Solution processed F doped ZnO (ZnO:F) for thin film

Journal of

Materials Chemistry C View Article Online

Published on 17 December 2014. Downloaded by National University of Singapore on 19/03/2015 14:15:29.

PAPER

Cite this: J. Mater. Chem. C, 2015, 3, 1787

View Journal | View Issue

Solution processed F doped ZnO (ZnO:F) for thin film transistors and improved stability through codoping with alkali metals† Jingjing Chang,a Zhenhua Lin,b Ming Lin,c Chunxiang Zhu,b Jie Zhang*c and Jishan Wu*ac This paper reports solution-processed metal oxide semiconductor thin film transistors (TFTs), which were produced using fluorine (F) doped ZnO-based aqueous solution. It was found that doping F into the ZnO film improves thin film transparency and TFT performance with an ultrahigh on/off ratio of 108. The F doped ZnO TFT devices showed no improvement in shelf-life stability but improved bias stress stability. Moreover, when the ZnO:F was co-doped with alkali metals like Li, Na, and K, the co-doped ZnO TFT

Received 8th October 2014 Accepted 15th December 2014

devices exhibited much higher electron mobility, in comparison with ZnO or the ZnO:F TFTs. In addition, the co-doped TFT device exhibited excellent shelf-life stability and bias stress stability. These results

DOI: 10.1039/c4tc02257b

suggest that F and alkali metal co-doping can be a useful technique to produce more reliable and low

www.rsc.org/MaterialsC

temperature solution-processed ZnO semiconductors for TFTs and their applications.

Introduction Recently, metal oxide semiconductors have been intensively studied as semiconductors in thin lm transistors.1–3 Among them, zinc oxide (ZnO) as a wide-band-gap semiconductor has received a lot of attention in recent years due to its superior electrical performance and good optical transmittance compared to organic semiconductors.1–4 Despite their superior performance, most metal oxide semiconductors are processed using expensive vacuum based deposition techniques such as sputtering,5 pulsed laser deposition,6 and physical vapour deposition,7 which are incompatible with large area and exible plastic substrates due to their high cost and the high temperatures used in the process. Therefore, solution-processable ZnO precursors are desirable for the formation of ZnO thin lm because of the ease of control over their chemical composition/ stoichiometry and simple printing processes.8 Moreover, it is convenient and inexpensive to prepare the precursor solutions with the desired compositions simply by dissolving different ratios of metal sources and bases in an appropriate solvent. Recently, solution processed ZnO thin lms have been prepared using low process temperatures through aqueous a

Department of Chemistry, National University of Singapore, 3 Science Drive 3, 117543, Singapore. E-mail: [email protected]; Fax: +65 6779 1691; Tel: +65 6516 2677

b

Department of Electrical and Computer Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260, Singapore

c Institute of Materials Research and Engineering, A*STAR, 3 Research Link, Singapore 117602, Singapore. E-mail: [email protected]; [email protected]

† Electronic supplementary information (ESI) available: Transfer IV curves of ZnO:F and FESEM images of ZnO:F. See DOI: 10.1039/c4tc02257b

This journal is © The Royal Society of Chemistry 2015

precursors.9–14 However, the intrinsic mobility of ZnO is too low for practical applications; therefore, ZnO doped with various elements has been developed. In general, doping by introducing an electron donor or acceptor into the host crystal is a successful approach in thin-lm electronic/optoelectronic devices. Various elements, such as In, Sn, Al, and Ga, as the dopant in ZnO have been extensively studied.1–3 Among them, uorine (F), as an anionic dopant, has lower lattice distortion as its radius is closer to oxygen as compared with Sn, Ga and In dopants. Recent studies reveal that doping ZnO thin lm with low concentrations of F has signicantly improved the optical transmittance in the visible region and the mobility of charge carriers.15,16 It has been proved that the doped F atom substitutes an oxygen atom to generate a free electron or occupies an oxygen vacancy site to eliminate an electron trap site. The dual effects of the F dopant results in enhancing the TFT electron mobility and the gate bias stability.15 On the other hand, alkali metal as a good n-type cation doping source for ZnO has been recently reported with the device performance reaching up to 50 cm2 V1 s1 using spray pyrolysis methods and solution processed methods.17 The alkali metal dopant prefers interstitial sites to substitutional sites in the bulk ZnO thin lm, which is different from the F doping. The question was arising, whether we can combine these two different kinds of dopants to give a synergistic doping effect on the pristine ZnO in our study. In this study, we report a simple low temperature solutionprocessed F doped ZnO and co-doping ZnO:F with alkali metals for thin lm transistors. NH4F and MOH (M ¼ Li, Na, and K) as dopants were introduced into zinc ammonia complex solution. Since NH4+ and OH ions act as the solvent, the dopant will not introduce any carbon impurity that needs high temperature

