You are on page 1of 6

Home Search Collections Journals About Contact us My IOPscience

Improved electrical stability of CdS thin film transistors through hydrogen-based thermal

treatments

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2014 Semicond. Sci. Technol. 29 085001

(http://iopscience.iop.org/0268-1242/29/8/085001)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 148.210.138.106
This content was downloaded on 11/06/2014 at 15:36

Please note that terms and conditions apply.


Semiconductor Science and Technology

Semicond. Sci. Technol. 29 (2014) 085001 (5pp) doi:10.1088/0268-1242/29/8/085001

Improved electrical stability of CdS thin film


transistors through hydrogen-based thermal
treatments
A L Salas-Villasenor1, I Mejia1, M Sotelo-Lerma2, Z B Guo3,
H N Alshareef3 and M A Quevedo-Lopez1,2
1
Department of Materials Science and Engineering, University of Texas at Dallas, Richardson, Texas,
75080, USA
2
Department of Polymer and Materials Science, University of Sonora, Hermosillo, Mexico
3
King Abdullah University of Science and Technology (KAUST), Thuwal 23955-6900, Saudi Arabia

E-mail: mquevedo@utdallas.edu

Received 28 February 2014, revised 24 April 2014


Accepted for publication 6 May 2014
Published 6 June 2014

Abstract
Thin film transistors (TFTs) with a bottom-gate configuration were fabricated using a
photolithography process with chemically bath deposited (CBD) cadmium sulfide (CdS) films as
the active channel. Thermal annealing in hydrogen was used to improve electrical stability and
performance of the resulting CdS TFTs. Hydrogen thermal treatments results in significant VT
instability (VT shift) improvement while increasing the Ion/Ioff ratio without degrading carrier
mobility. It is demonstrated that after annealing VT shift and Ion/Ioff improves from 10 V to 4.6 V
and from 105 to 109, respectively. Carrier mobility remains in the order of 14.5 cm2 V s−1. The
reduced VT shift and performance is attributed to a reduction in oxygen species in the CdS after
hydrogen annealing, as evaluated by Fourier transform infrared spectroscopy (FTIR).
Keywords: thin film transistors, electrical stability, solution-based
(Some figures may appear in colour only in the online journal)

Introduction stages of the deposition process using solution-based methods


[12]. Thus, it is essential to investigate the electrical stability
The ability to deposit semiconductor materials over large of solution-processed CdS TFTs to estimate device lifetime
areas using inexpensive deposition techniques with mobilities and further understand instability mechanisms in these devi-
higher than a:Si-H (0.1–1 cm2 V s−1) [1, 2] enables the pos- ces. In particular, in this paper we report CdS films deposited
sibility to fabricate high performance and low-cost thin film by chemical bath deposition (CBD) methods [9, 12].
transistors (TFTs) for large area and flexible electronics Electrical instability in TFTs is typically evaluated by
applications. In this sense, and opposite to organic TFTs, measuring the threshold voltage shift (ΔVT) that occurs when
TFTs fabricated with cadmium sulfide (CdS) will not degrade the device is subjected to constant voltage or drain current
when exposed to oxygen and moisture [3, 4]. Recently, much stress for a period of time [13]. Such instability is generally
progress has been made in several solution-based methods to ascribed to charge trapping [14] in either the gate dielectric,
fabricate CdS devices at low temperature; however, device semiconductor, or both [15]. Therefore, the physical
stability has been addressed only in a limited manner [5–11]. mechanisms responsible for charge trapping depended not
It is important to note that most solution-processed films, such only in the semiconductor and gate dielectric used in the TFT,
as CdS, might have impurities that could lead to device but also in the deposition method used to grow or deposit
reliability issues [7, 9]. In the case of CdS, some of these these films. For example, studies in TFTs with hydrogenated
impurities include hydroxides from the aqueous media, amorphous silicon (a:Si-H) as the semiconductor and silicon
unsatisfied sulfur or carbon bonds, and cyanamide formation nitride as the gate dielectric suggested that carriers are trapped
as a result of incomplete thiourea hydrolysis during the initial in the gate dielectric [14]. However, it was later demonstrated

