Title
Author(s)
Citation
Issued Date
URL
Rights
Fluorinated InGaZnO Thin-Film Transistor With HfLaO Gate
Dielectric
QIAN, L; Lai, PT
IEEE Electron Device Letters, 2014, vol. 35 n. 3, p. 363-365
2014
http://hdl.handle.net/10722/202928
IEEE Electron Device Letters. Copyright © IEEE
IEEE ELECTRON DEVICE LETTERS, VOL. 35, NO. 3, MARCH 2014
363
Fluorinated InGaZnO Thin-Film Transistor
With HfLaO Gate Dielectric
Ling Xuan Qian and Peter T. Lai, Senior Member, IEEE
Abstract— Fluorinated amorphous InGaZnO thin-film transistor with HfLaO gate dielectric has been studied by treating
InGaZnO film in a CHF3 /O2 plasma. The saturation carrier
mobility can be improved from 29.6 cm2 /V · s to as high as
39.8 cm2 /V · s. In addition, the passivation effect of the fluorination on the dominant donor-like traps at the InGaZnO/HfLaO
interface is observed, as reflected by suppression of hysteresis phenomenon and smaller subthreshold swing. Measurement
result of low-frequency noise further supports the improvement
in electrical properties by the plasma treatment.
Index Terms— Amorphous InGaZnO (a-IGZO), thin-film
transistor (TFT), HfLaO, fluorine, plasma.
I. I NTRODUCTION
N THE field of flat-panel displays (FPDs), thin-film transistors (TFTs) with high field-effect mobility (μFE ) are
required [1]. Accordingly, ZnO-based TFTs have been intensively investigated to replace conventional amorphous silicon
TFTs. Compared to polycrystalline ZnO (poly-ZnO) TFTs,
amorphous InGaZnO (a-IGZO) TFTs exhibit better spatial uniformity of device performance. In addition, the incorporation
of Ga ions, which act as carrier suppressors and network stabilizers in InZnO (IZO), is beneficial to achieving high carrier
mobility and excellent uniformity [2]. Hence, a-IGZO TFT
has been regarded as one of the most promising candidates
for the application in FPDs. Lee et al. reported a-IGZO TFT
with a high μFE of 28 cm2 /V · s using ZrO2 gate dielectric [3].
However, it had large subthreshold swing (SS, 0.56 V/dec) and
large threshold-voltage (VTH ) hysteresis (VH , 3 V), which
could slow down the switching and deteriorate the reliability
properties of the device respectively. In addition, Zan et al.
used nanometer dot doping to improve μFE to an average
value of 67.5 cm2 /V · s [4]. Nevertheless, if considering the
effective intrinsic channel length, the average intrinsic μFE is
33.8 cm2 /V · s. However, the SS of this device is as large as
0.92 V/dec, and a negative VTH of −2.94 V was achieved,
making the device unsuitable for switching applications.
a-IGZO TFT with Ta2 O5 gate dielectric achieved a high μFE of
61.5 cm2 /V · s, but also had a large SS (0.61 V/dec) [5]. It was
reported that the electrical characteristics of poly-ZnO TFTs
could be improved by fluorine implantation [6]. Unfortunately,
the improvements in mobility and SS were accompanied by a
change of operation mode from accumulation to depletion.
I
Manuscript received December 3, 2013; revised December 20, 2013;
accepted December 22, 2013. Date of publication January 24, 2014; date of
current version February 20, 2014. This work was supported by the University
Development Fund, Nanotechnology Research Institute of the University of
Hong Kong, under Grant 00600009. The review of this letter was arranged
by Editor A. Chin.
The authors are with the Department of Electrical and Electronic Engineering, the University of Hong Kong, Hong Kong (e-mail: laip@eee.hku.hk).
Color versions of one or more of the figures in this letter are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/LED.2013.2296895
Fig. 1.
Schematic diagrams of the device cross section: (a) during the
CHF3 /O2 plasma treatment; (b) a-IGZO TFT with HfLaO gate dielectric.
Furthermore, fluorine doping by spin coating of metalfluoride (ZnF2 and InF3 ) precursor aqueous solution was also
adopted to improve both the carrier mobility and stability of
IZO TFTs [7].
In this letter, the effects of fluorine incorporation in a-IGZO
on the characteristics of a-IGZO TFTs are studied for the first
time. Fluorine is introduced in a-IGZO by means of plasma
treatment, which is more compatible with device processing
on large-area substrate and also induces less damage to thin
film than implantation. Due to its superior properties, including
high dielectric constant, good thermal stability and low trap
density [8], high-k material HfLaO is selected as the gate
dielectric for the samples in this letter. Consequently, both high
saturation carrier mobility (μsat , 39.8 cm2 /V · s) and small SS
(0.21 V/dec) can simultaneously be obtained by the sample
with the plasma treatment. Moreover, this sample can operate
in enhancement mode with a small VH of −0.5 V.
