful in determining how the drain side electric field
18
2.3.2 Threshold Voltage Model
The threshold voltage Vth is that value of the gate voltage VGS at which a
conducting channel is induced at the surface of the SOI MOSFET. Therefore, the
threshold voltage is taken to be that value of gate source voltage for which φS ,min = 2φF ,
where φF is the difference between the extrinsic Fermi level in the bulk region and the
intrinsic Fermi level. Hence we can determine the value of threshold voltage as the value
γ1
φs1, min = A1 exp(ηxmin ) + B1 exp(−ηxmin ) +
β
A1 = K1 + K 2γ 1 and B1 = K 3 + K 4γ 1 (2.41)
−V + φ L1 exp(η L1 )
where K1 = bi
exp(2η L1 ) − 1
1 1 − exp(η L1 )
K2 =
β exp(2η L1 ) − 1
2φ − K exp(η x ) − K exp(−η x ) qN ε t w
Vth = VFB1 − F 1 min 3 min
− A si ox d (2.42)
1 + K exp(η x ) + K exp(−η x ) ε si 4ε ox
β 2 min 4 min
19
The dependence of the surface potential on the work functions and the length of the metal
gates mean that we can tune the surface potential and the threshold voltage of the
transistor. This mechanism has also been observed for the bulk MOSFETs [14] when the
The threshold voltage model derived does not take into account the presence of
mobile carriers in the channel. Hence the model cannot clearly demarcate the transition
between the weak and the strong inversion. However, such an analysis will lead to a
To verify the proposed analytical model, the 2-D device simulator MEDICI was
used to simulate the different aspects, viz. surface potential, electric field, threshold
voltage etc. and compare with the results predicted by the analytical model.
Fig. 2.2 shows the calculated and the simulated values of the surface potential for
different drain voltages along the channel for a channel length L = 0.15µm. It is clearly
seen that because of the step function profile of the surface potential, there is no
significant increase of the potential under gate M1 as the drain voltage is increased,
which means that gate M1 is effectively screened from the drain potential variations. In
other words, the drain voltage has very little influence on the drain current after
saturation, thus reducing the drain conductance. It follows from the figure that the
minimum surface potential point is independent of the applied drain bias. Because of this
feature, DIBL is reduced considerably in the DMG structure. The predicted values of the
20
3.0 VGS = 0.15V MEDICI V DS = 1.75V
NA = 1×1018cm-3 MODEL
Surface Potential (in volts)
2.5 L1= 0.05µm
L2 = 0.1µm V = 0.95V
DS
2.0
φM1 = 4.63V
0.5
Fig. 2.3 shows the calculated and simulated values of electric field along the
channel length at the drain end for the DMG-PD SOI MOSFET and the simulated values
for the SMG-PD SOI MOSFET for the same channel length. It is clearly seen that there is
a considerable reduction in the peak electric field at the drain end in the case of the DMG
structure when compared with the SMG. This reduction in the electric field reduces the
hot carrier effect, which is another important short channel effect. The agreement
between the model and the simulated results proves the accuracy of the model.
21
600
DMG-SOI SMG-SOI
550 φM1 = 4.63V φM = 4.63V
Electric Field (in kV/cm)
500 φM2 = 4.17V L = 0.15µm
L1 = 0.05µm
450
L2 = 0.1µm
400
350
300
DMG-(MEDICI)
250 SMG-(MEDICI)
DMG-(MODEL)
200
0.120 0.125 0.130 0.135 0.140 0.145 0.150
Position in channel (in µm)
Fig. 2.3: Electric field along the channel towards the drain end obtained from the
analytical model and MEDICI simulation in DMG-SOI and SMG-SOI MOSFETs with a
channel length L = 0.15µm and a drain bias VDS = 1.75V.
partially depleted DMG-SOI with film thickness tSi = 100 nm is shown Fig. 2.4. As
previously stated, Fig. 2.4 points out that the minimum channel potential is almost
constant for different channel lengths i.e., the shift in the minimum surface potential with
channel length is almost zero. This is due to the existence of a work function difference
in the case of the DMG-SOI MOSFETs. The close match between the analytical results
and the 2-D simulation results verifies the validity of the model for the minimum surface
22
0.7
Minimum Surface Potential (in volts)
0.6
0.5
0.4
0.3
φM1 = 4.63V
0.2 φM2 = 4.17V MEDICI
L1 = 0.05µm MODEL
0.1
0.10 0.15 0.20 0.25 0.30 0.35 0.40
Channel length (in µm)
Fig. 2.4: Variation of the channel minimum potential with channel length L (=L1+L2) for
partially depleted DMG-SOI MOSFETs with L1 constant at 0.05µm.
In Fig. 2.5, the simulated values of the threshold voltage as a function of channel
length are compared with those of the model values. It is seen that the threshold voltage
rolls-up with decreasing channel lengths for a fixed L1. This happens due to the increased
L1/L2 ratio at decreasing channel lengths since the portion of the larger work function gate
is increased as the channel length reduces. This unique feature of the DMG structure is an
added advantage when the device dimensions are continuously shrinking. From the
results it is clearly seen that the calculated values of the analytical model tracks the
23
0.7
0.5
0.4
0.3
φM1 = 4.63V
0.2 φM2 = 4.17V MEDICI
MODEL
L1 = 0.05µm
0.1
0.10 0.15 0.20 0.25 0.30 0.35 0.40
Channel length (in µm)
Fig. 2.5: Threshold voltage variation with channel length for DMG SOI device compared
between MEDICI simulations and model prediction with L1 fixed at 50nm.
Fig. 2.6 shows the threshold voltage variation against different channel lengths at a
fixed back gate bias φB. In the case of the DMG-SOI MOSFET, the threshold voltage
variation is quite less when compared with the SMG-SOI MOSFET. This figure clearly
depicts that the threshold voltage of the DMG SOI structure is much less dependent on
the substrate bias than the SMG SOI structure as the threshold voltage roll-up mechanism
24
0.9
Threshold voltage (in volts) 0.8
0.7
0.6
0.5
φ B = -1V
0.4 L = 0.05µm DMG-PD(MEDICI)
1
In Fig. 2.7(a) and Fig. 2.7(b), the threshold voltage for both the DMG and the
SMG is shown for different substrate biases for two different channel lengths. In Fig.
2.7(a) when the channel length L = 0.2µm, the threshold voltage variation in the DMG is
smaller than that of the SMG. However, in Fig. 2.7(b), when the channel length is
reduced to L = 0.1µm, the variation of the threshold voltage between the two is same. The
threshold voltage saturates as the negative substrate bias is increased for both the DMG
and the SMG. This condition is achieved for the low back gate bias when the channel
length is reduced. The threshold voltage saturates because of the hole accumulation at the
bottom of the silicon film. For the SMG-SOI MOSFET, the shift in the position of the
minimum surface potential towards the source increases with increasing values of φB and,
therefore, the DIBL increases. However, for the DMG structure, the shift in the position
25
of the minimum surface potential is almost zero and hence DIBL increase is far less in
0.9
Threshold voltage (in volts)
0.8
0.7
0.6
0.5
L = 0.05µm
0.4 1
L = 0.15µm DMG-PD(MEDICI)
0.3 2
φM1 = 4.63V DMG-PD(MODEL)
0.2 φ = 4.17V SMG-PD
M2
0.1
0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0
Substrate bias (in volts)
Fig. 2.7(a): Threshold voltage variation versus back gate bias φB with a channel length
L=0.2µm (L1=0.05µm, L2=0.15µm).
0.8
Threshold Voltage (in volts)
0.7
0.6
0.5
0.4
L1 = 0.05µm
0.3 DMG-PD(MEDICI)
L2 = 0.05µm
0.2 φM1 = 4.63V DMG-PD(MODEL)
φM2 = 4.17V SMG-PD(MEDICI)
0.1
0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0
Substrate Bias (in volts)
Fig. 2.7(b): Threshold voltage variation versus back gate bias φB with a channel length
L=0.2µm (L1 = 0.05µm, L2 = 0.05µm).
26
2.4.4 Gate Material Engineering
The dual-material gate (DMG) structure offers the benefit of SCE suppression in a
SOI device by virtue of gate material engineering, i.e., engineering the length and
workfunction of the two gate metals. Fig. 2.8 shows the variation of the threshold
voltage with workfunction difference at a fixed channel length of L = 0.25 µm for two
L1/L2 ratios as predicted by the analytical expression and the 2-D numerical simulations.
