I. INTRODUCTION
With the recent growth in information technology, next-generation semiconductors in
the steadily growing semiconductor market require high performance and reduced power
consumption. For displays (one of the key components in electronic devices), the market
is increasingly demanding higher resolution and larger sizes. Accordingly, recent
attention has been given to oxide thin-film transistor (TFT) with high charge mobility
and excellent electro-optical uniformity. Oxide TFTs are advantageous in creating
screens because of their low temperature processing, high charge mobility, and high
reliability, compared to conventional amorphous silicon (a-Si)-based TFTs or low-temperature
polycrystalline silicon TFTs (1-5).
To overcome the limitations of existing semiconductor materials, oxide semiconductors
having various composition ratios of zinc oxide (ZnO), indium gallium oxide, (S)-4-nitrostyrene
oxide, nitrosyl iodide, amorphous indium gallium zinc oxide (a-IGZO), and amorphous
indium zinc oxide have been studied for the channel layer in transistors. Notably,
a-IGZO has high charge mobility, maintains uniformity across large areas, and has
wide band gap energy. Furthermore, a-IGZO is widely used as a material for transparent
displays because it is easily deposited at low temperatures through physical vapor
deposition and sputtering deposition, and additionally has high light transmittance
(6-11).
To date, many studies have been conducted on electrical properties and transistor
performance, specifically examining various process conditions, such as insulating
layer thickness, channel layer thickness, annealing temperature, and surface treatment
of the channel layer (5,12-17). Recently, studies have been conducted to improve the electrical performance of oxide
TFTs through plasma treatment (18-21). Recent studies have shown that tetrafluoromethane (CF4) plasma treatment improves
performance by reducing contact resistance between the semiconductor channel layer
and the electrode, ultimately changing surface morphology (8,22-26).
In the case of IGZO, studies such as increasing the mobility of transistors through
oxygen plasma treatment and controlling internal oxygen vacancy have been conducted
(27-30).
Therefore, this study analyzes the effect of surface treatment on the performance
of TFTs according to the amount of oxygen injection by applying an oxygen plasma surface
treatment with differing amounts of oxygen injection into the a-IGZO channel layer
after the post annealing process. A device with a metal-insulator-metal (MIM) structure
of a silicon dioxide (SiO2) insulating layer and an a-IGZO oxide semiconductor between the Si gate electrode
and the aluminum (Al) source/drain electrode was fabricated. To observe the change
in electrical and surface area characteristics of the a-IGZO channel-layer oxide semiconductor
transistor based on oxygen plasma treatment with various amounts of oxygen injection,
a semiconductor parameter analyzer, and an atomic force microscope (AFM) were used
to identify the surface morphology of the transistor. In addition, the effects of
oxygen gas injection on current persistence and stability of a-IGZO channel layer
TFTs fabricated with oxygen plasma surface treatment were analyzed. Lastly, a basic
logic circuit was constructed, and a dynamic inverter test was conducted to evaluate
the applicability of the fabricated device as an active driving backplane display.
II. EXPERIMENT
Fig. 1 shows the structure of the a-IGZO channel layer TFT with the MIM structure. Heavily
doped n-type Si wafers were used as substrates and electrodes for gate contacts. A
thermal oxidation process applied a 100 nm thick SiO2 insulating film on the substrate. To remove the organic and inorganic impurities
from the surface, a sulfuric acid (H2SO4)-hydrogen peroxide (H2O2) mixture for cleaning (piranha etch, a.k.a. SPM) was applied for 20 min at a temperature
of 60 °C with the H2SO4 and H2O2 in a 3:1 ratio. The sample was soaked in deionized water, acetone, and an isopropyl
alcohol solution, and subsequently dried in a vacuum oven for one hour after 20 min
of ultra-sonication.
Fig. 1. Schematic representation of the bottom gated a-IGZO TFTs. In the device, Si
is used as a gate, SiO2 is used as a gate-insulator, a-IGZO is used as an active layer (a channel layer of
semiconductor), and Al is used as source and drain top electrodes.
Afterwards, sputtering was applied using an RF magnetron sputtering system to form
an IGZO channel layer on the SiO
2 insulating layer. A 1:1:1 ratio (In2O3:Ga2O3:ZnO) IGZO having a diameter of three
inches was used as the target, and the distance between the target and the substrate
was set to 8 cm. To remove impurities in the chamber, the initial vacuum setting in
the chamber was 3 × 10-6 torr or less, using a rotary pump and a turbo molecular pump,
and 30 standard cubic centimeters per minute (sccm) of argon gas was injected to maintain
a vacuum in the chamber at 1.5 × 10-2 torr. The substrate was rotated at 7 rpm for
uniform deposition of the thin film, and plasma was generated by applying 150 W of
RF power from the RF power generator to deposit a 50 nm a-IGZO channel layer for 6
min 40 s.