II. E XPERIMENTAL D ETAILS
The schematic diagrams of the device cross section during
the CHF3 /O2 plasma treatment and after all the fabrication
processes are shown in Fig. 1. P-type (100) silicon acts as
both substrate and gate electrode. Firstly, deposition of a
40-nm HfLaO film was done by a sputtering system.
An annealing treatment at 400 °C in N2 for 10 min followed.
Subsequently, the sample received a deposition of a 60-nm
a-IGZO active layer through radio-frequency (RF) sputtering
in an Ar/O2 mixed ambient. Then, the sample received the
CHF3 /O2 plasma treatment with the flow rate of 10 sccm/
1 sccm at a RF power of 20 W for 3 min. The addition of O2
was to remove carbon and increase fluorine concentration by
suppressing CFx radicals [9], [10]. The control sample without
plasma treatment was also fabricated. After that, a lift-off
process was utilized to form the source/drain electrodes, which
were composed of 20-nm Ti and 80-nm Au. The channel
width (W) and length (L) were 100 μm and 20 μm respectively. Finally, both samples were annealed in a forming-gas
(N2 : H2 = 95 : 5) ambient at 350 °C for 20 min. In addition,
metal-insulator-semiconductor (MIS) capacitor was prepared
beside the transistor to monitor the gate-oxide capacitance per
unit area (Cox , equal to 0.202 μ F/cm2 in this letter). For each
condition, 12 devices were measured, and the device with the
median value of μsat was selected to represent the condition.
0741-3106 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
364
IEEE ELECTRON DEVICE LETTERS, VOL. 35, NO. 3, MARCH 2014
Fig. 2. SIMS depth profiles in the stacks of IGZO/HfLaO/Si thin films for:
(a) Zn, Hf, La and Si in the plasma-treated sample before the forming-gas
annealing; (b) F in the plasma-treated sample before & after the forming-gas
annealing.
III. R ESULTS AND D ISCUSSION
Fig. 2 exhibits the secondary ion-mass spectrometry (SIMS)
depth profile in the stack of IGZO/HfLaO/Si thin films. The
ordinate refers to the atom counts in an analysis area of
56.6 μm × 56.6 μm while the abscissa is about the acquisition
time which reflects the depth from the surface of a-IGZO. For
the plasma-treated sample before the forming-gas annealing,
it is apparent that fluorine has been successfully introduced
into a-IGZO by the plasma treatment, and majority of fluorine
atoms piles up near the surface of a-IGZO. The fluorine peak at
the IGZO/HfLaO interface should be due to its blocking effect
on the fluorine incorporation. After the forming-gas annealing,
fluorine atoms have diffused towards the a-IGZO/HfLaO interface, and a fraction of them has spread into HfLaO.
Fig. 3(a) shows the transfer characteristics of the a-IGZO
TFTs with or without CHF3 /O2 plasma treatment for a-IGZO:
1/2
drain current (ID ) vs. VGS , and ID vs. VGS at a drainto-source voltage (VDS ) of 5 V. The important electrical
parameters are extracted from Fig. 3 and listed in Table I.
Among them, μ sat and VTH are calculated from a linear fitting
1/2
to the plot of ID vs. VGS (using a VGS range of 1 V), which
is based on the I–V equation of field-effect transistor operating
in the saturation region:
ID = (μsat Cox W/2L) (VGS −VTH )2
(1)
The inset of Fig. 3(a) reveals μsat is gate-bias dependent,
which is consistent with the result reported in [11], and
accordingly the maximum value of μsat during the VGS sweeping is selected for comparison purpose [6], [11]. Compared
to the control sample (μsat = 29.6 cm2 /V · s), the plasma1/2
treated sample possesses a steeper slope of ID vs. VGS and
2
accordingly a higher μsat (39.8 cm /V · s). It is well known
that fluorine and oxygen have similar ionic radius, and the
Zn–F bonding is more stable than the Zn–O one. Therefore,
it is easy to cause the substitution of the oxygen atoms in the
lattice (OX
O ) of a-IGZO by the fluorine atoms derived from the
CHF3 /O2 plasma. Moreover, the difference in electrovalency
between oxygen ion (O2− ) and fluorine ion (F− ) can induce
the substitution of an oxygen ion by a fluorine ion to generate
a free electron [7]:
−
•
−
OX
O +F → FO + e
(2)
Therefore, the electron concentration in a-IGZO is increased,
resulting in a shift of the zero-gate-bias Femi level towards
Fig. 3.
Transfer (a) and output (b) characteristics of the a-IGZO TFTs.