As shown in the figure, the threshold voltage increases with the increasing workfunction
difference. For a fixed workfunction difference, the threshold voltage is higher for a
higher L1/L2 ratio due to the increased proportion of the channel region controlled by a
1.0
Threshold Voltage (in volts)
0.5
0.4
MEDICI
0.3 MODEL
0.2
0.2 0.3 0.4 0.5 0.6 0.7 0.8
φM1- φM2(in volts)
Fig. 2.8: Threshold voltage variation with gate work function difference with φM2 fixed at
4.17eV for the DMG-SOI MOSFET with channel length L (=L1+L2) of 0.25µm.
27
2.3 Summary
structure has been examined for the first time by developing a 2-D analytical model. The
results obtained have been compared with the MEDICI simulations. The model results
agree well with the simulated results. It is also emphasized that the DMG structure in the
partially depleted SOI MOSFETs leads to a reduced short channel effect as the surface
potential profile shows a step at the interface of the two metals which the reduces drain
conductance and the DIBL. Also, the peak electric field at the drain end is reduced
minimizing the hot carrier effect. The threshold voltage shows a roll-up with the reducing
channel lengths. Thus the short channel behavior of the PD SOI MOSFETs is further
enhanced with the introduction of the DMG structure over their single gate counterparts.
Moreover, the immunity against the SCE is possible by a new way of “gate material
28
CHAPTER III
A NEW DUAL MATERIAL DOUBLE GATE NANOSCALE SOI MOSFET − TWO
DIMENSIONAL ANALYTICAL MODELING AND SIMULATION
3.1 Introduction
Double Gate (DG) MOS devices using lightly doped ultra thin silicon layers are
very promising for ultimate scaling of the CMOS technology because they exhibit the
best performance in terms of the short channel effects, the subthreshold slope, the current
drivability and the transconductance [27]. One of the problems affecting the DG
MOSFETs is the control of the threshold voltage that is hardly dependent on the doping
concentration and on the other hand, is strongly affected by the thickness of the silicon
layer. In particular, the when technological characteristics compatible with a 50nm gate
length are assumed, Vth is negative if two n+ poly gates are adopted, while it becomes too
large (approximately 1V) in the case of the p+ poly gates. Acceptable values for Vth (0.2 –
0.4V for VDD = 1.0 – 1.5V) are obtained when asymmetrical gate structure with one n+
poly and one p+ poly is realized [15]. Furthermore, device simulations demonstrate that
the asymmetric structure provides almost the same drain ON current as the symmetric
one [15]. For these reasons, it might be concluded that the asymmetric DG structure is
preferable because symmetric devices would require alternative gate materials with a
fabrication process.
However, the DG MOSFETs also suffer from considerable short channel behavior
in the sub 100nm regime. As demonstrated in the previous chapter, to enhance the
immunity against the short channel effects, the dual material gate (DMG) structure was
29
proposed. As the name suggests, the gate is composed of two materials in this structure
potential profile, along with an increase in the transconductance while suppressing the
short channel effects. However, the drive capability of this structure is not as good as that
To incorporate the advantages of both the DG and the DMG SOI MOSFETs we
propose a new structure called the dual material double gate (DMDG) SOI MOSFET.
The proposed structure is similar to that of an asymmetrical DG SOI MOSFET with the
exception that the front gate of the DMDG structure consists of two materials (p+ poly
and n+ poly). We present here, using two-dimensional simulation, the reduced short
channel effects exhibited by the DMDG structure below 100nm, while simultaneously
SOI MOSFET. The proposed structure exhibits the desired features of both the DMG and
peak electric field near the drain end, increased drain breakdown voltage and
transconductance, reduced drain conductance and threshold voltage “roll-up” even for
channel lengths far below 100 nm. An analytical model using Poisson’s equation also has
been presented for the surface potential leading to the threshold voltage model for the
DMDG SOI MOSFET. A complete drain current model [28] considering the impact
ionization [29], the velocity overshoot, the channel length modulation and the DIBL [30]
is also presented. The accuracy of the model is verified by comparing the model results
30
3.2 DMDG SOI structure and its parameters
Schematic cross-sectional views of both the asymmetrical DG and the DMDG SOI
MOSFET implemented using the 2-D device simulator MEDICI are shown in Fig. 3.1.
The front gate consists of dual materials M1 (p+ poly) and M2 (n+ poly) of lengths L1 and
L2 respectively, while the back gate is effectively an n+ poly gate. The doping in the p
type body and the n+ source/drain regions is kept at 1 x 1015cm-3 and 5 x 1019cm-3
thickness and the thin-film thickness are 2nm, 2nm and 12 nm respectively.
S D
p+
tf
L tsi
n+ n+
p
tb
n+
(a)
G
L1 L2
S D
p+ n+
tf
L tsi
n+ p n+
tb
n+
(b)
31
3.3 Model Formulation
Assuming that the impurity density in the channel region is uniform and the
influence of the charge carriers on the electrostatics of the channel can be neglected, the
potential distribution in the silicon thin-film, before the onset of strong inversion can be
expressed as
∂ 2φ ( x , y ) ∂ 2φ ( x , y ) qN A
+ = for 0 ≤ x ≤ L, 0 ≤ y ≤ t si (3.1)
∂x 2 ∂y 2 ε si
where NA is the uniform film doping concentration independent of the gate length, εsi is
the dielectric constant of silicon, tsi is the film thickness and L is the device channel
length. The potential profile in the vertical direction, i.e., the y-dependence of φ ( x, y ) can
be approximated by a simple parabolic function as proposed by [26] for the fully depleted
SOI MOSFETs as
φ ( x, y ) = φs ( x) + a1 ( x) y + a2 ( x) y 2 (3.2)
where φS ( x ) is the surface potential and the arbitrary coefficients a1 ( x ) and a2 ( x ) are
functions of x only.
In a DG-SOI MOSFET, the front gate consists of only one material i.e, p+ poly, but in the
DMDG structure, we have two different materials (p+ poly and n+ poly) with work
the p+ poly and n+ poly at the front gate would be different and they are given as
32
Eg
φSi = χ Si + + φF (3.4)
2q
where Eg is the silicon bandgap at 300K, χsi is the electron affinity of silicon,
carrier concentration. Since we have two regions in the front gate of the DMDG structure,
based on (3.2) the surface potential under p+ poly and n+ poly can be written as:
where φs1 and φs2 are the surface potentials under p+ poly (M1) and n+ poly (M2)
respectively and a11, a12, a21 and a22 are arbitrary coefficients.
The Poisson’s equation is solved separately under the two top front gate materials
1. Electric flux at the front gate-oxide interface is continuous for the dual material gate.
Therefore, we have
ε ox φs1 ( x) − VGS , f 1
'
∂φ1 ( x, y )
y =0 = under M1 (3.7)
∂y ε si tf
ε ox φs 2 ( x) − VGS , f 2
'
∂φ2 ( x, y )
y =0 = under M2 (3.8)
∂y ε si tf
where εox is the dielectric constant of the oxide, tf is the gate oxide thickness and
where VGS is the gate-to-source bias voltage, VFB,fp and VFB,fn are the front-channel
(3.3).
33
2. Electric flux at the back gate-oxide and the back channel interface is continuous for
both the materials of the front gate (p+ poly and n+ poly).
ε ox VGS ,b − φB ( x)
'
∂φ1 ( x, y )
y =tsi = under M1 (3.10)
∂y ε si tb
ε ox VGS ,b − φB ( x)
'
∂φ2 ( x, y )
y =tsi = under M2 (3.11)
∂y ε si tb
where tb is the back gate oxide thickness, φB ( x) is the potential function along the
back gate oxide-silicon interface, and VGS' ,b = VGS − VFB ,bn , where VFB ,bn is the back
3. Surface potential at the interface of the two dissimilar gate materials of the front gate
is continuous
4. Electric flux at the interface of the two materials of the front gate at y = 0 is
continuous
∂φ1 ( x, y ) ∂φ2 ( x, y )
x = L1 = x = L1 (3.13)
∂x ∂x
N AND
where Vbi = VT ln 2
is the built-in potential across the body-source junction
n
i
34
where VDS is the applied drain-source bias.