J. Mater. Chem. C, 2015, 3, 1787–1793 | 1787

View Article Online

Journal of Materials Chemistry C


annealing for removal. By controlling the F(M)/Zn ratio in the precursor solution, high performance TFTs with high current on/off ratio have been obtained. We found that aer co-doping with alkali metals, the TFT charge carrier mobility and device stability were further improved.

Experimental The precursor was prepared by dissolving ZnO nanopowder (Sigma Aldrich) in an ammonium solution to form a 0.1 M (Zn(NH3)42+) complex. The solution was refrigerated for several hours to promote the powder dissolution. The F-doped ZnO precursor solution was prepared by adding NH4F with different molar ratios to the aqueous ZnO precursor solution. For the alkali metal co-doped ZnO precursor solutions, LiOH, NaOH, and KOH with different volume ratios in DI water (10 mg ml1) were added to the 10 mol% ratio of F doped ZnO. For device fabrication, a heavily doped p-type Si wafer (purchased from Silicon Quest International, Inc.) was used as the substrate and gate electrode, and 200 nm thermally grown SiO2 was used as the dielectric layer. Prior to spin coating the ZnO precursors, the Si/SiO2 substrates were cleaned with acetone, isopropyl alcohol (IPA), and de-ionized water, and then treated with Ar plasma to facilitate metal oxide thin lm formation. Aer that, the ZnO precursors were spin coated at 3000 rpm for 30 s to form the thin lm. Then, the substrates with the thin lms were annealed at 250 C or 300 C for 1 h in ambient air. Finally, Al source/drain electrodes (100 nm thick) were deposited on top of the ZnO thin lm with a shadow mask to dene the TFT channel length (L ¼ 100 mm) and width (W ¼ 500 mm or 1000 mm). The eld-effect mobility of the fabricated transistor was extracted using the following equation in the saturation regime from the gate sweep: ID ¼ W/(2L)Cim(VG VT)2, where ID is the drain current in the saturated regime, m is the eld-effect mobility, Ci is the capacitance per unit area of the gate dielectric layer (SiO2, 200 nm, Ci ¼ 17 nF cm2), VG and VT are gate voltage and threshold voltage, and W and L are channel width and length, respectively. The transistors were characterized using a Keithley 4200 parameter analyzer in a N2-lled glove box or under ambient conditions. For the stability test, each device was stored in ambient air with 40% relative humidity and measured over a period of 4-weeks. For the bias stress conditions, a gate bias voltage of 20 V was applied at the gate contact for 1000 s. The surface properties of the ZnO lm were characterized using 2D GADDS X-ray diffraction (XRD) (Bruker-AXS D8 DISCOVER GADDS). The transmittance spectra of the ZnO lms deposited onto glass substrates were characterized using a UV3600 Shimadzu UV-Vis-NIR spectrophotometer. The surface morphology and the roughness of the ZnO lm deposited on a Si substrate were observed using tapping-mode Atomic Force Microscopy (TM-AFM), which was performed on a Bruker ICONPKG AFM. The lm thickness and crystal structure of the ZnO and doped ZnO thin lms were investigated using the crosssection high-resolution transmission electron microscope (HRTEM) (Philips CM300 FEGTEM). X-ray photoelectron spectroscopy (XPS) experiments were carried out on the Escalab 220i using monochromatic Al-Ka (1486.6 eV) as the radiation source.

1788 | J. Mater. Chem. C, 2015, 3, 1787–1793

Paper

Time of ight secondary ion mass spectroscopy (TOF-SIMS) was performed under the following conditions: Bismuth gun, 25 keV, raster at 300 300 mm2 for 300 scans and positive polarity (good ion yield for Li ions, but poor ion yield for F). Scanning electron microscopy/energy dispersive X-ray spectroscopy (SEM/ EDX) images were obtained using a Zeiss Supra-40 SEM equipped with an in-lens detector and EDX detector.

