- Email: [email protected]
Fabrication of solution-processed nitrogen-doped niobium zinc tin oxide thin film transistors using ethanolamine additives
Journal of Alloys and Compounds 729 (2017) 370e378
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: http://www.elsevier.com/locate/jalcom
Fabrication of solution-processed nitrogen-doped niobium zinc tin oxide thin film transistors using ethanolamine additives Jiann-Shing Jeng*, Chi-Min Wu Department of Materials Science, National University of Tainan, Tainan 700, Taiwan
a r t i c l e i n f o
a b s t r a c t
Article history: Received 7 June 2017 Received in revised form 5 September 2017 Accepted 14 September 2017 Available online 18 September 2017
In this study, we fabricated the nitrogen-doped NbZnSnO channel layers by using the sol-gel method. Monoethanolamine (MEA) was used as a nitrogen additive. From the XPS results, the concentration of oxygen vacancies changes as a function of MEA/Nb ratio. A NbZnSnO film with MEA/Nb of 0.2 shows the lowest amount of oxygen vacancies. TFT electrical performance also shows a device with an MEA/Nb ratio of 0.2 possesses a high carrier mobility (7.4 cm2 V1s1) and good bias stress stability. In addition, we also investigated the effect of the aging time of precursor solution on the electrical characteristics of the TFT. After adding MEA, the annealing temperature of the NbZnSnO channels can be reduced, pertaining to the acceleration of the hydrolysis and condensation reaction. © 2017 Elsevier B.V. All rights reserved.
Keywords: Solution process Monoethanolamine Thin film transistor Nitrogen Oxide
1. Introduction In recent years, ZnSnO films have received much attention in the application for channels of thin film transistors (TFTs) [1]. The devices show a high carrier mobility in ambient air, a large on-to-off current ratio, and low production cost in comparison with Sibased TFTs [2,3]. In addition, solution-derived thin films have been widely investigated, which possesses many merits such as cost reduction, low-temperature fabrication, the simplicity of the process, the precise control of the stoichiometry, and high yield as compared with vapor deposition techniques. Therefore, solution processed ZnSnO has been utilized as a channel material in microelectronic. It is well known that the bias stress instability of TFTs is related to the amounts of oxygen vacancies in channel layers [4]. To improve the bias stress instability during device operation, some works have investigated ZnSnO channels added with various transition metal cations [5,6]. Our previous study [7] showed that introducing Nb into the ZnSnO channel materials could enhance the electrical performance of TFTs. Some literature [8,10] indicated that nitrogen anions can successfully substitute to O and eliminate the amounts
of oxygen vacancies in metal oxide channel layers because the ionic radii of N2 (0.129 nm) and N3 (0.132 nm) are similar to that of O2 (0.126 nm). The effective substitution of oxygen ions by nitrogen ions is based on the Hume-Rothery rules [9] where the ionic radii difference between the solute and solvent atoms should be no more than 15%. However, the N-doped metal oxide channel layers in the previous studies were fabricated by vacuum deposition techniques instead of solution processes [10,11]. In addition, monoethanolamine (MEA) was often used as a stabilizing agent for fabrication of solution-processed metal oxide channel layers [12,13]. In this study, we attempt to use MEA as a nitrogen additive for adding N into the host NbZnSnO channel layers to enhance the electrical performance of solution-processed NbZnSnO TFTs. We demonstrate the production of NbZnSnO device with MEA/Nb of 0.2 shows a mobility of 7.4 cm2 VV1s1, an on-to-off current ratio of 108 and a small subthreshold swing (~0.3 V/decade) by using a solution process technique. At the same time, we try to adjust the oxygen vacancies of the N-doped NbZnSnO TFTs using different amounts of MEA additives. In addition, the effect of MEA on the aging time of NbZnSnO precursor is also investigated. Finally, the correlation among the MEA-doping concentration, aging time, and device performance is discussed. 2. Material and methods
* Corresponding author. E-mail address: [email protected] (J.-S. Jeng). http://dx.doi.org/10.1016/j.jallcom.2017.09.151 0925-8388/© 2017 Elsevier B.V. All rights reserved.