The constants a11 ( x ) , a12 ( x ) , a21 ( x ) and a22 ( x ) in equations (3.5) and (3.6) can be
From (3.5), (3.7) and (3.10), we can obtain the following relations for the region under p+
poly (L1):
ε ox φ S1 ( x ) − VGS , f 1 φ S1 ( x ) − VGS' , f 1
'
ε
a11 ( x ) = = Cf where C f = ox (3.17)
ε Si tf ε Si tf
ε ox VGS .b − φ B ( x )
'
V ' −φ ( x) ε ox
a11 ( x ) + 2a12 ( x ) tSi = = Cb GS ,b B where Cb = (3.18)
ε Si tb tb tb
Similarly for the region under n+ poly (L2), we obtain the following expressions using
ε ox φS 2 ( x ) − VGS , f 2 φ S 2 ( x ) − VGS' , f 2
'
ε
a21 ( x ) = = Cf where C f = ox (3.20)
ε Si tf ε Si tf
ε ox VGS ,b − φ B ( x ) V ' −φ ( x)
'
ε ox
a21 ( x ) + 2a22 ( x ) tSi = = Cb GS ,b B where Cb = (3.21)
ε Si tb tb tb
C C C C
VGS' ,b + VGS' , f 1 f + f − φS 1 ( x ) 1 + f + f
a12 ( x ) = Cb CSi Cb CSi
C
tSi2 1 + 2 Si
Cb
35
Thus substituting the values of a11 ( x ) and a12 ( x ) in (3.5) and using φ1 ( x, y ) in (3.1) we
∂ 2φS 1 ( x )
− αφS 1 ( x ) = β1 (3.22)
∂x 2
where
2 (1 + C f Cb + C f CSi )
α= and
tSi2 (1 + 2 CSi Cb )
qN A C C + C f CSi 1
β1 = − 2VGS' , f 1 2 f b − 2VGS ,b 2
'
ε Si tSi (1 + 2 CSi Cb ) tSi (1 + 2 CSi Cb )
β1
φS 1 ( x ) = A1 exp (η x ) + B1 exp ( −η x ) − (3.23)
α
where A1, B1 are constants and η = α . Now using the boundary condition (3.14), we
obtain
β1
A1 + B1 − = Vbi (3.24)
α
C C C C
VGS' ,b + VGS' , f 2 f + f − φ S 2 ( x ) 1 + f + f
a22 ( x ) = Cb CSi Cb CSi
C
tSi2 1 + 2 Si
Cb
Thus substituting the values of a21 ( x ) and a22 ( x ) in (3.6) and using φ2 ( x, y ) in (3.1), we
36
∂ 2φS 2 ( x )
− αφS 2 ( x ) = β 2 (3.25)
∂x 2
qN A C C + C f CSi 1
β2 = − 2VGS' , f 2 2 f b − 2VGS ,b 2
'
ε Si t Si (1 + 2 CSi Cb ) t Si (1 + 2 CSi Cb )
β2
φS 2 ( x ) = A2 exp (η ( x − L1 ) ) + B2 exp ( −η ( x − L1 ) ) − (3.26)
α
where A2, B2 are constants and η = α . Now using boundary condition (3.15), we obtain
β2
Vbi + VDS = A2 exp (η L2 ) + B2 exp ( −η L2 ) − (3.27)
α
Using boundary conditions (3.12) and (3.13), we get the following expressions
where
qN A C f Cb + C f CSi 1
σ 1 = − β1 α = − VGS' , f 1 − VGS ,b
'
(3.30)
ε Si 1+ C C + C C 1 + C C + C C
f b f Si f b f Si
and
qN A C f Cb + C f CSi 1
σ 2 = − β2 α = − VGS' , f 2 − VGS .b
'
(3.31)
ε Si 1+ C C + C C
f b f Si 1 + C f Cb + C f CSi
Solving (3.28) and (3.29), we obtain the relationship among the coefficients A1, B1, A2
and B2 as
37
A2 = A1 exp (η L1 ) +
(σ 1 − σ 2 ) and
2
B2 = B2 exp ( −η L1 ) +
(σ 1 − σ 2 )
2
A2 = A1 exp (η L1 ) +
(σ 1 − σ 2 ) and B2 = B1 exp ( −η L1 ) +
(σ 1 − σ 2 )
2 2
The electric field pattern along the channel determines the electron transport
velocity through the channel. The electric field component in the x–direction, under p+
poly is given as
∂φ1 ( x, y ) ∂φS 1 ( x )
E1 ( x ) = = = A1η exp (η x ) − B1η exp ( −η x ) (3.32)
∂x y =0
∂x
Similarly the electric field pattern, in x–direction, under n+ poly (front gate) is given as
∂φ2 ( x, y ) ∂φS 2 ( x )
E2 ( x ) = = = A2η exp ( λ1 ( x − L1 ) ) − B2η exp ( −η ( x − L1 ) ) (3.33)
∂x y =0
∂x
The above two equations are quite useful in determining how the drain side electric field
In the proposed DMDG SOI MOSFET, we have tf and tb as the front gate and
back gate oxide thicknesses, and the same gate voltage, VG, is applied to both the gates.
The channel doping is uniform with an acceptor concentration of 1015 cm-3 as in [28]. The
38
threshold voltage, Vth for the DMDG SOI structure is derived from the graphical
approach as has been done for the DG SOI MOSFETs [28]. When the potential
distribution dependence on the gate voltage is studied, it is seen that first an inversion
layer is formed on the inside surface of the back gate n+ polysilicon. Then the potential
distribution changes linearly while the surface potential is fixed. After this, an inversion
layer on the inside surface of the p+ polysilicon is formed and then the potential
distribution in the channel is invariable and the applied voltage is sustained by both gate
oxides. This analysis concludes that this structure has two different threshold voltages
From [28], the expression for the front and the back gate threshold voltage of the
(3.34)
where Vth1 is the threshold voltage for the back gate with n+ poly and Vth2 is the threshold
voltage for the front gate with p+ poly and n+ poly. VFB,fp and VFB,fn are given by (3.3),
γ = ε si ε ox , Qsi = qN Atsi is the channel acceptor charge and ∆VFB is the difference
between the flatband voltages associated with the front and the back gates and is given by
In the above models, both the induced and the depleted charges have been
considered in the channel region. However, for short channel devices, we neglect both the
39
charges in the derivation of the threshold voltage model. Low doping concentration of
double-gate SOI MOSFETs makes this a good approximation [31]. This approximation
∂ 2φ ( x , y ) ∂ 2φ ( x , y ) qN A
+ = ≈0 (3.37)
∂x 2 ∂y 2 ε si
As shown in [31], the above equation can be solved using the parabolic potential profile
(3.5) and (3.6) and with the help of the boundary conditions (3.7)-(3.15). The short
channel threshold voltage shift ∆Vth of DMDG SOI MOSFET can be written as
∆Vth = 2 η Sη L1 e −ζ (3.38)
where
∆VFB
η S = Vbi − VGS' , f 1 + (3.39)
t
2 1 + si
2γ t f
L L
1 ( )
Vbi + VDS − VGS' , f 2 sinh 1 + ηS sinh 2
λ λ
ηL1 = (3.40)
2
L L
L L
cosh 1 sinh 2 + sinh 1 cosh 2
λ λ λ λ
ζ = L1 2γ tsi t f (3.41)
Therefore, the expression for the threshold voltage of the DMDG SOI MOSFET is given
by
where VthL can be either Vth1 or Vth2. Equation (3.42) does not predict any threshold
voltage roll-up, whereas the simulation results of the DMDG structure exhibit a threshold
40
empirical correction factor, θ. The empirical relation used for this parameter is given by
L1 − L2
θ = 1− (3.43)
ρ L1
Here the value of ρ, when compared with the simulated results has been obtained as
ρ = kL1 − 2.25 , where L1 is in µm and k=185/µm. Therefore, the final expression for the
It is to be noted that when L1 = L2, θ is equal to unity and when the channel length is
assumed here that the length of M1 is always greater than that of M2, which is reasonable
Using the threshold model given by (3.44), the DIBL of the DMDG structure can now be
expressed as
where Vth,lin and Vth,sat are the threshold voltages in the linear and the saturation regimes,
respectively.