Results and discussion F doping effect Most of the dopants for ZnO are metallic cationic dopants, while a few studies have been reported using halogen anionic dopants. Among the anionic dopants, F is the most effective dopant for a ZnO matrix. Firstly, the F ions have a similar size to ˚ O2: 1.38 A), ˚ which can avoid lattice distorO ions (F: 1.31 A, tions to a great extent. Secondly, when an F ion substitutes an oxygen ion, it can produce a free electron, which acts as a charge carrier to enhance the charge mobility of the TFTs. Furthermore, the F ion could occupy an oxygen vacancy site to eliminate an electron trap site, therefore improving the bias stress stability of the TFTs. Bottom-gate top-contact TFT architecture was employed in this study to evaluate the electrical properties of the F doped thin lms. Fig. 1 shows the typical transfer plots for TFTs fabricated on Si/SiO2 at different molar ratios of F doped ZnO annealed at 300 C. These TFTs exhibited an excellent n-type response with high charge carrier mobility and large Ion/Ioff ratio. As the doping concentration has important effects on the nal device performance, we investigated ZnO:F lms fabricated from the precursor solutions varying in F/Zn ratios. Fig. 1b shows the TFTs charge carrier mobility as a function of the F/Zn ratio for the ZnO:F based TFT devices. The detailed electrical parameters are summarized in Table 1. Fig. 1 clearly shows that when the F content was increased from 0% to 10%, the electron mobility in saturation increased from 1.3 to 2.6 cm2 V1 s1 with an excellent Ion/Ioff of 108 at a doping ratio of 10 mol%. The low conversion temperature, at 300 C, of metal amine complex to metal oxide lattice is desirable for ZnO:F semiconducting thin lm formation. For the optimum dopant concentration (F/Zn ¼ 10 mol%), the thin lms were formed at different annealing temperatures, and it was found that when the annealing temperature decreased from 300 C to 250 C, the mobility decreased from 2.6 to 0.4 cm2 V1 s1 due to less ZnO lattice formation (Fig. S1, ESI†). Previously, rst-principles calculation has been used to study the doping effect on the electrical properties.15 According to the calculation, it was found that F substitutes O (FO) in an energetically favorable process and could be dominant in ZnO:F. Due to this substitution process, the oxygen vacancies could be eliminated, which can be conrmed by the decrease in the lm absorption in the visible region caused by oxygen vacancies.18 When the F doping concentration increases to a certain high level, the interstitial F (Fi) can appear and act as a deep acceptor, and decrease the charge carrier mobility. The surface morphology of ZnO and the ZnO:F thin lms with different concentrations of F on Si substrates were


View Article Online


Paper


Fig. 1 (a) The transfer characteristics of the undoped ZnO and F doped ZnO TFTs and (b) the average charge carrier mobility of the undoped ZnO and F doped ZnO TFTs.

investigated by FESEM and TM-AFM. The FESEM images (Fig. S2, ESI†) and AFM images (Fig. 2) reveal that these ZnO:F lms are compact, dense, uniform, and fairly smooth. The root mean square (rms) values of surface roughness are 0.410 nm, 0.414 nm, 0.473 nm, 0.485 nm, 0.515 nm, and 0.767 nm on Si/SiO2 substrates for the different doping ratios. The surface roughness increased with increase in the doping level. In order to investigate the F doping effect on the ZnO thin lm microstructure properties, the ZnO:F lm microstructures on Si/SiO2 substrates were investigated by XRD. Fig. 3 shows the XRD patterns of ZnO:F lms with different molar ratios annealed at 300 C. The diffraction patterns exhibited substantial c-axis-oriented grain growth with (002) predominant orientation, consistent with the hexagonal wurtzite structure of ZnO. This is consistent with previously reported results as the grain growth tends to favor the low energy (002) surface.19 The intensity of the (002) diffraction peak observed at 2q ¼ 34.3 decreased slightly at a higher F content, indicating that ZnO crystallization was suppressed due to the scattering process. This is same as the charge carrier mobility decrease at a higher doping ratio of F. To gain further insight about the microstructural properties of the thin lms aer F doping HR-TEM was performed. Fig. 4 shows a series of HR-TEM images of the ZnO and ZnO:F cross section from lower to higher magnication. The polycrystalline domains with clear lattice fringes can be observed and a lattice spacing of around 0.275 nm could be deduced from these images. These results are in accordance with the

Table 1

Atomic force microscopy (AFM) images of the F-doped ZnO films with different molar ratios.

