marroun2021 (1)

Vol.:(0123456789)
1 3
Transactions on Electrical and Electronic Materials
https://doi.org/10.1007/s42341-020-00275-z
REGULAR PAPER
Physical Modeling onEffective Traps Density Near theConduction
Band Dependence ofElectrical Characteristics ofAmorphous Indium
Gallium Zinc Oxide Thin‑Film Transistors
AbdelhadMarroun1 · NaimaAmarTouhami1· Taj‑eddinElHamadi1
Received: 10 August 2020 / Revised: 12 November 2020 / Accepted: 21 December 2020
© The Korean Institute of Electrical and Electronic Material Engineers 2021
Abstract
An (a-IGZO TFT) is modeled through experimental-based (a-IGZO TFTs) using (TCAD) simulator. A parametric study
is performed on the numerical fit of the designed (a-IGZO TFT) current–voltage (I/V) characteristics, to obtain the near
conduction band defects parameters optimal values, and to investigate the effect of the near band defects caused by oxygen
vacancies on the (a-IGZO TFTs) output parameters. A new model approach is proposed for simulating (a-IGZO) electrical
properties. The proposed model is known as a density of state models (DOS), and it is composed of two principal compo-
nents, conduction band tail
(
gA
ct
(E)
)
and Gaussian distributed donor-like
(
gD
G
(E)
)
. The study of the presented (DOS) models
is based on both conduction band tail elements that are known as tail acceptor density (
gta
) as well as tail acceptor energy
(
Ea
), and Gaussian distributed donor-like elements which are donor gaussian energy (
ED
) and donor gaussian distribution
(
ggd
). Results show that the tail acceptor states defects (
gta
,
Ea
) near the conduction band is the cause of the mobility and
gaussian donor distribution degradation near the conduction band, and it has a major impact on changes that occur in the
subthreshold region data [threshold voltage (Vth), subthreshold swing (SS), and on-state/off-states current ratio (Ion/Ioff)].
Keywords Numerical simulation· Thin film transistors TFTs· a-IGZO TFT· Electrical characteristics· Defects
1 Introduction
Amorphous Indium-Gallium-Zinc-Oxide (a-IGZO) is con-
sidered as one of the most important amorphous oxide semi-
conductors (AOS) because of its large electron mobility. It
became the center of attention of a group of studies that
adapt amorphous Indium-Gallium-Zinc-Oxide (a-IGZO)
thin film transistors (TFTs) as the next generation of (TFT)
technology (high-resolution active-matrix organic light-
emitting diode displays (AMOLEDs) [1, 2], liquid crys-
tal displays (AMLCD), RFID Tags, and sensors [3, 4] …)
because of its outstanding characteristics of a field-effect
mobility µeff ranging from 8 to 20cm2/Vs, a threshold volt-
age (Vth) around 0V, an off current (Ioff) below 1E-12 A,
high on-state/off-states current ratio (Ion/Ioff)
≥
1E6, and
a subthreshold swing (SS) below 0.2V/Dec [5–8]. The
most important characteristic that plays an important role
in developing this type of technology is the stability of
(a-IGZO TFTs) that is defined by the shift of subthreshold
region data (threshold voltage (Vth), subthreshold swing
(SS) and on-state/off-states current ratio (Ion/Ioff)) [9–11].
Hence, many types of research study the factors that
affect the changes in the subthreshold region data. Among
the possible causes of these changes are oxygen vacancies
near the conduction band minimum (CBM) states [12–14].
Refer to some studies her. The numerical simulation is an
essential and necessary tool for the understanding of the
device physics and the (TFT) operation principles. However,
there are barely any publications that discuss the numerical
simulation of the (a-IGZO TFT) even with the incredible
advances in device fabrication.