3.3.3 IV Model
parallel, each having its own threshold voltage Vth1 and Vth2 relating to back gate and
W µneff Cox
(VGS − Vthi )VDS − VDS2
1
I ch = ∑
i =1,2 V 2
in linear region (3.46)
L 1 + DS
LEC
41
W µneff Cox
(VGS − Vthi )VDS ,sati − VDS2 ,sati in saturation region
1
I ch = ∑
i =1,2 V 2
(3.47)
L 1 + DS , sati
LEC
where Vthi corresponds to the back gate and the front gate threshold voltage for i=1, 2,
respectively, EC is the critical electric field at which electron velocity ( υns ) saturates and
where µneff is the effective mobility (Watt’s mobility model [32]) of the inversion layer
electrons given by
1 1 1
= + (3.49)
µneff µ ph µ sr
where µph is the mobility associated with the phonon scattering and µsr is the mobility
associated with the surface roughness scattering as discussed in [28] and are given by
α1 α2
Eeff E
µ ph = µ ref 1 6 µ sr = µ ref 2 6 eff
10 V / cm 10 V / cm
where µ ref 1 , µ ref 2 , Eeff , α1 and α 2 are the parameters for Watt’s mobility model and the
However, (3.46) and (3.47) do not include the short channel effects, the parasitic
BJT effects and the impact ionization. To develop an accurate analytical drain current
Depending on the biasing condition, the impact ionization and the parasitic BJT
42
effects strongly affect the current conduction of the device. As the lateral electric field in
the device is large in the saturation region, the impact ionization and the BJT effects are
important. In the inversion layer at the oxide-silicon interface, there is a channel current
(Ich), which is due to the drifting of the electrons. In the high electric field region near the
drain, the drifting electrons collide with the lattice, resulting in electron and hole pairs.
The generated electrons and holes move in the opposite direction as a result of the
electric field − the generated electrons move towards the drain contact and the generated
holes move in the source direction. Thus, the impact ionization results in the generated
electron and hole current (Ih), equal in magnitude. For a very short-channel SOI MOS
device, the parasitic BJT with its emitter at the source and its collector at the drain cannot
be overlooked. A portion of the impact ionization current (KIh) is directed towards the
source. As a result, the holes get accumulated in the thin film, which leads to the
activation of the parasitic bipolar transistor. As the bipolar device is activated, these holes
recombine with the electrons in the base region. In the parasitic bipolar device, a portion
of the collector current (K’IC), which is mainly composed of electrons, is toward the high
electric field as result of the vertical electric field. These electrons also collide with the
lattice, and consequently also generate electron and hole pairs as described above.
The drain current is composed of the channel current (Ich), the impact ionization
current (Ih), and the collector current (IC) of the parasitic bipolar device [29]:
ID = Ich+Ih+IC (3.50)
The collector current can be expressed in terms of the emitter current (IE):
I C = α I E + I CBO (3.51)
where ICBO, which is a function of the gate voltage, is the leakage current between the
43
collector and the base with the emitter open:
I SO
I CBO = Wtsi (3.52)
1 + θ1 (VG − Vthi )
where ISO is the leakage current per unit cross section in the collector-base junction. The
source current can be expressed as the sum of the channel current (Ich), a portion of the
impact ionization current [(1-K)Ih], and the emitter current (IE) of the bipolar device:
I S = I ch + (1 − K ) I h + I E (3.53)
The impact ionization current is a function of the current components, which include the
channel current (Ich) and a portion of the collector current (K’IC) flowing through the high
β
M − 1 = α (VDS − VDS , sat ) exp − (3.55)
V −V
DS DS , sat
where α and β are process-dependent fitting parameters [33]. Note that at the onset of
saturation, M-1=0. For current continuity, the source current should be equal to the drain
current. Therefore, from eqns (3.50) - (3.54), one obtains the emitter current and the
K ( M − 1) 1 + KK '( M − 1)
IE = I ch + I CBO (3.56)
1 − [1 + KK '( M − 1)]α o 1 − [1 + KK '( M − 1)]α o
1 − αo K'
I h = ( M − 1) I ch + I CBO (3.57)
1 − [1 + KK '( M − 1)]α o 1 − [1 + KK '( M − 1)]α o
From eqns (3.50), (3.51), (3.56) and (3.57), the drain current is given by
44
( M − 1) 1 − (1 − K ) α 0 1 + K ' ( M − 1)
where G = 1+ H=
1 − 1 + KK ' ( M − 1) α 0 1 − 1 + KK ' ( M − 1) α 0
The values of K, K’, α o , ISO, ICBO [29] are given in Table 1. The drain current is equal to
Non-local effects are becoming more and more prominent as the MOSFET
dimensions shrink to the deep submicrometer regime. Velocity overshoot is one of the
most important effects from the practical point of view as it is directly related with the
increase of the current drive and the transconductance [34-37]. It has been shown that an
electric field step causes the electron velocity to overshoot the value that corresponds to
the higher field for a period shorter than the energy relaxation time. Therefore, as the
longitudinal electric field increases, the electron gas starts to be in non-equilibrium with
the lattice with the result that electrons can be accelerated to velocities higher than the
saturation velocity. This result has been shown for channel lengths under 0.15µm.
Using (3.46), (3.47) and considering the channel length modulation [38], the
velocity overshoot effects [39] and the DIBL [30], the final expression for the channel
W µ neff Cox 1 2 W 1 2
I ch = ∑
l (
)
VGS − Vthi' VDS − VDS + λa 2 ( )
VGS − Vthi' VDS − VDS
i =1,2
L 1 − d + DS
V 2 ( L − ld ) 2
L LEC
(3.59)
45
W µneff Cox 1 2 W 1 2
I ch , sat = ∑
( )
VGS − Vthi' VDS , sati − VDS , sati + λa 2 (VGS − Vthi' ) VDS , sati − VDS , sati
i =1,2 l VDS , sati
L 1 − d +
2 ( d)
L − l 2
L LEC
(3.60)
where (3.59) corresponds to the current in the linear region and (3.60) corresponds to the
current in the saturation region, λa is a parameter that takes into account velocity
Vthi' = Vthi − DIBL and ld is the channel length modulation factor as given in [38].
Using (3.58) and (3.60), the total drain current of the DMDG SOI MOSFET is given by
the expression
W µneff Cox 1 2 W 1 2
I D,sat = G∑
(
l VDS , sati GS
'
)
V − Vthi VDS ,sati − VDS ,sati + λa ( )
V − Vthi VDS ,sati − VDS ,sati + HICBO
2 GS
'
i =1,2
L 1 − +
2 ( L − ld ) 2
d
L LEC
(3.61)
Equation (3.61) corresponds to the drain current in the saturation region. Drain current in
46
Table 3.1: Device parameters used in the model and the simulation of the DMDG and the
DG SOI MOSFETs.
Parameter Value
Length of source/drain
100 nm
regions
Distance between
50 nm
source/drain contact and gate
K 0.85
K’ 0.85
ISO 80 A°/cm2
αo 0.997
θ1 6
To verify the proposed analytical model, the 2-D device simulator MEDICI was
used to simulate the different aspects, viz. the surface potential, the electric field, the
threshold voltage etc. and compare with the results predicted by the analytical model.
47
3.4.1 Surface Potential and Electric Field
Typical dimensions used for both the DMDG and the DG structures are
summarized in Table 3.1. Surface potential distribution within the silicon thin-film was
simulated with MEDICI. Fig. 3.2 shows the calculated and simulated surface potential
profile for a channel length of 100nm at the silicon-oxide interface of the DMDG
structure along with the simulated potential profile of the DG structure. It is clearly seen
that the DMDG structure exhibits a step function in the surface potential along the
channel. Because of this unique feature, the area under p+ poly front gate of the DMDG
structure is essentially screened from the drain potential variations. This means that the
drain potential has very little effect on the drain current after saturation [14] reducing the
drain conductance and the drain induced barrier lowering (DIBL) as discussed below.
The predicted values of the model (3.23) and (3.26) agree well with the simulation
results.
Fig. 3.3 shows the surface potentials at the front and back gates of the DMDG and
DG SOI MOSFETs along the channel using MEDICI. It can be observed that DMDG
structure exhibits a step in the surface potential profile at the front gate as well as at the
back gate. The step, which is quite substantial in the front gate surface potential, occurs
because of the difference between the workfunctions of n+ poly and p+ poly. Though the
back gate in DMDG structure consists of a single material, n+ poly, a small step can be
seen in the back gate surface potential when the silicon film thickness is thin enough.