After depositing the a-IGZO channel layer through the sputtering process, a post annealing
process was performed to planarize the surface by reducing both the crystallization
of the a-IGZO channel-layer thin film and any defects present in the thin film. The
post annealing process was performed for one hour at a temperature of 350 °C under
atmospheric conditions. Then, oxygen plasma surface treatment was applied using a
gun-type plasma cell to analyze the surface area and electrical properties of the
a-IGZO thin film from differing amounts of oxygen injection. Treatment of the a-IGZO
channel layer was accomplished by applying 120 W of RF power from the RF generator
for three min. Gun-type plasma cells can provide a locally uniform treatment on the
target sample. As the gas collected in the crucible is ionized by the RF power applied
through the coil wound around the crucible, it is released toward the crucible nozzle,
and changes into a radical state to generate dense plasma. After the surface was treated
with oxygen plasma, Al source/drain electrodes with a channel length of 200 μm and
width of 2,000 μm were deposited using a DC magnetron sputtering system. To generate
vacuum plasma in the chamber, direct current at 150 W was applied for 10 min to deposit
a 100 nm Al source/drain electrode.
This study fabricated an a-IGZO channel layer TFT with no oxygen plasma surface treatment
after its post annealing as a control, and we further fabricated a-IGZO channel layer
TFTs with oxygen plasma surface treatment by injecting oxygen at 3 sccm, 6 sccm, or
9 sccm after the post annealing. Electrical performance and stability of the TFTs
were evaluated using a semiconductor parameter analyzer (Keithley 2636, Keithley Instruments
LLC) in a dark room at room temperature. To better understand how varying amounts
of oxygen plasma surface treatment on the surface of a-IGZO channel layers influenced
the surface microstructure of a-IGZO channel layers, AFM (ICON, Bruker Corporation)
was used. The current persistence and stability of the fabricated device was analyzed
through the retention current stability gate bias stress (GBS) tests using a semiconductor
parameter analyzer (Keithley 4200, Keithley Instruments). Finally, an inverter circuit
was configured to analyze the switching characteristics of the device to confirm its
application to backplane display devices.
III. RESULTS AND DISCUSSION
In this study, varying amounts of oxygen injection were used to improve the electrical
and surface performance of an a-IGZO channel layer TFT. This was investigated by applying
plasma surface treatment to the a-IGZO thin films with differing amounts of oxygen
injected at 3 sccm, 6 sccm, or 9 sccm after post annealing in the process of manufacturing
a-IGZO channel layer TFTs.
Fig. 2 shows the output characteristic curves of the as-deposited a-IGZO TFT and the a-IGZO
TFTs injected with 3 sccm, 6 sccm, and 9 sccm of oxygen for plasma surface treatment.
Ids is measured by sweeping 0.5 V increments starting from Vds of 0 V up to 30 V.
The gate bias voltage was applied in 10 V increments from 0 V to 30 V. In Fig. 2(b) and (c), which are the output characteristic curves of the a-IGZO TFTs injected with 3 sccm
and 6 sccm of oxygen, the output curves are saturated at higher currents. This is
due to the decrease in the oxygen vacancy in the a-IGZO channel layer and the decrease
in the defects causing trapping, resulting in increased charge mobility and improved
leakage current despite the decrease in the free electron concentration.
Fig. 2. Output characteristics of Ids-Vds curves at four different Vgs levels in TFTs
with a-IGZO channel layers treatedwith oxygen plasma with gun-type plasma cell based
on the amount of oxygen gas injected (a) as-deposited, (b) at 3 sccm, (c) at 6 sccm,
(d) at 9 sccm.
However, in
Fig. 2(d), which is the output characteristic curve of the a-IGZO TFT injected with 9 sccm
of oxygen, the output curve is saturated at significantly lower currents, compared
to
Fig. 2(a) to (c), and there is no difference, even if the gate bias voltage is applied differently.
This is due to the considerable reduction of free electron concentration caused by
excessive injection of oxygen, leading to a removed oxygen vacancy that is more than
necessary in the a-IGZO channel layer, and an increase in surface roughness. Oxygen
vacancy in the thin film generally acts as a donor of the carrier, and further serves
as a trap site to disturb carrier flow. Oxygen vacancy causes trap charges or leakage
currents, which greatly affect the reliability of transistors. However, a decrease
in oxygen vacancy can lead to reduced charge mobility and increased resistance due
to a decrease in carrier density. Therefore, it is important to maintain an appropriate
amount of oxygen vacancy for TFT performance
(31-38).
Fig. 3 shows the transfer characteristic curves of the as-deposited a-IGZO TFT and the a-IGZO
TFTs injected with 3 sccm, 6 sccm, or 9 sccm of oxygen for plasma surface treatment.