0268-1242/14/085001+05$33.00 1 © 2014 IOP Publishing Ltd Printed in the UK


Semicond. Sci. Technol. 29 (2014) 085001 A L Salas-Villasenor et al

that bias stress also modifies the a-Si:H microstructure by


creating unsaturated valence states into which electrons are
trapped [14, 15]. Carriers trapped in these valence states
remained trapped after removing the gate bias and the VT shift
is persistent. To reverse this bias-induced threshold voltage
shift, annealing at temperature above 180 °C is
required [16, 17].
In contrast, bias stress-induced VT shift in organic TFTs
is usually not persistent, i.e., carriers trapped during bias
Figure 1. Schematic cross section of the CdS TFT fabricated for this
stress are released after removing the gate-source bias without
study.
the need of additional thermal annealing. A number of studies
have shown that the bias stress effect in organic TFTs is
relatively independent of the gate dielectric material. These (ALD) as the gate dielectric. The CdS active layer was
studies concluded that the bias stress-induced carrier trapping deposited immediately after the gate dielectric by CBD in a
in organic TFTs occurs only in the semiconductor or at the cadmium chloride/sodium citrate/thiourea solution system
semiconductor/dielectric interface [18]. However, distin- [8, 9]. For the CdS deposition by CBD, the substrates were
guishing between volume effects and trapping at the semi- immersed in the chemical bath solution at 70 °C. The
conductor/dielectric interface is complicated. On the other resulting CdS films have a thickness of approximately 70 nm.
hand, studies in CBD-deposited CdS films reported substantial After CdS deposition, 500 nm a hard mask of parylene-C
improvement in the electrical properties of CdCl2-thiourea (poly-p-xylylene) was deposited by chemical vapor deposi-
bath deposited CdS after thermal annealing [10, 19–21]. The tion (at room temperature). 100 nm thick Al film deposited by
improvement in the electrical properties is attributed to the e-beam evaporation was used as S/D electrodes. After the
change in larger grain size, lattice parameter and crystalline TFT structure is completed, the substrate is cleaved in three
structure along with a reduction in series resistance (contact pieces to maintain the same dielectric/semiconductor interface
resistance) of the CdS/electrode structure. Hence, the oppor- and compare three conditions: as-deposited, 150 °C and
tunity to control electrical properties of CBD-deposited CdS 300 °C annealing. Post-annealing of the devices was per-
by using a proper heat-treatment is of great interest. formed in a Thermco MiniBrute atmospheric furnace system
In particular, this paper evaluates the impact of thermal in forming gas (10%H2 + 90%N2) atmosphere at 150 °C and
annealing on the final electrical properties of TFT transistors 300 °C for one hour.
fabricated using CBD-deposited CdS films. The effect of Electrical characteristics of the TFTs before and after
thermal treatments and its correlation with the resulting TFT annealing were measured using current–voltage (I–V) mea-
performance and reliability is evaluated. The chemical and surements at room temperature in a Keithley 4200 semi-
structural correlation between the CdS thin films before and conductor characterization system. In order to have a
after post-deposition annealing treatment and the character- consistent methodology to assess the reliability of the TFTs, a
istics of TFTs is reported. Three different conditions were measurement procedure previously reported by our group was
evaluated: as-deposited, 150 °C and 300 °C annealing in used [22]. In this method, the (ID–VG) characteristics are
forming gas, (FG, 10%H2 + 90%N2) atmosphere for 1 h. The initially measured three consecutive times, followed by the
maximum annealing temperature was maintained at 300 °C output characteristic (ID–VD) measurements. Then, the same
due to potential CdS decomposition at higher temperatures. (ID–VG) sequence is used to obtain measurements three
The results indicate that improved device performance and additional ID–VG 4 measurements. This procedure allows
reliability is achieved after annealing in hydrogen as a con- detecting the bias stress effect on the device by monitoring the
sequence of reduced oxygen species in the films. Fourier- VT change between the 1st and 6th ID–VG electrical mea-
transform infrared spectroscopy (FTIR) shows that thermally surement. In the CdS devices reported here the threshold
treated CdS has a lower concentration of oxygen species than voltage reached a quasi-steady-state with no further changes
as-deposited samples, which results in an improvement in VT observed after the fourth ID–VG measurement.
stability and Ion/Ioff ratio without compromising device The CdS films before and after annealing were studied
performance. using scanning electron microscopy (SEM), crystalline
structure was studied using x-ray diffraction (XRD) and film
composition was determined using Fourier transform infrared
Materials and methods spectroscopy (FTIR).