This step occurs because of the coupling between the front channel and the back channel
of the DMDG structure. This step in the back gate surface potential profile enables the
DMDG structure to exhibit the reverse short channel effect (RSCE) i.e., the threshold
48
voltage rolls-up with decreasing channel lengths. Though the coupling between surface
potential profiles of the front gate and the back gate is present in the DG SOI MOSFET
also, reverse short channel effect cannot be observed. The back gate surface potential
SOI structure, since the back gate consists of n+ poly which has a workfuction smaller
than that of the p+ poly. Since the back gate surface potential of the DMDG structure has
a step profile without having used two materials in the back gate, the coupling in the
Fig. 3.4 shows the back gate surface potential profiles of the DMDG structure
along the channel for different film thicknesses. It can be clearly seen that as the film
thickness increases, the step in the potential profile decreases and diminishes when the
film thickness is thick enough. This is because when the film thickness is large, the
coupling between the front and the back gate surface potentials is minimal. This result
clearly demonstrates that to observe the RSCE or the threshold voltage roll-up with
decreasing channel lengths, the film thickness should be as small as possible so that
maximum coupling between the surface potential of the front gate and the back gate
occurs.
Fig. 3.5 shows the calculated and simulated values of the electric field along the
channel length at the drain end for the DMDG SOI MOSFET and the simulated values
for the DG SOI MOSFET for the same channel length. Because of the discontinuity in
the surface potential of the DMDG structure, the peak electric field at the drain is reduced
substantially, by approximately 40%, when compared with that of the DG structure that
leads to a reduced hot carrier effect. The agreement between the model (3.33) and
49
simulated results proves the accuracy of the model.
2.0
DMDG (MEDICI)
DMDG (MODEL)
Surface Potential (V)
1.5 DG (MEDICI)
VGS = 0.15V
1.0 VDS = 0.75V
0.5
0.0
0 20 40 60 80 100
Position along the channel (nm)
Fig. 3.2: Surface potential profiles of DMDG and DG-SOI MOSFETs for a channel
length L = 0.1µm (L1 =L2 = 0.05µm).
2.0
L1 = L2 = 50 nm DMDG
tsi = 12 nm DG
Surface Potential (V)
1.5 tf = tb = 2 nm
VGS = 0.1 V
VDS = 0.75 V
1.0 Back gate
Front gate
0.5
0.0
0.00 0.02 0.04 0.06 0.08 0.10
Position along the channel (µm)
Fig. 3.3: Surface potential profiles at front gate and back gate for DMDG and DG-SOI
MOSFETs for a channel length L = 0.1 µm (L1 = L2 = 0.05 µm) using MEDICI.
50
Back Gate Surface Potential (V)
1.6 L1 = L2 = 50 nm
tf = tb = 2 nm
1.4 VGS = 0.1 V
VDS = 0.75 V
1.2
1.0
tsi = 12 nm
0.8 tsi = 20 nm
tsi = 25 nm
0.6
0.00 0.02 0.04 0.06 0.08 0.10
Position along the channel (µm)
Fig. 3.4: Back gate surface potential profiles for different film thicknesses for DMDG
SOI MOSFETs for a channel length L = 0.1 µm (L1 = L2 = 0.05 µm) using MEDICI.
1.8
DMDG (MEDICI)
Electric Field (MV/cm)
0.6
0.3
0.0
70 80 90 100
Position along the channel (nm)
Fig. 3.5: Electric-field variation at the drain end along the channel at the Si-SiO2 interface
of DMDG and DG SOI MOSFETs for a channel length L = 0.1µm (L1 =L2 = 0.05µm).
51
3.4.2 Threshold Voltage and Drain Induced Barrier Lowering
In Fig. 3.6, the threshold voltage of the DMDG structure as a function of channel
length is compared with that of the DG MOSFET and the proposed model (3.44). It can
be observed clearly that the proposed DMDG structure exhibits a reverse short channel
effect (RSCE), i.e., the threshold voltage “roll-up” with fixed L1, while the threshold
voltage of the DG structure rolls-down with decreasing channel lengths. The threshold
voltage roll-up occurs because as the channel length decreases, keeping L1 fixed, the
proportion of the channel length influenced by p+ poly of the front gate at the back gate
increases which leads to an increase in the threshold voltage. This unique feature of the
DMDG structure is an added advantage when the device dimensions are continuously
shrinking. With decreasing channel lengths, it is very difficult to fabricate precise channel
lengths. A threshold voltage variation from device to device is least desirable. The
DMDG structure exhibits a threshold voltage that is almost constant with decreasing
channel lengths. From the results it is clearly seen that the calculated values of the
Fig. 3.7 shows the DIBL variation along the channel for both the DMDG and the
DG SOI MOSFETs. The simulated DIBL results are calculated as the difference between
the linear threshold voltage (Vth,lin) and the saturation threshold voltage (Vth,sat). The
parameters, tf, tb and tsi used here are 2nm, 3nm and 20nm respectively. The linear
1V where the critical current is defined as the drain current when VGS = Vth,lin [40]. Again
it can be observed clearly that the DIBL increase in the DMDG is far less when compared
52
0.30
0.20
50
DMDG (MEDICI)
DMDG (MODEL)
DIBL, Vth,lin -V th,sat (mV)
40 DG (MEDICI)
VDS = 1V
30
20
10
60 70 80 90 100
Channel length (nm)
Fig. 3.7: DIBL of DMDG and DG SOI MOSFETs is plotted for different channel lengths,
L=L1 + L2 where L1 = L2. The parameters used are tox =2nm, tb = 3nm, tsi = 20nm.
53
with the DG MOSFET with decreasing channel lengths. The step profile ensures that the
drain potential is screened and the surface potential minima at the source end remains
3.4.3 IV Characteristics
Drain current characteristics of both the DMDG and the DG MOSFETs are shown
in Fig. 3.8. In the case of the DMDG structure, the results obtained from the model are
also shown. Fig. 3.8 demonstrates that the DMDG structure exhibits an improved
voltage. This enhancement in the performance is because of the step function of the
surface potential profile along the channel, which reduces the DIBL and the peak electric
field at the drain end. In the drain current analytical model various short channel effects
such as the channel length modulation, the DIBL, the velocity overshoot have been
considered along with the breakdown mechanisms involved: the parasitic BJT effects and
Fig. 3.9 and Fig. 3.10 show the transconductance (gm) and the drain conductance
(gd) for both the structures for different channel lengths. The gm is extracted from the
slope of ID-VGS between VGS = 1V and 1.5V at VDS = 0.75V while gd is extracted from the
slope of ID-VDS between VDS = 0.5V and 0.75V at VGS = 1.5V for both simulation and the
model predicted values. Fig. 3.11 shows the voltage gain of the DMDG and the DG SOI
MOSFETs. Because of the increase in the transconductance and decrease in the drain
conductance, the voltage gain (gm/gd) predicted for the DMDG structure is much higher
54
DMDG (MEDICI)
4 DMDG (MODEL)
Drain Current, ID (mA) DG (MEDICI)
3 L1 = L2 = 50nm
VGS = 1.5V
2 VGS = 1V
1
VGS = 0.5V
0
0.0 0.5 1.0 1.5 2.0
Drain Voltage, VDS (V)
Fig. 3.8: ID - VDS characteristics of the DMDG and DG-SOI MOSFETs for a channel
length L = 0.1µm
1.5
1.4
gm (mS/µ m)
1.3
1.2
DMDG (MEDICI)
1.1 DMDG (MODEL)
VDS = 0.75V DG (MEDICI)
1.0
60 70 80 90 100
Channel length (nm)
Fig. 3.9: Variation of gm with different channel lengths, (L1 = L2) for DMDG and DG
SOI MOSFETs.
55
0.25
gd (mS/µ m) 0.20
0.15
0.10
20
16
Voltage gain, gm/gd
12
DMDG (MEDICI)
4
DMDG (MODEL)
DG (MEDICI)
0
60 70 80 90 100
Channel length (nm)
Fig. 3.11: Variation of voltage gain with different channel lengths, (L1 = L2) for DMDG
and DG SOI MOSFETs.