TABLE I
E XTRACTED E LECTRICAL PARAMETERS OF IGZO TFTs
the conduction band. Accordingly, a larger fraction of the
acceptor-like traps in a-IGZO and/or at the a-IGZO/HfLaO
interface can be filled by the generated free electrons.
As a result of the filling of acceptor-like traps at/near the
a-IGZO/HfLaO interface, trap-induced scattering on channel electrons is suppressed, and thus carrier mobility is
increased [12]. However, even with an increased electron
concentration in a-IGZO, VTH is slightly increased from
2.5 V to 2.8 V by the plasma treatment, possibly due to
the introduction of negative fluorine ions in the gate oxide.
Fig. 3(b) displays the output characteristics of two samples.
Both samples clearly exhibit the n-type enhancement mode.
Moreover, ID increases linearly with VDS in the region of low
VDS , and current saturation can be observed in the region of
high VDS . Due to the increase in μsat , the on current (Ion ) at
VGS = 10 V and VDS = 5 V for the plasma-treated sample
(Ion = 1149 μA) is improved by over 20% compared to the
control sample (Ion = 934 μA). However, the off current (Ioff )
is also increased, which should be ascribed to the conductivity
increase of the ZnO-based material after fluorine incorporation
demonstrated by Equation (2) [6], [7]. As a whole, a slight
improvement of Ion /Ioff (from 9.0 × 105 to 9.2 × 105 ) can
still be achieved by the plasma treatment.
As shown in Fig. 4, both samples present a hysteresis
phenomenon in the counter-clockwise direction, which reveals
the dominant role of donor-like traps at the a-IGZO/HfLaO
interface. Nevertheless, the reduction of VH from −1.3 V to
QIAN AND LAI: FLUORINATED InGaZnO TFT WITH HfLaO GATE DIELECTRIC
Fig. 4. Transfer characteristics of the a-IGZO TFTs measured under the
forward and reverse sweepings.
365
It is found that μsat can be increased from 29.6 cm2 /V · s to as
high as 39.8 cm2 /V · s by the plasma treatment. Accordingly,
Ion is increased by over 20%. In addition, the passivation
effect of plasma treatment on the dominant donor-like traps
at the a-IGZO/HfLaO interface is also observed. As a result,
the hysteresis phenomenon is suppressed while SS is reduced.
Accordingly, both high saturation carrier mobility and small
SS (0.21 V/dec) can simultaneously be achieved by the
plasma-treated sample, which can also operate in enhancement
mode with a small VH of −0.5 V. The LFN measurement
result further supports the improvement in electrical properties.
In summary, fluorinated a-IGZO is a promising material for
making high-performance TFT used in the field of FPDs.
R EFERENCES
Fig. 5.
The plot of SiD /I2D versus frequency of the a-IGZO TFTs.
−0.5 V reveals that the counter-clockwise hysteresis can be
obviously suppressed by the CHF3 /O2 plasma treatment for
a-IGZO. Therefore, it is believed that the CHF3 /O2 plasma
treatment can effectively passivate the dominant donor-like
traps at the a-IGZO/HfLaO interface. A similar effect on
donor-like traps after fluorine incorporation in ZnO-based
material was reported [6]. In addition, the subthreshold region
shown in Fig. 3(a) is steeper for the plasma-treated sample,
and thus a smaller SS (0.21 V/dec) is obtained as compared
to the control sample (SS = 0.31 V/dec). It was reported that
the donor-like traps existing at the intermediate energy levels
in the bandgap of ZnO could influence the subthreshold swing
with a hump in the transfer curve of ZnO TFTs [13]. Moreover,
Raja et al. claimed that the unpredictable subthreshold swing
with a hump-shaped transfer curve in a-IGZO TFTs could
be suppressed through passivating the natural intrinsic donor
states [14]. Hence, the decrease of SS further verifies the
passivation effect on the donor-like traps at the a-IGZO/HfLaO
interface by the CHF3 /O2 plasma treatment.
As shown in Fig. 5, the normalized noise power spectral
density (SiD /I2D ) is measured at a fixed (VGS – VTH ) of 5.0 V
in the linear region (VDS = 1.0 V) for each sample. The lowfrequency noise (LFN) of a-IGZO TFTs can be influenced
by carrier number fluctuation (due to interface traps) and/or
mobility fluctuation (due to Coulomb scattering) [15]. Since
the LFN of the plasma-treated sample is smaller than that of
the control sample, the improvement in electrical properties
(smaller SS, smaller VH and higher μsat ) by the CHF3 /O2
plasma treatment is further supported.
IV. C ONCLUSION
In this letter, the effect of fluorine incorporation in a-IGZO
by the CHF3 /O2 plasma treatment on the characteristics of
a-IGZO TFT with HfLaO gate dielectric has been investigated.