The CdS TFTs used for the studies reported here were fab-
ricated on Si wafers with 500 nm of SiO2. The TFTs were Results and discussion
fabricated using an inverted staggered with bottom-gate, top-
contact structure using a photolithography process previously The XRD results for as-deposited, 150 and 300 °C annealed
reported by our group (figure 1) [10]. In this process, 10 nm/ films are shown in figure 2(a). Regardless of the annealing
100 nm Cr/Au thin film is used as the gate electrode with temperature all the films showed hexagonal phase with pre-
90 nm HfO2 films deposited by atomic layer deposition ferential orientation in the (002) direction with no significant

2
Semicond. Sci. Technol. 29 (2014) 085001 A L Salas-Villasenor et al

Figure 2. (a) XRD spectra and (b) SEM pictures for the CdS films annealed under different temperatures for one hour in forming gas. CdS
films show no change in crystalline structure or morphology after annealing.

difference between 150 and 300 °C annealing. Crystallite forming gas annealing (3600–3000 cm−1). The CBD method
size, as evaluated from the XRD results, is independent of the uses an alkaline aqueous solution and, therefore, the solution
annealing conditions. The evolution of the CdS surface contains OH−1 ions. From this fact, it is easy to explain the
morphology was also analyzed as a function of the annealing presence of OH−1 in the CdS films, as demonstrated by the
conditions. Figure 2(b) shows the SEM results of the CdS FTIR results in figure 3. Reduction of this OH−1 at the
films before and after annealing. No significant change in the interface is desired to minimize VT shift; however, at this
morphology or grain size was observed. This indicates that point we cannot identify if the OH−1 groups are located at the
neither the crystalline structure nor the film’s morphology is interface between the dielectric/CdS [12] and/or the bulk of
affected by the thermal annealing and that any changes the CdS films [23, 29–31]. We discuss this later in the paper.
observed in the resulting TFT behavior can be attributed to The peak at 2300 cm−1 in figure 3 is attributed to S-H
chemical changes in the CdS films as a result of the annealing stretching. This S-H bond is only observed in annealed CdS
process. films, suggesting incorporation of hydrogen species in the
To further study the impact of annealing temperature in CdS films after annealing. The hydrogen incorporation can be
the CdS films FTIR analyses were carried out using a Nicolet due to the passivation of oxygen containing species such as
6700 Thermo Fisher Scientific Inc. spectrophotometer in the sulfur bonds in the CdS film. The formation of SO2− 4 in sulfide
range of 4000–400 cm−1. Each spectrum was obtained using materials deposited by CBD has already been observed by
500 scans with 4 cm−1 resolution. The FTIR spectra of the other authors [24, 28, 32, 33]. Therefore, after annealing in H2
CdS films are shown in figure 3. The broad absorption band in rich environment, the reduction and passivation of this oxy-
the range from 3600–3000 cm−1 is attributed to OH−1 gen species is expected. The peak at about 2000 cm−1 is
stretching [23] and is consistent with other reports [24–28]. attributed to the carbon-nitrogen (C-N) stretching vibrations
figure 3 shows that the amount of OH−1 is reduced after of cyanamide [23, 29–31, 34]. Recently, we proposed the