56
3.5 Summary
The concept of the Dual-Material-Gate has been applied to the Double Gate
structure and the features exhibited by the resulting new structure, Dual Material Double
Gate, have been examined for the first time by developing an analytical model. The
results obtained from the model agree well with the MEDICI simulation results. The
results show that the DMDG structure leads to reduced short channel effects as the
surface potential profile shows a step at the interface of the two materials of the front
gate, which reduces the drain conductance and the DIBL. Moreover, the peak electric
field at the drain end is reduced, minimizing the hot carrier effect. The threshold voltage
shows a roll-up with reducing channel lengths. In addition, it has been demonstrated that
the DMDG MOSFET offers higher transconductance and drain breakdown voltage. All
these features should make the proposed DMDG SOI MOSFET a prime candidate for the
future CMOS ULSI chips. Because of the asymmetric nature of the DMDG structure, it
may pose few challenges while integrating with the present CMOS technology. But Zhou
[14] suggested two fabrication procedures requiring only one additional mask step with
which a dual material gate can be obtained. As the CMOS processing technology is
maturing and already into the sub-100 nm [41] regime, fabricating a 50 nm feature gate
length should not hinder the possibility of achieving the potential benefits and excellent
immunity against the SCE’s that the DMDG SOI MOSFET promises.
57
CHAPTER IV
INVESTIGATION OF THE NOVEL ATTRIBUTES OF A SINGLE HALO
DOUBLE GATE SOI MOSFET: 2D SIMULATION STUDY
4.1 Introduction
The scaling of CMOS technology in deep submicron regime has enabled the
system-on-chip (SoC) applications that integrate the RF/analog and the digital logic
functions. Most of the analog applications rely on pure digital CMOS technology [42].
However deeply scaled analog CMOS device design for SoC applications is particularly
challenging, since the analog and digital circuit requirements are often conflicting. High
performance logic devices like the double pocket structures, with good current drive-to-
off state leakage tradeoff and good control of short channel effects, often have poor
drive current ratio, and matching properties [43]. Recently, single pocket (SP) or
asymmetric channel structures are suggested for mixed mode applications [44-45]. They
have shown good digital performance compared to the double pocket or the super halo
devices as well as the super steep retrograde devices [46], due to their laterally
asymmetric channel doping profile that suppresses SCE and exploiting velocity overshoot
The single halo MOSFET structures have been introduced for the bulk [18] as well
as the SOI MOSFETs [19] to adjust the threshold voltage and improve the SCE. The halo
implantation devices show excellent output characteristics with low DIBL, no kink,
higher drive currents, flatter saturation characteristics, and slightly higher breakdown
voltages compared to the conventional MOSFET. However, no such attempt has been
59
reported on the DG MOSFET. In this paper, for the first time we have investigated the
performance of the DG structure with halo implantation using 2-D numerical simulations.
The unique features of the double gate single halo (DG-SH) device are explored and
(Vth) variation with the channel length and the film thickness, the drain-induced barrier
lowering (DIBL), the on-current (Ion), the off-current (Ioff), and the ratio of
S D
p
+ tf
L tsi
n+ p n+
tb
n
+
(a)
G
S D
p
+ tf
L tsi
n+ p+ p n+
tb
n
+
LP (b)
Fig.4.1 : Cross-sectional view of (a) DG-SOI MOSFET (b) DG-SH SOI MOSFET
60
4.2 DG-SH structure and its parameters
implemented in the 2-D device simulator MEDICI are shown in Fig. 4.1. The doping in
the p-type body and the n+ source/drain regions is kept at 1x1015cm-3 and 5x1019cm-3
respectively. Pocket implantation, NAP, on the source side of the DG-SH MOSFET is
kept at 8x1017cm-3 while the length of the pocket, Lp, is fixed at 4nm. Typical value of the
front-gate oxide thickness (tf) and the back-gate oxide thickness (tb) is taken as 2.5nm
while the thin-film thickness (tsi) is kept at 20nm respectively. The values of different
Table 4.1: Device parameters used for the simulation of the DG-SH and the
DG SOI MOSFETs
Parameter Value
61
4.3 Results and Discussion
Simulation studies are performed to explore the characteristics of the DG-SH SOI
MOSFET with the conventional DG structure. Comparisons between the two structures
are drawn out on the basis of the threshold voltage variation with the channel length, the
DIBL, the saturation current (Ion), the off-state leakage current (Ioff), the transconductance
and the drain conductance. The DIBL is measured as the difference of the linear
threshold (Vth,lin) voltage and the saturation threshold voltage (Vth,sat). The linear threshold
ID – VGS to zero) at VDS = 50mV. The saturation threshold voltage is based on a modified
constant-current method at VDS = 0.75V [40]. The saturation current is the drain current at
VDS = 0.75V. The leakage current is the drain current at VGS = 0V and VDS = 0.75V (or VDS
= 0.05V, as stated). The transconductance, gm, is extracted from the slope of ID - VGS at
VGS = VDS = 0.75 V. The drain conductance (gd) is extracted from the slope of ID - VDS
To gain an insight into the physical mechanism responsible for the improved
performance of the DG-SH structure, the surface potential profile is plotted at the
interface of the front-gate oxide and the thin film. Fig. 4.2 shows the surface potential
plots for both the DG and the DG-SH MOSFETs for channel lengths 0.1µm and 0.2µm.
As can be observed from the figure, a small step on the source side occurs in the surface
potential profile in the case of DG-SH structure. This step function in the case of DG-SH
structure is due to the higher doping near the source end. Because of this step in the
surface potential profile, there is almost a zero shift in the minimum surface potential
62
2.0
VGS = 0.25V
1.6
Surface Potential (V)
VDS = 0.75V
1.2
0.8
0.4
DG
DG-SH
0.0
0.00 0.04 0.08 0.12 0.16 0.20
Position along the channel (µm)
Fig. 4.2: Surface potential profiles of DG-SH and DG-SOI
MOSFETs for channel lengths 0.1µm and 0.2µm with a film
thickness of 20nm.
which leads to a reduced DIBL and an extended threshold voltage roll-off is observed.
Fig. 4.3 shows the threshold voltage variation with channel length for the DG and
the DG-SH down to 80nm for a fixed film thickness, tsi = 20nm. The so-called reverse
short channel effect [47-50] is clearly seen here as the threshold voltage slightly increases
with decrease in channel length for the DG-SH MOSFET. This is because of the
screening of the drain voltage as the channel length decreases due to the step function in
the surface potential profile and the surface potential minima is essentially least affected.
On the other hand, the threshold voltage rolls-off rapidly for the DG MOSFET. In the
case of the DG structure, as the channel length decreases, the drain bias shifts the surface
potential minimum which leads to the roll-off in the threshold voltage. It can also be
63
noted that the threshold voltage of the DG-SH structure is slightly higher than that of the
DG structure due to the increase in the doping at the source end in the former. Fig. 4.4
shows the threshold voltage variation with film thickness, tsi, for a fixed channel length, L
= 100nm. Due to the presence of the reverse short channel effect, it can be observed that
the threshold voltage dependence on film thickness in the DG-SH MOSFET is much less
compared to the DG structure. Another important short channel effect is the DIBL.
Fig. 4.5 shows the DIBL parameter for both the DG and the DG-SH MOSFETs
for channel lengths down to 80nm. This figure demonstrates that the DIBL is far less in
the DG-SH structure when the channel length is reduced below 120nm than in the case of
the DG structure.
0.25
Threshold Voltage, Vth (V)
0.20
0.15
0.10
64
0.30
30
DG
DIBL, Vth,lin-V th,sat (mV)
25
DG-SH
20
15
10
0
0.09 0.12 0.15 0.18 0.21
Channel length (µm)
Fig. 4.5: DIBL of DG-SH and DG SOI MOSFETs is plotted for different channel lengths
for a film thickness of 20nm.
65
4.3.3 Subthreshold Slope and On/Off Currents
Fig. 4.6 shows the subthreshold slope of the DG and the DG-SH MOSFETs. It is
clear from the figure that the subthreshold behavior of both the structures is almost
identical and the subthreshold slope is around 60mV/dec. The subthreshold slope of the
DG-SH is expected to degrade because of the increase in the doping near the source end,
but as can be seen, the increase is very small. The increase in subthreshold slope in DG-
SH MOSFET gives rise to a notion that the off-state leakage current of this structure will
be greater than that of the DG MOSFET. But due to the screening of the drain bias by the
step function in the surface potential profile, the increase in the leakage current is
suppressed and the off-state leakage current is in fact less than that of the DG MOSFET.