In this paper, A design of (a-IGZO TFT) is proposed
and studied. This study is based on the optimization of
the current–voltage (I/V) characteristics of the (a-IGZO
TFT) numerical fit. Thereafter, a more accurate modeling
approach which is named density-of-state (DOS) models
is proposed for simulating (a-IGZO) electrical properties
using a two-dimensional (2D) technology computer-aided
Online ISSN 2092-7592
Print ISSN 1229-7607
* Abdelhafid Marroun
marroun.abdelhafi[email protected]
1 Faculty ofScience, Abdelmalek Essaadi University,
M’Hannech II, 93000Tétouan, Morocco

Transactions on Electrical and Electronic Materials
1 3
design (TCAD) simulator. These models contain two com-
ponents near the conduction band minimum (CBM) states,
(
gA
ct
(E)
)
and (
gD
G
(E
)
) that are known as conduction band-
tail and gaussian-distributed donor-like, respectively. So, to
investigate the (DOS) components effect on the subthreshold
region data ((Vth), (SS), and (Ion/Ioff)). A study of the model
tail acceptor density (
gta
) and tail acceptor energy (
Ea
) for
the conduction band-tail
(
g
A
ct
(E)
)
on one hand, and donor
Gaussian energy (
ED
) and donor Gaussian distribution (
ggd
)
for the gaussian-distributed donor-like (
gD
G
(E
)
) on the other
hand, is required since the mathematical (DOS) models are
an exponential function.
2 Experimental Methods
The schematic structure of the (a-IGZO TFT) is a bottom-
gate staggered structure. The device is composed of a heav-
ily doped
(n++ )
silicon wafer that acts as a gate, and a
100nm of (SiO2) layer placed on top of it that is used as a
gate insulator. Also, a 20nm thick (a-IGZO) active layer was
deposited on the unheated (SiO2) layer. Finally, the (Au/Ti)
stacked layer
40∕5 nm
thick was deposited as the source
and drain electrodes. Noting that the channel width (W) and
length (L) are 180μm, 30μm, respectively.
For the operation of a thin film transistor, the following
standard Eqs.(1)–(3) are used [15, 16]. The threshold volt-
age (Vth) and field-effect mobility (µsat) were extracted from
the expression of the operation of a field-effect transistor,
which is a function of the capacitance and applied voltage:
where Ids is the drain current, (W) is the channel width, (L)
is the channel length, (Vgs) is the gate voltage, (Vds) is the
drain bias voltage, (Cox) is the capacitance per unit area of
the gate insulator, and (Vth) and (µsat) are the threshold volt-
age and Saturation mobility that were extracted from the
linear fit of Eq.1.
Subthreshold swing (SS) is a gate voltage required for the
decade increase in drain current at a constant drain voltage.
Typically, the (SS) of an (a-IGZO TFT) is in the range of
0.2–0.5V/decade [17], and it can be further reduced to be
lower than 0.1V/decade by, for example, reducing the chan-
nel bulk trap density, applying high-K dielectrics, or adopt-
ing fully depleted states. Subthreshold swing is obtained
from the equation given below:
The (SS) was also extracted from the subthreshold region
data at the maximum slope point using Fig.4 transfer
(1)
I
DS =
W
L
Cox 𝜇sat
(
VGS −Vth
)
,VDS ≥VDS −V
th
(2)
SS
=
𝜕V
gs
𝜕
(
logI
ds)|
|
|
|
|
Vds =
Constant
characteristics in the logarithmic scale. Saturation mobil-
ity is defined based on how quickly an electron can move
through the active layer, which is obtained by the following
equation:
2.1 2D Numerical Simulation
The two-dimensional cross-section along the channel of
the (a-IGZO TFT) structure used in this work is shown in
Fig.1. It is designed to match with the actual (TFT) used in
this study. The structure consists of a 20nm thick (a-IGZO)
channel layer, a heavily doped n-type poly-silicon substrate
that also acts as a gate, a 100nm thick gate insulator (SiO2)
layer, and drain and source Ohmic contacts that are the low
resistance (Au/Ti)
40∕5 nm
thick stacked layer. The channel
width (W) is 180μm, and the channel length (L) is 30μm,
respectively.
Contacts between (S/D) electrodes and the (a-IGZO)
layer are both assigned as Schottky in this work. The (S/D)
metal work function (
𝜙m=4.33eV)
and the electron affin-
ity of (a-IGZO)
(
𝜒
a−IGZO)
were included in the calculation.
Both thermionic emission and tunneling current were con-
sidered [18].
(
𝜒
a−IGZO)
is estimated from a simple linear
relation between electron affinities of its three elementary
compounds.
where a, b, and c are molar percentages.
(
𝜒
a−IGZO)
was cal-
culated to be 4.16eV. Throughout this paper, the Schottky
contact model is used as a default in numerical simulation.