3
Semicond. Sci. Technol. 29 (2014) 085001 A L Salas-Villasenor et al

the S-H bonds are indicative of effective defect reduction


during the annealing in forming gas.
Field-effect mobility and threshold voltage, for devices
before and after annealing were evaluated from a linear fitting
of a plot of VG versus the square root of ID using the fol-
lowing equation [37]:
W 2
ID = Ci μ ( VG − VT ) (1)
2L FET
where VG is the gate voltage and Ci is the capacitance per unit
area of the gate dielectric. Table 1 summarizes these
characteristics.
The improved Ion/Ioff ratio, shown in table 1, for annealed
devices is the result of the combination of the reduced Ioff
current, smaller subthreshold slope (SS) and the reduction in
VT that results in higher saturation current when the transistor
is in saturation regime. Oxygen species can act as donors in
CdS film [35]. Therefore, we attribute the reduction in Ioff to a
reduction in the CdS film conductivity. At this point, the
reduction in oxygen species is likely happening throughout
the entire CdS film, affecting the grain boundaries [10], the
semiconductor/contact interface [10, 38] and at the dielectric/
Figure 3. FTIR absorption spectra of CdS film for different thickness semiconductor interface [12, 39]. Subthreshold slope is also
showing three regions: 3600–3000 cm−1 attributed to OH-1 stretch- reduced from 1.21 to 0.21 V dec−1 due to the reduced defects
ing at the dielectric/CdS interface, 2200 cm−1 attributed to the C-N (traps) at the semiconductor/dielectric interface after anneal-
stretching vibrations of cyanamide in the bulk of the CdS film and
2400 cm−1 attributed to S-H in the bulk of CdS film.
ing. The VT for the first ID–VG measurement (before electrical
stress) was similar for all three conditions (1 V). However,
after electrical stress, reduced VT is observed for annealed
formation of cadmium cyanamide as a possible impurity devices when compared with not-annealed devices. This is
phase in the CBD grown CdS films due to the decomposition also consistent with the reduction of semiconductor/dielectric
reaction of thiourea during initial stages of the CBD [12]. The interface traps with the FG gas thermal treatment. More
presence of C-N does not affect the performance of CdS importantly, ΔVT was reduced from 10 V (no annealing) to
TFTs, as previously reported [32]. Nevertheless, the reduction 7.4 V and 4 V for films annealed at 150 and 300 °C, respec-
in the band intensity after annealing at 300 °C indicates that tively. The residual VT shift after annealing indicates that
the cadmium cyanamide might be reduced when annealing some defects are still present in the TFTs and longer
the films at 300 °C. annealing times might be necessary to eliminate oxygen
The CdS TFT transfer curves for as deposited, 150 °C, species still trapped at the grain boundaries resulting in
and 300 °C annealed for 1 h in forming gas (10%H2 + 90%N2, interface traps that affect VT and, potentially, ΔVT.
FG) are shown in figure 4. The two curves correspond to the
first and sixth ID–VG measurement cycles. Gate current for all
the cases remained in the range of 10 pA to 100 pA for the Conclusions
entire voltage sweep, which means the gate dielectric leakage
current is not affected by the thermal annealings. In general, In summary, the effect of thermal annealing on the electrical
the ID–VG showed a shift to positive voltages after con- stability of CdS TFTs was studied. Thermal annealing
secutive ID–VG measurements indicating negative charge improves the device characteristics due to the reduction in
trapping in the TFTs. Clearly, the VT shift (ΔVT) is greatly oxygen species and incorporation of S-H bonds in the CdS
reduced after annealing at 300 °C. No higher temperature films. Annealing results in excellent SS of 0.21 V dec−1 and
anneals were performed because annealing at higher tem- high Ion/Ioff ratio of 109, as well as high mobility of 14.6
perature (>300 °C) decomposes the CdS films and no TFT cm2 V s−1 and reduced ΔVT (4.6 V). This annealing is a
(channel modulation) behavior was observed [35]. Since simple and efficient method to improve reliability in
reduced ΔVT and better sub-threshold slope (SS) is observed CdS TFTs.
in devices annealed at 300 °C, then the reduction in the OH−
species and additional S-H bonding is occurring at the
interface between the dielectric and the semiconductor films Acknowledgements
(HfO2/CdS) [14, 15, 36]. The OH− species affect the devices
properties by either doping the CdS film [35] or acting as trap Authors would like to thank CONACyT project 158281, The
sites [23, 29–31] for the carriers resulting in increased ΔVT AFOSR project FA9550-10-1-0183, and COSMOS for par-
and SS. Therefore, the reduction in OH−1 and the increase in tially supporting this work.

4
Semicond. Sci. Technol. 29 (2014) 085001 A L Salas-Villasenor et al

Figure 4. Output and transfer characteristics for CdS TFT with HfO2 gate dielectric and Al S/D electrodes for (a) as deposited, (b) after
150 °C for 1 h in FG, and (c) 300 °C for 1 h in FG. Leakage current is in the order of 10–11 A.

Table 1. Electrical characteristics for CdS TFTs under various annealing conditions extracted after the annealing. VT before stress was about
1 V for all three conditions.
Annealing conditions μ (cm2 Vs−1) VT (V) SS (V dec−1) IOn/IOff ratio ΔVT (ΔV)
As deposited 14.8 ± 3.1 11.9 ± 1.2 1.21 ± 0.1 105 10.0 ± 1.7
150 °C 14.2 ± 3.6 8.7 ± 2.0 0.38 ± 0.05 107 7.5 ± 2.0
300 °C 14.6 ± 3.1 5.9 ± 0.5 0.21 ± 0.03 109 4.6 ± 0.5