64
Subthreshold slope, S(mV/dec)
DG
DG-SH
63
VDS = 50mV
62
61
60
0.08 0.12 0.16 0.20
Channel length (µm)
Fig. 4.6: Subthreshold slope of DG-SH and DG SOI MOSFETs is plotted for different
channel lengths for a film thickness of 20nm.
66
Off state Leakage current, Ioff (A/µm)
10-7 1.5
6×10-8 0.9
0 0.0
0.08 0.12 0.16 0.20
Channel length (µ m)
Fig. 4.7(a): Variation of Ioff and Ion with channel length for DG-SH and DG SOI
MOSFET for a film thickness of 20nm.
3.0 ×105
DG
2.4 ×105
DG-SH
1.2 ×105
6×104
0
0.08 0.12 0.16 0.20
Channel length (µm)
Fig. 4.7(b): Ratio of Ion and Ioff with channel length for DG-SH and DG SOI MOSFET for
a film thickness of 20nm at VDS= 0.75V.
67
Fig. 4.7(a) shows the saturation current (Ion) and the off-state leakage current (Ioff)
for both the configurations. It can be seen that the saturation current of the DG-SH
MOSFET is slightly lower than that of the DG MOSFET due to the increased threshold
voltage of the former structure. It can also be observed that the off-state leakage current
in the DG-SH structure is much less and there is nearly an order of magnitude difference
in this current for a drain voltage, VDS= 0.75V, when the channel length is in the sub
100nm regime for the two structures. In other words, the on-off current ratio, (Ion/Ioff), of
the DG-SH MOSFET is large compared to that of a DG MOSFET. This can be clearly
observed from Fig. 4.7(b), where the Ion/Ioff ratio of both the DG and the DG-SH is
shown.
4.4 IV Characteristics
Output characteristics of the DG and the DG-SH SOI devices are compared for the
same channel length, L = 0.1µm in Fig. 4.8. It is evident that the drive capability of the
DG-SH is slightly less when compared with conventional the DG structure. This is
because of the increase in the threshold voltage of the DG-SH when compared to the DG
structure. It can be easily interpreted that while the transconductance is almost same for
both the structures, the drain conductance is much less in the case of the DG-SH
MOSFET, which leads to an increased voltage gain. The other advantages of the DG-SH
MOSFET are the absence of kink in the drain characteristics and a slight increase in the
Fig. 4.9 shows the transconductance (gm) and the drain conductance (gd) of the
DG and the DG-SH MOSFETs. This figure clearly demonstrates that while the
transconductance of both the structures is identical, the drain conductance is much less in
68
the case of the DG-SH MOSFET. This is primarily due to the reverse short channel effect
and the reduced DIBL in the case of the DG-SH MOSFET. Fig. 4.10 shows the voltage
gain of both the structures. Because of the drastic improvement in the drain conductance,
a considerable increase in the voltage gain (gm/gd) can be observed for the DG-SH
3.0
Drain Current, ID (mA/µm)
VGS=1.5V
2.5
2.0
VGS=1.0V
1.5
1.0
0.5 DG
VGS=0.5V DG-SH
0.0
0.0 0.5 1.0 1.5 2.0
Drain Voltage, VDS (V)
Fig. 4.8: ID-VDS characteristics of the DG-SH and DG-SOI MOSFETs for a channel
length L = 0.1µm with a film thickness of 20nm.
69
3.0
DG
gm , gd (mS/µm) 2.5 DG-SH
2.0
gm
1.5
1.0
gd
0.5
0.0
0.08 0.12 0.16 0.20
Channel length (µm)
Fig. 4.9: Variation of gm, gd with different channel lengths for DG-SH and DG SOI
MOSFETs.
10
8
Voltage gain, gm /gd
2 DG
DG-SH
0
0.08 0.12 0.16 0.20
Channel length (µ m)
Fig. 4.10: Variation of voltage gain with different channel lengths DG-SH and DG SOI
MOSFETs.
70
4.4 Summary
gate (DG) SOI MOSFET. The features of the resulting structure, the double gate single
halo (DG-SH) SOI MOSFET have been studied in the context of its potential integration
in the current CMOS technology and a comparison has been drawn out with the
conventional asymmetric DG SOI structure. The unique features of the DG-SH MOSFET
that are not easily available in the conventional asymmetric DG SOI devices include:
threshold voltage roll-up, reduced DIBL, kink free output characteristics and an
71
CHAPTER V
CONCLUSIONS
The work presented in this thesis can be divided into three categories, namely, (1)
Analytical Model for a partially depleted (PD) dual-material gate (DMG) SOI MOSFET,
(2) Analytical model for a dua1-material double gate (DMDG), and (3) 2-D numerical
simulation studies of a double gate single halo (DG-SH) doped SOI MOSFET.
Analytical model of the surface potential along the channel and of threshold
voltage in a DMG-PD SOI MOS device is developed by solving the 2-D Poisson’s
1. The 2-D surface potential model predicts a step in the channel potential profile
due to the presence of the two different gate materials with the finite
2. The shift in the surface channel potential minima position is negligible with the
increasing drain biases. This leads to an excellent immunity against the short-
channel effects (SCE) like the drain-induced barrier lowering (DIBL) and the
3. The peak electric field at the drain end is lowered leading to a reduced hot-carrier
effect.
4. The threshold voltage, Vth, rolls-up with decreasing channel length in a DMG SOI
MOSFET down to 0.1µm if the length and workfunction of the gate materials are
chosen appropriately.
73
b) DMDG SOI MOSFET
Analytical model of the surface potential, the threshold voltage and the
drain characteristics is developed and the short channel performance of the proposed
DMDG structure is compared with the asymmetrical DG SOI MOSFET. The conclusions
are:
1. The surface potential model predicts the creation of a step in the channel potential
compared with the DG SOI MOSFET. The short channel behavior was analyzed
with respect to the threshold voltage and the DIBL variation with channel length.
The proposed structure exhibits a threshold voltage roll-up and a reduced DIBL
when compared with the DG structure. Also the peak electric field at the drain
effect.
3. The proposed IV model and the simulated characteristics for the DMDG structure
increase in the voltage gain. The IV characteristics of the proposed structure also
2-D MEDICI simulations were used to explore and compare the novel attributes
offered by the double gate single halo (DG-SH) doped structure with a conventional
asymmetrical DG SOI in terms of Vth variation with the decreasing channel length,
74
the drain-induced barrier-lowering (DIBL), the leakage current, the drive current,
the transconductance, the drain conductance and the voltage gain. The conclusions
are:
1. The surface potential profile exhibits a small step because of the increased doping
2. The unique features of the DG-SH are: Vth roll-up with decreasing channel length,
1. The proposed structures can be applied at the circuit level (e.g. inverter) and the
2. A lot of scope lies in studying the effect of the proposed structures on devices
proposed structures.
75
76
APPENDIX A
77
MODELS CONMOB SRFMOB CONSRH FLDMOB AUGER BGN
COMMENT Solve using refined grid save solution for later use
SYMB CARRIERS=0
SOLVE OUT.FILE=DMPDSOI1S_10
78
COMMENT Gate Characteristics for a 0.1 µm DMG SOI MOSFET.
COMMENT Extract MOS parameters like the linear threshold voltage, channel length, etc
EXTRACT MOS.PARA IN.FILE=DMPDSOI1GI GATE=Gate1
79
APPENDIX B
80
+ IN.FILE=DDMDSOI1DS_20
+ OUT.FILE=DDMDSOI1MS_20
PLOT.2D GRID TITLE="SOI-NMOSFET - POTENTIAL REGRID" FILL SCALE
COMMENT Solve using refined grid save solution for later use
SYMB CARRIERS=0
SOLVE OUT.FILE=DDMDSOI1S_20
81
COMMENT Drain characteristics for DMDG SOI MOSFET
82
APPENDIX C
83
PLOT.2D GRID TITLE="SOI-NMOSFET - POTENTIAL REGRID" FILL SCALE
COMMENT Solve using refined grid save solution for later use
SYMB CARRIERS=0
SOLVE OUT.FILE=DMDSOI1S_20
84
REFERENCES
85
stopper (TIPS),” IEDM Tech. Dig., pp. 75-78, 1994.