The channel is made of (a-IGZO) which is an amorphous
n-type semiconductor. Disordered materials [like a-Si and
(a-IGZO)] contain a large number of defect states continu-
ously distributed within the bandgap of the material. The
(3)
𝜇
sat =⎡
⎢
⎢
⎢
⎣
d√Ids
dVgs
1
�
CiW
2L
⎤
⎥
⎥
⎥
⎦
2
(4)
𝜒a−IGZO
=a
(
𝜒
In2O2)
+b
(
𝜒
Ga2O3)
+c
(
𝜒
ZnO)
Fig. 1 Structure of an a-IGZO TFT in a bottom gate (top-contact)
staggered configuration

Transactions on Electrical and Electronic Materials
1 3
density of states is a combination of exponentially decaying
band tail states and Gaussian distributions of mid-gap states.
The conduction band-tail (CB) states and the valence band-
tail (VB) states are given by exponential function decay:
where
(gta)
and
(
g
td)
(
cm−3eV−1
) are the effective density at
(Ec)
,
(Ea)
is the characteristic slope energy of the conduction
band-tail states, and
(
E
v)
and
(Ed)
are the characteristic slope
energy of the valence band-tail states.
In addition to tail states, Gaussian-distributed donor-like
and acceptor-like defect states,
(
g
D
G)
and
(
g
A
G)
, respectively
are also considered in the energy gap.
where
(
g
gd)
and
(
g
ga )
are the total density (
cm
−3
eV−1
),
(
𝜎
D)
and
(𝜎A)
are the standard deviation, and
(ED)
and
(
E
A)
are
the peak energy of the Gaussian distribution. Usually, the
(a-IGZO) density of gap states is formed by donor tail state
(
g
D
vt
(E)
)
, donor Gaussian distribution
(
g
D
G
(E)
)
with a maxi-
mum located at
2.9 eV
, and acceptor tail state
(
gA
ct
(E)
)
[19].
Figure2. Shows the different components of the density of
states in (a-IGZO), and their values are presented in Table1.
3 Results andDiscussion
Figures3 and 4 show the optimized fittings to the experi-
mental output and the transfer characteristics of a typical
depletion-type (a-IGZO TFT).
As depicted in these figures, the output characteristics
of (a-IGZO TFTs) is performed for different voltage val-
ues (
(Vgs)
from 4 to 20V) and (Vds = 0.1V). Results show
that the threshold voltage (Vth), the on-current (Ion), and the
sub-threshold swing (SS) are 1.5V, 3.8E − 7A, and 0.19V/
decade, respectively. According to the accuracy of the (2D)
simulation results, a study of the influence of each density-
of-state (DOS) parameter is taken into account. this study
is based on the evaluation of the tail acceptor states effect
that is defined by its energy decay
(Ea)
and its density
(gta)
.
So, by examining the effect of
(
E
a)
and
(
g
ta)
on the transfer
characteristics and the output parameters, only one (DOS)
(5)
g
A
ct(E)=gtaexp
(
E
−
Ec
E
a)
(6)
g
D
vt(E)=gtdexp
(
Ev−E
E
d)
(7)
g
D
G(E)=ggdexp
−
E−ED
2
𝜎2
D
(8)
g
A
G(E)=ggaexp
−
EA−E
2
𝜎2
A
parameter is varied, while the other parameters are kept the
same as those listed in Table1.
Figures5 and 6 displays tail acceptor density
(
g
ta)
and tail
acceptor energy
(Ea)
of the acceptor tail state, which ranges
from 0.7E20 to 2.5E20 and 0.005 to 0.016, respectively.
Based on these figures, it is obvious that the effect of each of
gta
and
(
E
a)
does not appear on the subthreshold region data,
since the sub-threshold swing (SS) and the threshold volt-
age (Vth)keep the same values which are 1.5V, and 0.19V/
decade, respectively. However, according to these previous
various values, we notice a great effect in terms of mobil-
ity and (Ids)on (Vgs = 20V). According to saturation mobil-
ity (µsat) values presented in Eq.(6), and constructed from
Fig. 2 Proposed DOS model for a-IGZO
Table 1 The physical parameters of the a-IGZO TFT used in this
work
Symbol Description Value
Nc(
cm
−3)
Effective DOS in the conduction band 5E18
Nv(
cm
−
3
)
Effective DOS in the valence band 5E18
Eg(eV)
Bandgap 3.05
𝜒(eV)
Electronic affinity 4.16
𝜀
Permittivity 10
gta(
cm−3eV
−1)
The density of tail states at
E=Ec
1.55E20
gtd(
cm
−3
eV
−1)
The density of tail states at
E=Ev
1.55E20
gd(
cm
−3
eV
−1)
The peak of the donor-like Gaussian
states
6.5E16
Ea(meV)
Conduction-band-tail slope 13
Ed(meV)
Valence-band-tail slope 120
𝜇n(
cm
2
∕Vs
)
Free electron mobility 15
𝜇p(
cm2∕Vs
)
Free hole mobility 0.1

Transactions on Electrical and Electronic Materials
1 3
both Figs.5 and 6, more specifically from the square root
of the drain current (Ids) plot, we note that the higher is the
value of
(
g
ta)
and
(
E
a)
, the lower is the mobility. This decline
could be due to the unrest created by oxygen disorder, which
determines the nature of defects near the Conduction Band
Minimum (CBM) states.