Fig. 2

Fig. 3 XRD patterns for ZnO and the F doped ZnO with different F doping concentrations.

XRD results. Aer F doping, the crystal lattice did not show any remarkable lattice distortion, which is the same as our previous discussion.

The electrical characteristics of solution processed ZnO:F TFTs as a function of the F doping content

F doping ratio

Annealing temperature ( C)

m [cm2 V1 s1] (mmax)

VT [V]

On/off

ZnO ZnO 1% 5% 10% 10% 15% 20%

250 300 300 300 250 300 300 300

0.21 (0.29) 1.3 (2.4) 1.5 (2.3) 1.9 (2.0) 0.40 (0.54) 2.6 (3.5) 2.3 (3.3) 1.4 (2.0)

35–40 28–32 27–30 24–27 25–28 26–30 25–29 26–30

106 106 107 107 107 107 107 107


J. Mater. Chem. C, 2015, 3, 1787–1793 | 1789

View Article Online



Cross-sectional HR-TEM images of the solution-processed ZnO films on SiO2 substrates with F doping levels of (a–c) 0.0 mol%, (d–f) 10 mol%.

Paper

Fig. 4

Fig. 6 XPS spectra of the O1s and C1s core level lines for the solutionprocessed ZnO films with F doping levels of (a) 0.0 mol%, (b) 10 mol%. The three peaks correspond to the oxide lattice without oxygen vacancies, the oxide lattice with oxygen vacancies, and hydroxide.

The electrical characteristics of solution processed alkali metal co-doped ZnO:F TFTs as a function of the doping content

Table 2

Doping source

Doping ratio

m [cm2 V1 s1] (mmax)

VT [V]

On/ off

ZnO:F Li

0% 1% 5% 10% 1% 5% 10% 1% 5% 10%

2.6 (3.5) 4.0 (4.7) 5.8 (6.6) 6.9 (8.7) 3.9 (4.2) 5.0 (5.5) 3.8 (4.4) 3.7 (4.0) 4.6 (5.2) 3.1 (3.4)

28–32 28–32 29–33 28–32 28–32 26–30 32–36 26–30 26–30 33–37

107 108 107 107 107 107 107 106 107 107

Na

K

Fig. 5 The optical transmittance spectrum of F-doped ZnO films on the glass substrate. (b) The (ahn)2 vs. photon energy plot of the Fdoped ZnO films with various doping concentrations.

The transparency of the ZnO:F thin lms with different doping concentrations were investigated by UV-Vis spectroscopy. As shown in Fig. 5, all of the present ZnO:F lms are colorless and highly optically transparent and exhibit an average transparency of above 90% in the visible light region for lms fabricated on glass substrates. From the transmittance spectra, the absorption coefficient as a function of photon energy is plotted in Fig. 5b, from which the optical band gap of each ZnO lm can be determined. The absorption coefficient a

1790 | J. Mater. Chem. C, 2015, 3, 1787–1793

can be calculated as follows: T ¼ A exp(ad),20 where T is the transmittance of the ZnO lm, A is a constant and approximate unity, and d is the lm thickness. The optical band gap of the ZnO lms was determined by applying the Tauc model21 and the Davis– Mott model22 in the high absorbance region: ahn ¼ D(hn Eg)n, where hn is the photon energy, Eg is the optical band gap, D is a constant, and n is equal to 1/2. The direct optical band gap of the ZnO thin lms with/without F doping was obtained by plotting (ahn)2 versus hn, as shown in Fig. 5b. The Eg value can be obtained by extrapolating the linear portion to the photon energy axis. It can be seen that the optical band gap values of the F doped ZnO lms increased from 3.421 eV to 3.423 eV (1%), 3.427 eV (5%), 3.436 eV (10%), 3.446 eV (15%) to 3.454 eV (20%). The widening of the optical band gap could be explained by the Burstein–Moss effect.23 XPS measurements were taken to analyse the chemical and structural evolution of the thin lm aer F doping. Fig. 6 shows the O1s XPS spectra for ZnO and the ZnO:F thin lm annealed at 300 C. The peaks centered at 529.9, 531.2, and 532.1 eV can be assigned to oxygen in the oxide lattice without vacancies, oxygen vacancies, and oxygen in hydroxide, respectively. The


View Article Online


Paper


Fig. 7 Transfer and output characteristics of the Li doped ZnO:F (a and d), Na doped ZnO:F (b and e), and K doped ZnO:F (c and f) TFTs.