[18] S. Odanaka, and A. Hiroki, “Potential Design and Transport Property of 0.1 µm
MOSFET with Asymmetric Channel Profile,” IEEE Trans. Electron Devices,
Vol. ED-44, pp. 595-600, 1997.
[19] B. Cheng, V. Ramgopal Rao, J. C. S. Woo, “Sub 0.18 µm SOI MOSFETs using
lateral symmetric channel profile and Ge pre-amorphization salicide
technology,” Proc. IEEE pp. 113-114, 1998.
[20] L. A. Greenfield and R. W. Dutton. “Nonplanar VLSI device analysis using the
Poisson’s equation,” IEEE Trans. Electron Devices, vol. ED-27, pp. 1520-
1532, 1980.
[21] H. S. Lee, “An analysis of the threshold voltage for short-channel IGFETs,”
Solid State Electron., vol. 16, pp. 1407-1417, 1973.
[22] G. W. Taylor, “Subthreshold conduction in MOSFETs,” IEEE Trans. Electron
Devices, vol. ED-25, pp. 337-350, 1978.
[23] T. Toyabe and S. Asai, “Analytical models of threshold voltage and breakdown
voltage of short-channel MOSFETs derived from the two-dimensional analysis
methods,” IEEE Trans. Electron Devices, vol. ED-26, pp. 453-461, 1979.
[24] D. R. Poole and D. L. Kwong, “Two dimensional analysis modeling of
threshold voltage of short-channel MOSFETs,” IEEE Electron Device Lett.,
vol. EDL-5, pp. 443-446, 1984.
[25] J. D. Kendall and A. R. Boothroyd, “A two-dimensional analytical threshold
voltage model for MOSFETs with arbitrarily doped substrate,” IEEE Electron
Device Lett., vol. EDL-7, pp. 401-403, 1986.
[26] K. K. Young, “Short-Channel Effect in Fully Depleted SOI MOSFET’s,” IEEE
Trans. Electron Devices, vol. ED-36, pp.399-402, 1989.
[27] H. S. Wong, D. Frank, and P. Solomon, “Device design considerations for
Double-Gate, Ground-Plane and Single-Gated Ultra-Thin SOI MOSFETs at the
25nm channel length generation,” IEDM Tech. Dig., p. 407, 1998.
[28] K. Suzuki, and T. Sugii, “Analytical models for n+-p+ double gate SOI
MOSFET’s,” IEEE Trans. Electron Devices, vol. ED-42, pp. 1940-1948, 1995.
[29] Y. G. Chen, J. B. Kuo, Z . Yu, and R. W. Dutton, “An analytical drain current
model for short-channel Fully-Depleted ultrathin silicon-on-insulator NMOS
devices,” Solid State Elelctron., vol. 38, pp. 2051-2057, 1995.
[30] J. B. Roldan, F. Gamiz, J. A. Lopez-Villanueva, P. Cartujo, and J. E. Carceller,
“A Model for the Drain Current of Deep Submicrometer MOSFET’s including
Electron-Velocity Overshoot,” IEEE Trans. Electron Devices, vol. ED-45, pp.
2249-2251, 1998.
[31] K. Suzuki, Y. Tosaka, and T. Sugii, “Analytical threshold voltage model for
short channel Double Gate MOSFET’s,” IEEE Trans. Electron Devices, vol.
ED-43, pp. 732-738, 1996.
[32] J. T. Watt and J. D. Plummer, “Universal mobility-field curves for electrons
and holes in MOS inversion layers,” Symp. VLSI Tech. Dig. pp. 81-82, 1987.
[33] S. M. Sze, Physics of Semiconductor Devices, 2nd edn., pp. 45-47, John Wiley
& Sons, Canada, 1981.
[34] G. A. Sai-Halasz, M. R. Wordeman, D. P. Kern, S. Rishton, and E. Ganin,
“High transconductance and velocity overshoot in NMOS devices at the 0.1µm
86
gate-length level,” IEEE Electron Device Lett., vol. EDL-9, p. 464, 1988.
[35] S. Y. Chou, D. A. Antoniadis, and H. I. Smith, “Observation of electron
velocity overshoot in sub-100-nm-channel MOSFET’s in silicon,” IEEE
Electron Device Lett., vol. EDL-6, p. 665, 1985.
[36] G. G. Shahidi, D. A. Antoniadis, and H. I. Smith, “Electron velocity overshoot
at room and liquid-nitrogen temperatures in silicon inversion layers,” IEEE
Electron Device Lett., vol. EDL-9, p. 94, 1988.
[37] F. Assaderaghi, P. D. Ko, and C. Hu, “Observation of velocity overshoot in
silicon inversion layers,” IEEE Electron Device Lett., vol. EDL-14, p. 484,
1993.
[38] N. D. Arora, R. Rios, C. L .Huang, and K. Raol, “PCIM: A Physically Based
Continuous Short-Channel IGFET Model for Circuit Simulation,” IEEE Trans.
Electron Devices, vol. ED-41, pp. 988- 997, 1994.
[39] J. B. Roldan, F. Gamiz, J. A. Lopez-Villanueva, and J. E. Carceller, “Modeling
effects of electron velocity overshoot in a MOSFET,” IEEE Trans. Electron
Devices, vol. ED-44, pp. 841-846, 1997.
[40] X. Zhou, K. Y. Lim, and D. Lim, “A simple and unambiguous definition of
threshold voltage and its implications in deep-submicron MOS device
modeling,” IEEE Trans. Electron Devices, vol. ED-46, pp. 807-809, 1999.
[41] J. M. Hergenrother et. al., “The vertical replacement-gate (VRG) MOSFET: A
50-nm vertical MOSFET with lithography-independent gate length,” IEDM
Tech. Dig., pp. 75-79, 1999.
[42] K. Bult, “Analog broadband communication circuits in pure digital deep sub-
micron CMOS,” IEEE ISSCC Tech. Dig., pp. 76–77, 1999.
[43] E. Morifuji, A. Oishi, K. Miyashita, S. Aota, M. Nishigori, H. Ootani, T.
Nakayam, K. Miyamoto, F. Matsuoka, T. Noguchi, and M. Kakumu, “An 1.5 V
high performance mixed signal integration with indium channel for 130 nm
technology node,” IEDM Tech. Dig., pp. 459–462, 2000.
[44] A. Chatterjee, K. Vasanth, D. T. Grider, M. Nandakumar, G. Pollack, R.
Aggarwal, M. Rodder, and H. Shichijo, “Transistor design issues in integrating
analog functions with high performance digital CMOS,” in VLSI Tech. Dig.,
pp. 147–148, 1999.
[45] B. Cheng, A. Inani, R. Rao, and J. C. S. Woo, “Channel engineering for high
speed sub-1.0 V power supply deep sub-micron CMOS,” in VLSI Tech. Dig.,
pp. 79–80, 1999.
[46] Y. V. Ponomarev, P. A. Stolk, A. van Brandenburg, C. J. J. Dachs, M. Kaiser,
A. H. Montree, R. Roes, J. Schmitz, and P. H. Woerlee, “A 0.13 µm CMOS
technology for low-voltage analog applications,” in Proc. ESSDERC, pp. 180–
183, 1999.
[47] H.I. Hanafi, W.P. Noble, R.S. Bass, K. Varahramyan, Y. Lii, A.J. Dally, “A
model for anomalous short-channel behavior in submicron MOSFET’s,” IEEE
Electron Device Lett., vol. EDL-14, pp. 575-577, 1993.
[48] J. Tanaka, T. Toyabe, H. Matsuo, S. Ihara, H. Masuda, F. Otsuka, “Reverse
short-channel effect of threshold voltage in LOCOS parasitic MOSFET’s,”
Solid State Electron., vol. 38, pp. 567-572, 1995.
[49] S.T. Hsu, I.H. Kalish, K. Suzuki, R. Kawabata, H. Shibayama, “Physical
87
mechanism of the reverse short-channel effect in MOS transistors,” Solid State
Electron., vol. 34, pp. 605-608, 1991.
[50] N. D. Arora, and M. S. Sharma, “Modeling the anomalous threshold voltage
behavior of submicrometer MOSFET’s,” IEEE Electron Device Lett., vol.
EDL-13, pp. 92-94, 1992.
88
LIST OF PUBLICATIONS
89