[1] T. Kamiya, K. Nomura, and H. Hosono, “Present status of amorphous
In-Ga-Zn-O thin-film transistors,” Sci. Technol. Adv. Mater., vol. 11,
no. 4, pp. 044305-1–044305-23, Sep. 2010.
[2] J. S. Park, W. J. Maeng, H. S. Kim, et al., “Review of recent developments in amorphous oxide semiconductor thin-film transistor devices,”
Thin Solidi Films, vol. 520, no. 6, pp. 1679–1693, Jan. 2012.
[3] J. S. Lee, S. Chang, S. M. Koo, et al., “High-performance a-IGZO TFT
with ZrO2 gate dielectric fabricated at room temperature,” IEEE Electron
Device Lett., vol. 31, no. 3, pp. 225–227, Mar. 2010.
[4] H. W. Zan, W. W. Tsai, C. H. Chen, et al., “Effective mobility
enhancement by using nanometer dot doping in amorphous IGZO thinfilm transistors,” Adv. Mater., vol. 23, pp. 4237–4242, Aug. 2011.
[5] C. J. Chiu, S. P. Chang, and S. J. Chang, “High-performance a-IGZO
thin-film transistor using Ta2 O5 gate dielectric,” IEEE Electron Device
Lett., vol. 31, no. 11, pp. 1245–1247, Nov. 2010.
[6] Z. Ye and M. Wong, “Characteristics of thin-film transistors fabricated
on fluorinated zinc oxide,” IEEE Electron Device Lett., vol. 33, no. 4,
pp. 549–551, Apr. 2012.
[7] J. S. Seo, J. H. Jeon, Y. H. Hwang, et al., “Solution-processed flexible
fluorine-doped indium zinc oxide thin-film transistors fabricated on
plastic film at low temperature,” Sci. Rep., vol. 3, no. 2085, pp. 1–9,
Jun. 2013.
[8] N. C. Su, S. J. Wang, C. C. Huang, et al., “Low voltage-driven flexible
InGaZnO thin-film transistor with small subthreshold swing,” IEEE
Electron Device Lett., vol. 31, no. 7, pp. 680–682, Jul. 2010.
[9] K. Takahashi, M. Hori, and T. Goto, “Fluorocarbon radicals and surface
reactions in fluorocarbon high density etching plasma. I. O2 addition
to electron cyclotron resonance plasma employing CHF3 ,” J. Vac. Sci.
Technol. A, vol. 14, no. 4, pp. 2004–2010, Jul./Aug. 1996.
[10] B. Chapman, Glow Discharge Processes Sputtering and Plasma Etching,
New York, NY, USA: Wiley, 1980, pp. 314–316.
[11] K. H. Choi and H. K. Kim, “Correlation between Ti source/drain contact
and performance of InGaZnO-based thin film transistors,” Appl. Phys.
Lett., vol. 102, pp. 052103-1–052103-5, Feb. 2013.
[12] H. Q. Chiang, B. R. McFarlane, D. Hong, et al., “Processing effects
on the stability of amorphous indium gallium zinc oxide thin-film
transistors,” J. Non Cryst. Solids, vol. 354, nos. 19–25, pp. 2826–2830,
May 2008.
[13] M. Furuta, Y. Kamada, M. Kimura, et al., “Analysis of hump characteristics in thin film transistors with ZnO channels deposited by sputtering
at various oxygen partial pressures,” IEEE Electron Device Lett., vol. 31,
no. 11, pp. 1257–1259, Nov. 2010.
[14] J. Raja, K. Jang, N. Balaji, et al., “Negative gate-bias temperature
stability of N-doped InGaZnO active-layer thin-film transistors,” Appl.
Phys. Lett., vol. 102, pp. 083505-1–083505-4, Feb. 2013.
[15] H. S. Choi, S. Jeon, H. Kim, et al., “Verification of interface state
properties of a-InGaZnO thin-film transistors with SiN x and SiO2
gate dielectrics by low-frequency noise measurements,” IEEE Electron
Device Lett., vol. 32, no. 8, pp. 1083–1085, Aug. 2011.
[16] L. F. Lan and J. B. Peng, “High-performance indium-gallium-zinc oxide
thin-film transistors based on anodic aluminum oxide,” IEEE Trans.
Electron Device, vol. 58, no. 5, pp. 1452–1455, May 2011.
[17] L. Yuan, X. Zou, G. Fang, et al., “High-performance amorphous indium
gallium zinc oxide thin-film transistors with HfO x N y /HfO2 /HfO x N y
tristack gate dielectrics,” IEEE Electron Device Lett., vol. 32, no. 1,
pp. 42–44, Jan. 2011.