XPS data indicates that the ZnO thin lm annealed at 300 C contains large amounts of oxide lattice with or without oxygen vacancies and small amounts of metal hydroxide. The ratios for these three peaks were 60.6%, 28.3%, and 11.1%, respectively. Upon thermal annealing at higher temperature, the metal hydroxide can gradually be converted into oxide via a thermally driven condensation process. In the ZnO thin lm transistors, the oxygen vacancies and metal hydroxide in the thin lm can act as the trap sites, which limit the charge carrier mobility. Aer F doping, it was found that the oxygen vacancy content decreased, as shown in Fig. 6, and the ratios changed to 62.4%,

24.8%, and 12.8%, respectively. This is because for the F doped ZnO thin lm, in which F can occupy the oxygen vacancy sites to reduce the electron trap sites, which increases the carrier mobility and carrier concentration. On the other hand, the F ion can form hydrogen bonding with hydroxyl groups (–OH) due to its high electronegativity. This can passivate the trap sites of the hydroxyl groups to further improve the mobility and stability. For the C1s core level ionization, the 284.6 eV feature is attributed to C–C and C–H moieties and is usually used as a reference peak. The higher binding energy peaks at 286.0 and 288.6 eV can be attributed to carbon oxide groups. These three C1s peaks

Fig. 8 Transfer characteristics of the ZnO (a), ZnO:F (b), and Li co-doped ZnO:F (c) TFTs after storage under ambient conditions (40% RH).


J. Mater. Chem. C, 2015, 3, 1787–1793 | 1791

View Article Online



Paper

Fig. 9 The bias stress stability of solution-processed ZnO (a), ZnO:F (b), and Li co-doped ZnO:F (c) TFTs before and after applying electrical stress with a gate bias of 20 V for 500 s.

(284.6, 286.0 and 288.6 eV) were attributed to unintentional surface contamination arising from (CHx)-like carbon and carbon oxides. Since the starting precursor was carbon-free, these peaks cannot origin from the carbon residues from the bulk thin lms. Therefore, the doping sources have little effect on the C1s spectra.

dominant (Li / LiZn + h+). This substitutional doping of Li neutralizes the carriers in the ZnO semiconductor and reduces the charge carrier mobility of the ZnO TFTs. The incorporation of Li and F in the ZnO thin lms could be observed by TOF-SIMS and SEM/EDX surface mapping (Fig. S3 and S4†).

The stability of the doped ZnO TFT devices Alkali metal doping effects In the case of alkali metal doped ZnO, it has been reported that an appropriate amount of Li dopant can improve the preferred orientation of the ZnO crystallites along the c-axis. In addition, since Li has a smaller ionic radii and ionic valence (68 pm, +1) than Zn (74 pm, +2), a small amount of Li can be interstitially incorporated within the ZnO matrix, which will generate the charge carriers and enhance the device performance of Li doped ZnO TFTs. The electrical properties of the solution processed alkali metal co-doped ZnO:F TFTs as a function of alkali metal doping concentration was investigated. As shown in Table 2 and Fig. 7, all of the alkali metal doped ZnO:F TFTs exhibited high eld effect mobility and on/off current ratio in the nchannel enhancement mode. Aer Li doping, the charge carrier mobility was improved signicantly compared to the mobility obtained from pristine ZnO and the ZnO:F TFTs. The highest mobility (8.7 cm2 V1 s1) was achieved at a Li concentration of 10 mol% for this ZnO:F system. The TFTs with the optimized alkali metal co-doped ZnO:F thin lm exhibited an excellent average mobility of 6.9 cm2 V1 s1, which is better than the un-doped ZnO TFT (1.3 cm2 V1 s1), ZnO:F TFT (2.6 cm2 V1 s1) and Li doped ZnO (5.3 cm2 V1 s1) (shown in Table S1†). Upon a further increase in the Li ratio, the performance degraded. This is similar to our previously used LiF doped device.11 For Na and K doped ZnO:F, the maximum mobility was around 5.0 cm2 V1 s1 and 4.6 cm2 V1 s1, respectively. The alkali metal doping mechanism, using Li as an example, has proved that Li in the ZnO matrix prefers the interstitial sites to substitutional sites because of low electron affinity. When Li occupies an interstitial site (Li / Lii+ + e), one electron is released to increase the carrier concentration, which results in high electron mobility. However, when the Li content was further increased, the substitutional site could be