The graph measures Ids by sweeping 0.5 V increments from 0 V up to 30 V, and shows
Ids and square root of Ids values when Vds is fixed at 30 V.
Fig. 3. Transfer characteristics of Ids-Vgs curves and square root Ids-Vgs with Vds
= 30 V in TFTs with a-IGZO channel layers treated with oxygen plasma with gun-type
plasma cell based on the amount of oxygen gas injected (a) as-deposited, (b) at 3
sccm, (c) at 6 sccm, (d) at 9 sccm.
The electrical characteristics of the transistors were extracted based on the transfer
characteristic curves, and the results are summarized in
Table 1. The a-IGZO channel layer TFTs with oxygen plasma surface treatment demonstrated
an increase in charge mobility and current on-off ratios, and furthermore, showed
that subthreshold swing (S/S) values improved. In
Fig. 3(b) and (c), which are the transfer characteristic curves of the a-IGZO TFTs injected with 3
sccm and 6 sccm of oxygen, the drain off-current is lower than the transfer characteristic
curve of the as-deposited a-IGZO TFT in
Fig. 3(a). The charge mobility increased by more than 3.7 cm
2/Vs, and the current flashing ratio increased between tenfold and a hundredfold. The
subthreshold swing of the transmission curve is more linear, and the on-current voltage
shifted positively to near 0 V.
Table 1. Summary of electrical properties of a-IGZO channel layers treated with oxygen
plasma based on the amount of oxygen gas injected, including μsat, Ion/Ioff, Vth,
and S/S
|
Oxygen gas injected (sccm)
|
μsat (cm2/Vs)
|
Ion/Ioff
|
Vth (V)
|
S/S (V/dec)
|
Nit (cm-2)
|
|
as-deposited
|
10.6
|
1.3 × 107
|
2.8
|
0.9
|
4.5 × 1012
|
|
3
|
14.3
|
1.2 × 109
|
2.4
|
0.5
|
2.2 × 1012
|
|
6
|
14.4
|
1.1 × 108
|
4.5
|
0.7
|
3.3 × 1012
|
|
9
|
9.9
|
3.6 × 108
|
8.8
|
0.6
|
2.8 × 1012
|
Meanwhile, the improved electrical performance of 6 and 9 sccm sample is related to
stable S/S value, the relationship between the S/S and interface trap density (Nit)
is given in equation.
where q is the electronic charge, k the Boltzmann constant, and T the absolute temperature.
The calculated Nit values of a-IGZO TFTs from S/S are 4.5 × 1012, 2.2 × 1012, 3.3
× 1012 and 2.8 × 1012 cm−2, respectively. (as shown in
Table 1) These results demonstrate that the plasma irradiation with suitable oxygen could
reduce the Nit, which is helpful to reduce the electron capture behavior at trap sites
by the channel surface, thereby the electrical performance including mobility is improved.
For the a-IGZO TFT injected with 3 sccm of oxygen gas, the threshold voltage (Vth)
was significantly lower relative to the control, and the subthreshold swing improved
by about half compared to the TFT without plasma treatment. For the a-IGZO TFT injected
with 9 sccm of oxygen gas, the overall electrical characteristics of the transistor
degraded as shown in the output characteristic results of Fig. 2. The current flashing ratio and the S/S value improved, but the charge mobility was
reduced, the Vth was greatly increased, and the on-current voltage shifted negatively.
The charge mobility seemed to decrease due to the decrease in oxygen vacancy, which
is likely due to excessive injection of oxygen during the treatment.
Fig. 4 shows the results of the bias test from applying constant voltage to the gate to
evaluate the current continuity, stability, and reliability of the as-deposited a-IGZO
TFT and the injected a-IGZO TFTs. When Vgs and Vds were applied at 30 V each, the
retention current stability was measured for about 10 min by dividing the value of
Ids by the initial measurement value, Ids0. For the a-IGZO TFT with no oxygen plasma
surface treatment, the measured current value decreased sharply over time from the
initial value, and after about 10 min, Ids dropped to about 17 % of Ids0. On the other
hand, when the a-IGZO TFTs were injected with 3 sccm and 6 sccm of oxygen, the current
was maintained at about 80 % of Ids0 over time. It seems that oxygen plasma surface
treatment removes impurities and defects on the a-IGZO channel layer surface and reduces
oxygen vacancy and hydrocarbons to alleviate the reactivity to moisture and oxygen
in the atmosphere.
Fig. 4. Retention characteristic curves of TFTs with a-IGZO channel layers treated
with oxygen plasma with gun-type plasma cell based on the amount of oxygen gas injected
at 0 sccm, 3 sccm, 6 sccm, and 9 sccm when Vds = 30 V, and Vgs = 30 V.