References [19] Jayakrishnan R, Jaya P N, Annie K B, Kulkarni S K and


Pandey R K 1996 Semicond. Sci. Technol. 11 116
[20] Vigil O, Zelaya-Angel O and Rodrìguez Y 2000 Semicond. Sci.
[1] Street R A, Salleo A, Chabinyc M and Paul K 2004 J. Non- Technol. 15 259
Cryst. Solids 338–340 607 [21] Hiie J, Dedova T, Valdna V and Muska K 2006 Thin Solid
[2] Cerdeira A, Estrada M, Garcı ́a R, Ortiz-Conde A and Films 511–512 443
Garcı ́a Sánchez F J 2001 Solid-State Electron. 45 1077 [22] Garcia R, Mejia I, Molinar-Solis J E, Salas-Villasenor A L,
[3] Choi M H, Han S H, Lee S H, Choo D J, Jang J and Kwon S K Morales A, Garcia B, Quevedo-Lopez M A and Aleman M
2009 Organic Electron. 10 421 2013 Appl. Phys. Lett. 102 203505
[4] Han S H, Kim Y H, Lee S H, Choi M H, Jang J and Choo D J [23] Malinowska B, Rakib M and Durand G 2005 Sol. Energy
2008 Organic Electron. 9 1040 Mater. Sol. Cells 86 399
[5] Fortunato E, Correia N, Barquinha P, Pereira L, Goncalves G and [24] Ortega-Borges R and Lincot D 1993 J. Electrochem. Soc.
Martins R 2008 IEEE Electron Device Lett. 29 988 140 3464
[6] Chang Y J, Munsee C L, Herman G S, Wager J F, Mugdur P, [25] Hammond J S, Gaarenstroom S W and Winograd N 1975 Anal.
Lee D H and Chang C H 2005 Surf. Interface Anal. 37 398 Chem. 47 2193
[7] Khallaf H, Oladeji I O, Chai G and Chow L 2008 Thin Solid [26] Riggs W M, Wagner C D, Davis L E, Moulder J F and
Films 516 7306 Muilenberg G E 1979 Handbook X-Ray Photoelectron
[8] Salas-Villasenor A L, Mejia I, Hovarth J, Alshareef H N, Cha D K, Spectroscopy (St Paul, MN: Heyden and Son Ltd)
Ramirez-Bon R, Gnade B E and Quevedo-Lopez M A 2010 [27] Setty M S and Sinha A P B 1986 Thin Solid Films 144 7
Electrochem. Solid-State Lett. 13 H313 [28] El Maliki H, Bernède J C, Marsillac S, Pinel J, Castel X and
[9] Mejia I, Salas-Villasenor A L, Avendano-Bolivar A, Horvath J, Pouzet J 2003 Appl. Surf. Sci. 205 65
Stiegler H, Gnade B E and Quevedo-Lopez M A 2011 IEEE [29] Bhattacharya R N, Ramanathan K, Gedvilas L and Keyes B
Electron Device Lett. 32 1086 2005 J. Phys. Chem. Solids 66 1862
[10] Mejia I, Salas-Villasenor A L, Dongkyu C, Alshareef H N, [30] Kylner A and Wirde M 1997 Japanese J. Appl. Phys. 36 2167
Gnade B E and Quevedo-Lopez M A 2013 IEEE Trans. [31] Park W-D 2010 Trans. Electr. Electron. Mater. 11 170
Electron Devices 60 327 [32] Chaparro A M, Maffiotte C, Gutiérrez M T and Herrero J 2000
[11] Meth J S, Zane S G, Sharp K G and Agrawal S 2003 Thin Solid Thin Solid Films 358 22
Films 444 227 [33] Herrero J, Gutiérrez M T, Guillén C, Doña J M, Martı ́nez M A,
[12] Salas-Villasenor A L, Mejia I, Sotelo-Lerma M, Chaparro A M and Bayón R 2000 Thin Solid Films
Gnade B E and Quevedo-Lopez M A 2012 Appl. Phys. Lett. 361–362 28
101 262103 [34] Baldinozzi G, Malinowska B, Rakib M and Durand G 2002
[13] Gupta D, Seunghyup Y, Changhee L and Yongtaek H 2011 J. Mater. Chem. 12 268
IEEE Trans. Electron Devices 58 1995 [35] Goto F, Shirai K and Ichimura M 1998 Sol. Energy Mater. Sol.
[14] Powell M J 1983 Appl. Phys. Lett. 43 597 Cells 50 147
[15] Hepburn A R, Marshall J M, Main C, Powell M J and [36] Wang S D, Minari T, Miyadera T, Aoyagi Y and Tsukagoshi K
van Berkel C 1986 Phys. Rev. Lett. 56 2215 2008 Appl. Phys. Lett. 92 063305
[16] Libsch F R and Kanicki J 1993 Appl. Phys. Lett. 62 1286 [37] Sze S M 1981 Physics of Semiconductor Devices 2nd edn
[17] Kaneko Y, Sasano A and Tsukada T 1991 J. Appl. Phys. (New York: Wiley)
69 7301 [38] Yun D-J and Rhee S-W 2008 Organic Electron. 9 551
[18] Zschieschang U, Weitz R T, Kern K and Klauk H 2009 Appl. [39] Kumaki D, Umeda T and Tokito S 2008 Appl. Phys. Lett. 92
Phys. A 95 139 093309

You might also like