1792 | J. Mater. Chem. C, 2015, 3, 1787–1793

The stability of the TFTs was evaluated through precursor, device shelf-life stabilities, and device bias stress stabilities. The stability of the precursor solution is important for solution based metal oxides. The TFTs fabricated from pristine and three month aged aqueous solutions were compared for the investigation on the precursor storage stability. The F doped ZnO solution is stable for several months without any precipitation being observed. However, the 3 month aged solution resulted in a signicant increase in the TFT off current (Fig. S5, ESI†) because of the charge state change of the ZnO,24 whereas for the alkali metal co-doped solution, the off current measured from the TFT made from 3 month aged solution showed almost no change with only a slight decrease in current. The good storage stability is benecial to the practical applications in industry. The device shelf life stability under ambient conditions is important for the metal oxide TFTs since pristine ZnO without encapsulation is easily oxidized in air, which affects the on/off ratio for the device. This poor switch is un-desirable for practical applications, e.g. driving active matrix display. In order to check the shelf-life stability of the thin lms, the TFT devices were tested upon exposing the transistor devices to the ambient air with controlled humidity of 30–40%. It was found that the pristine ZnO showed poor stability upon 4 weeks storage with the off current increased from 1012 A to 108 A. The ZnO:F TFTs were also unstable, even with a reduction of the active defect sites by the F dopant. While for alkali metal co-doped ZnO:F, the off current only slightly increased and was without much threshold voltage shi aer being stored under ambient conditions for 4 weeks (Fig. 8). The bias stress stability is another important parameter for metal oxide based devices. It is usually determined by the threshold voltage (Vth) shi with time under a constant bias; moreover, it is related with the density of electronic defects.


View Article Online


Paper

According to previously reported studies, the device stability can be improved by the incorporation of metal cations, which have a higher ionic valence and stronger oxygen affinity than Zn2+.25 These transition metals, such as Ga3+, Y3+, Hf4+, and Sn4+, can effectively suppress oxygen vacancy formation, which reduces the charge carrier concentration and improves the device stability.25 In our study, we have found F and alkali metal co-doping can also effectively enhance the device stability. The experiments were carried out by applying a constant gate voltage to the tested TFT devices. Three groups of TFT devices, ZnO, ZnO:F, and Li doped ZnO:F, were evaluated, and the results are shown in Fig. 9. It was found that the F doped ZnO exhibited a reduced threshold voltage shi with DVth ¼ 5.1 V compared to pristine ZnO (DVth ¼ 9.6 V). This is the result of the defect reduction due to trap lling by F substitution. However, the stress stability of the ZnO:F devices are still inferior to the devices with additional transition cation dopants. When co-doped with alkali metal in the ZnO:F thin lms, it was found that the bias stability was further improved to DVth ¼ 3.8 V. This is understood by the charge trapping model. The hydroxyl groups in the metal oxide systems act as the trapping source, which inuence the bias stress stability.26 The F dopant and additional alkali metal co-dopant could passivate the surface traps and hence increase the bias stress stability.

Conclusions In conclusion, we have reported solution-processed metal oxide semiconductor thin lm transistors (TFTs), which were produced using an F doped ZnO-based aqueous solution. It was found that F doped ZnO could improve thin lms transparency and enhance electrical performance. Moreover, the TFT devices showed poor shelf-life stability but improved bias stress stability. Most importantly, when co-doped with alkali metals like Li, Na, and K, the TFT devices revealed enhanced electron mobility than the non-doped or F doped ZnO TFTs. Furthermore, the device shelf-life stability and bias stress stability were improved signicantly compared to the nondoped or F doped devices. Our results suggest that F and alkali metal co-doping can be a useful technique to produce more reliable and low temperature solution-processed oxide semiconductor TFTs.

Acknowledgements This work was nancially supported by MOE Tier 2 grant (MOE2011-T2-2-130), IMRE Core Funding (IMRE/13-1C0205), A*STAR SERC TSRP grant (Grant #102 170 0137), and A*STAR IMRE/10-1P0508.

Notes and references 1 M. G. Kim, M. G. Kanatzidis, A. Facchetti and T. J. Marks, Nat. Mater., 2011, 10, 382. 2 K. Banger, Y. Yamashita, K. Mori, R. Peterson, T. Leedham, J. Rickard and H. Sirringhaus, Nat. Mater., 2011, 10, 45.