Fig. 5 shows the results of the GBS test to evaluate the applicability and reliability of
fabricated a-IGZO channel layer TFTs for display devices. A negative bias voltage
stress (NBS) test was conducted to evaluate the turn-off state of the a-IGZO channel
layer TFTs injected with oxygen plasma surface treatment. The NBS test measured the
transfer characteristic curves when −20 V was applied to Vgs for 100 s, 200 s, and
300 s. Then, Vgs was swept from −20 V to 20 V at 0.5 V increments. Analyzing the change
in Ids over time when bias voltage was applied to Vgs, we found the transfer characteristic
curve shifted negatively. This result is exaggerated as the time for applying negative
bias voltage is extended. The results demonstrate that Ids for Vgs is constant, despite
the decrease in Vth, and the slope of the transmission characteristic curve is also
constant. This seems to be a phenomenon caused by the movement of the positive holes
in the a-IGZO channel layer toward the gate due to the negative voltage across the
gate. In the a-IGZO TFT injected with 3 sccm of oxygen, the characteristic curve strongly
shifted in the negative direction after applying a negative bias voltage. For the
a-IGZO TFT injected with 6 sccm of oxygen, the negative shift was attenuated, compared
to TFTs under other conditions.
Fig. 5. NBS stability of TFTs with a-IGZO channel layers treated with oxygen plasma
with gun-type plasma cell based on the amount of oxygen gas injected (a) as-deposited,
(b) at 3 sccm, (c) at 6 sccm, (d) at 9 sccm. The stressing conditions were Vgs = −20
V.
Fig. 6 shows the results of AFM measurements (size: 2 μm × 2 μm) of the surface morphology
of the as-deposited a-IGZO channel layer and the injected a-IGZO channel layers to
analyze the effect of plasma surface treatment on the a-IGZO channel layer. Furthermore,
the change in root mean square (RMS) value based on the amount of oxygen injection
is shown in
Fig. 6(e). As the amount of injected oxygen increased, the step difference of the surface of
the a-IGZO channel layer decreased.
Fig. 6(c) shows the measurement image of the a-IGZO channel layer injected with 6 sccm of oxygen,
with the smallest RMS value and surface curvature.
Fig. 6. AFM images on 2 μm × 2 μm a-IGZO channel layers treated with oxygen plasma
with gun-type plasma cell based on the amount of oxygen gas injected (a) as-deposited,
(b) at 3 sccm, (c) at 6 sccm, (d) at 9 sccm, (e) The change in RMS value depending
on the amount of oxygen gas injected.
However, when the a-IGZO channel layer was injected with 9 sccm of oxygen, as shown
in
Fig. 6(d), the step difference on the surface was reduced due to the oxygen plasma, but the
RMS value increased. Larger RMS values indicate that the roughness of the surface
increases. Increase in surface roughness causes trap charges that impede the movement
of electrons. When trap charges occur, leakage current is induced, which is the main
cause of reduced charge mobility in the transistor
(39-41).
Therefore, as the amount of oxygen injected during the oxygen plasma surface treatment
increases, the step difference on the surface of the a-IGZO channel layer decreases.
Accordingly, the greater the amount of oxygen injected, the better the electrical
performance of the transistor. However, when too much oxygen is injected, the surface
roughness of the a-IGZO channel layer can be increased, causing current leakage, which
will subsequently lower the charge mobility. Accordingly, the a-IGZO channel layer
TFT injected with 6 sccm of oxygen performed best, indicating that the optimal amount
of oxygen injection is 6 sccm.
Fig. 7(a) shows a logic circuit constructed using the a-IGZO channel layer TFT injected with
6 sccm oxygen. The circuit is composed of a power supply, an oscilloscope, and a function
generator. The load resistance of the circuit was 1 МΩ, and VDD was 5 V. When input
voltage (Vin) was 0 V, the gate was short-circuited so that current flowed to the
output voltage (Vout), resulting in 5 V output at Vout. When Vin was 5 V, the gate
was open so that current flowed to ground, resulting in 0 V at Vout. Fig. 7(b) shows the results of the dynamic test after a resistive load inverter configuration
using the same a-IGZO channel layer TFT. Vin was applied in the range −5 V to 5 V
with a frequency of 1 kHz, and Vout over time was measured. We determined that the
gate was open, and inverting occurred properly at about 4.5 V. Based on these findings,
we suggest oxygen plasma surface treatment is applicable to creating a backplane device
for active driving displays in the future.
Fig. 7. Dynamic inverter test results from configuring the logic circuit for an a-IGZO
channel layer TFT treated with oxygen plasma in which 6 sccm of oxygen gas was injected
(a) schematic diagrams, VDD = 5 V, and Vin = −5 to 5 V, (b) when input frequency =
1 kHz, VDD = 5 V, and Vpp = 10 V.