The peak energy of donor Gaussian distribution (
ED
) and
total density of donor Gaussian distribution (
ggd
) param-
eters of the Gaussian effect on the transfer characteristics
and output parameters are examined. Figure7 shows that the
proposed (DOS) models for (a-IGZO), near the Conduction
Band Minimum (CBM) for peak energy of donor Gauss-
ian distribution (
ED
) are varied from 1.8 to 2.9eV, while
donor Gaussian distribution (
ggd
) are kept fixed at 6.5E16
cm−3eV−1
.
Based on the results shown in Fig.8 that illustrates the
(a-IGZO TFT) simulated linear region (Vds = 0.1V) transfer
characteristics in both logarithm and Square root scale for
different values of donor gaussian distribution peak energy,
we observe that the position of donor Gaussian distribu-
tion affects the (TFT) transfer characteristics, since a slight
change is observed at the subthreshold region data, while
Fig. 3 Output characteristics of the experimental a-IGZO TFT along
with optimized fittings of simulation
Fig. 4 Transfer characteristics of the experimental a-IGZO TFT along
with optimized fittings of simulation in linear and logarithmic scale
Fig. 5 a-IGZO TFT simulated linear region (Vds = 0.1 V) transfer
characteristics curves in logarithm scale and Square root scale for
various tail acceptor density (
gta
) Real experimental data also showed
for reference
Fig. 6 a-IGZO TFT simulated linear region (Vds = 0.1 V) transfer
characteristics curves in logarithm scale and Square root scale for
various tail acceptor energy (
Ea
) Real experimental data also showed
for reference

Transactions on Electrical and Electronic Materials
1 3
a great change has occurred on the donor Gaussian distri-
bution [peak energy of donor Gaussian distribution (
ED
)]
position.
Figure9shows that the proposed (DOS) models for
(a-IGZO), near the Conduction Band Minimum (CBM)
for peak energy of donor Gaussian distribution (ED)
arekept fixed at2.9 eV, while donor Gaussian distri-
bution (ggd) are varied from 3.5E16 cm−3eV−1 to
7E17cm−3eV−1.The simulation results of the total
density donor Gaussian distribution (
ggd
) dependence
on transfer characteristics is represented in Fig.10. As
noticed, when donor-like Gaussian density (
ggd
) traps
are increased, the (Ids/Vgs) is shifted to negative gate bias
direction. And it also tends to be a conductor. However,
when donor-like Gaussian density attains 7E17
cm−3eV−1,
the simulated transfer characteristics curves that corre-
spond to the experimental one. To sum up, we can con-
clude that the tail acceptor states defects near the conduc-
tion band is the reason behind the mobility as well as the
gaussian donor distribution near conduction band degrada-
tion, which has by its turn a major impact on changes that
occur in the subthreshold region data [threshold voltage
(Vth) and subthreshold swing (SS)].
Fig. 7 The Proposed DOS model for a-IGZO for various peak energy
of donor Gaussian distribution (
ED
)
Fig. 8 The a-IGZO TFT simulated linear region (Vds = 0.1V) trans-
fer characteristics curves in logarithm scale and Square root scale for
various peak energy of donor Gaussian distribution (
ED
)
Fig. 9 The Proposed DOS model for a-IGZO for the various total
density of donor Gaussian distribution (
ggd
)
Fig. 10 a-IGZO TFT simulated linear region (Vds = 0.1 V) transfer
characteristics curves in logarithm scale and Square root scale for the
various total density of donor Gaussian distribution (
ggd
)