3 Y.-H. Kim, J.-S. Heo, T.-H. Kim, S. Park, M.-H. Yoon, J. Kim, M. S. Oh, G.-R. Yi, Y.-Y. Noh and S. K. Park, Nature, 2012, 489, 128. 4 U. Ozgur, Y. I. Alivov, C. Liu, A. Teke, M. A. Reshchikov, S. Dogan, V. Avrutin, S. J. Cho and H. Morkoc, J. Appl. Phys., 2005, 98, 041301. 5 P. F. Carcia, R. S. McLean, M. H. Reilly and G. Nunes, Appl. Phys. Lett., 2003, 82, 1117. 6 J. Nishii, F. M. Hossain, A. Takagi, T. Aita, K. Saikusa, Y. Ohmaki, I. Ohkubo, S. Kishimoto, A. Ohtomo, T. Fukumura, F. Matsukura, Y. Ohno, H. Koinuma, H. Ohno and M. Kawasaki, Jpn. J. Appl. Phys., 2003, 42, 347. 7 J. Jo, O. Seo, H. Choi and B. Lee, Appl. Phys. Express, 2008, 1, 041202. 8 J. J. Schneider, R. C. Hoffmann, J. Engstler, O. Soe, W. Jaegermann, A. Issanin and A. Klyszcz, Adv. Mater., 2008, 20, 3383. 9 S. T. Meyers, J. T. Anderson, C. M. Hung, J. Thompson, J. F. Wager and D. A. Keszler, J. Am. Chem. Soc., 2008, 130, 17603. 10 K. Song, J. Noh, T. Jun, Y. Jung, H. Y. Kang and J. Moon, Adv. Mater., 2010, 22, 4308. 11 J. Chang, Z. Lin, C. Zhu, C. Chi, J. Zhang and J. Wu, ACS Appl. Mater. Interfaces, 2013, 5, 6687. 12 S. Y. Park, B. J. Kim, K. Kim, M. S. Kang, K.-H. Lim, T. I. Lee, J. M. Myoung, H. K. Baik, J. H. Cho and Y. S. Kim, Adv. Mater., 2012, 24, 834. 13 J. Chang, K. L. Chang, C. Chi, J. Zhang and J. Wu, J. Mater. Chem. C, 2014, 2, 5397. 14 Z. Lin, J. Chang, C. Jiang, J. Zhang, J. Wu and C. Zhu, RSC Adv., 2014, 4, 6646. 15 B. Liu, M. Gu, X. Liu, S. Huang and C. Ni, Appl. Phys. Lett., 2010, 97, 122101. 16 J.-S. Seo, J.-H. Jeon, Y. H. Hwang, H. Park, M. Ryu, S.-H. K. Park and B.-S. Bae, Sci. Rep., 2013, 3, 2085. 17 G. Adamopoulos, S. Thomas, P. H. Wobkenberg, D. D. C. Bradley, M. A. McLachlan and T. D. Anthopoulos, Adv. Mater., 2011, 23, 1894. 18 S. Ilican, Y. Caglar, M. Caglar and F. Yakuphanoglu, Appl. Surf. Sci., 2008, 255, 2353. 19 P. O. Brien, T. Saeeda and J. Knowles, J. Mater. Chem., 1996, 6, 1135. 20 S. T. Tan, B. J. Chen, X. W. Sun and W. J. Fan, J. Appl. Phys., 2005, 98, 013505. 21 J. Tauc, Amorphous Liq. Semicond., 1974, 159–220. 22 E. A. Davis and N. F. Mott, Philos. Mag., 1970, 22, 903. 23 B. E. Sernelius, K. F. Berggren, Z. C. Jin, I. Hamberg and C. G. Granqvist, Phys. Rev. B: Condens. Matter Mater. Phys., 1988, 37, 10244. 24 S. M. Rozati, S. Moradi, S. Golshahi, R. Martins and E. Fortunato, Thin Solid Films, 2009, 518, 1279. 25 J.-S. Park, K. Kim, Y.-G. Park, Y.-G. Mo, H. D. Kim and J. K. Jeong, Adv. Mater., 2009, 21, 329. 26 S. Jeong, Y.-G. Ha, J. Moon, A. Facchetti and T. J. Marks, Adv. Mater., 2010, 22, 1346.

J. Mater. Chem. C, 2015, 3, 1787–1793 | 1793