NbZnSnO channel layers with adding different amounts of
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
371
Fig. 1. Cross-section TEM photographs of the NbZnSnO films with MEA/Nb molar ratios of (a) 0, (b) 0.2, and (3) 0.5 on SiO2/pþeSi. (d) GIAXRD patterns of 10-layers NbZnSnO channel layers doped with various MEA/Nb ratios.
monoethanolamine (MEA) were fabricated via a combined sol-gel and spin-coating method. From our previous study [7], an optimum Nb-doping concentration of solution-processed NbZnSnO TFTs is 3%, which possesses the best characteristics of the device. Therefore, the molar ratio of Nb: Zn: Sn in NbZnSnO precursors was chosen as 0.03:1:0.97. All reagents, anhydrous niobium chloride (NbCl5), hydrated Zinc nitrate (Zn(NO3)2$6H2O) and hydrated tin chloride dihydrate (SnCl2$2H2O) were dissolved in 20 ml ethylene glycol monomethyl ether (EGME) for preparing the NbZnSnO precursors. For MEA dopants, different amounts of MEA used as starting materials were weighed according to the molar ratios of MEA/Nb (i.e., 0, 0.1, 0.2, 0.3, and 0.5). After stirring for 20 h at room temperature, the NbZnSnO precursors with and without adding MEA achieved chemical homogeneity. Additionally, the precursor solutions were stirred for 2, 6, and 20 h, based on the color variation of NbZnSnO precursor, to proceed with the aging experiment by Fourier transform infrared absorption spectroscopy (FTIR) analysis and TFT characteristics. The NbZnSnO precursors were spin-coated onto a heavily doped p-type Si wafer with a 100 nm thick SiO2 layer, which functions as the gate electrode and gate dielectric. The spin-coated NbZnSnO active layers with and without MEA additives were rapidly baked at 110 C for 3 min and subsequently annealed on a heating plate at 430 C for 1 h under an air atmosphere. To verify the center of the plate surface controlled by the temperature controller, the temperature calibration of the heating plate was performed by a thermocouple. Finally, the 300-nm-thick Al source and drain electrodes were deposited on top of the active layer using an e-beam evaporation technique through a shadow mask to define transistor with the channel width (2000 mm) and length (100 mm). The thickness of the NbZnSnO film with and without MEA additives was determined using a transmission electron microscope (TEM, PHILIPS CM-200). To prepare the TEM-specimen, a focused ion beam instrument (FIB) was used to thin the samples for electron transparency. A protective layer of carbon was deposited on the NbZnSnO films with and without MEA additives
to prevent beam-induced damage during the milling process. The chemical compositions and oxygen vacancies were carried out with the x-ray photoelectron spectroscopy (XPS, JEOL JAMP9500F). A Hitachi U-2001 UV/Visible Spectrophotometer was used to measure the optical transmittance of the NbZnSnO composite films with and without adding MEA that were deposited on quartz glass substrates. Fourier transform infrared absorption spectroscopy (FTIR) was used to characterize the bonding configurations of NbZnSnO channels with and without MEA additives that were deposited on a KBr substrate. The currentevoltage characteristics of the NbZnSnO TFTs were measured under air ambient in a dark box using an Agilent 4156 C semiconductor parameter analyzer.
3. Results and discussion To determine the thickness of NbZnSnO films with and without adding monoethanolamine (MEA), the FIB-prepared thin samples were measured by transmission electron microscope (TEM). Fig. 1 presents the cross-section TEM images of solution-processed NbZnSnO films with different amounts of MEA (we designated it as NNZTO in Fig. 1). From Fig. 1(a) ~ 1(c), the thickness of NbZnSnO films without and with adding MEA is around 5~7 nm. The crystalline characteristics of all NbZnSnO films were performed by glancing incidence angles x-ray diffraction (GIAXRD). 10-layers NbZnSnO films were fabricated to obtain sufficient pattern intensity. All NbZnSnO films, no matter with or without adding MEA, are amorphous, as shown in Fig. 1(d). A diffraction peak at around 54 is due to the Si(100) plane of the Si substrate. The main idea of this study is to fabricate nitrogen-doped NbZnSnO films by using the sol-gel method and utilizing MEA additives. Therefore, the chemical composition and the chemical state of the NbZnSnO films with various ratios of MEA/Nb, prepared from precursor solution under the aging time of 20 h, were performed with X-Ray Photoelectron Spectroscopy (XPS). In order to avoid the surface contamination, sputter cleaning with 1 keV Ar
372
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
Fig. 2. XPS N1s spectra of the NbZnSnO thin films with various ratios of MEA/Nb.
ions was made for 4s to clean the NbZnSnO samples. For calibration of peak position, all the binding energies values were referenced to adventitious C1s at 284.8eV. Fig. 2 presents the XPS spectra of the N 1s peak for NbZnSnO films with various MEA/Nb ratios. From Fig. 2, N 1s XPS peak centered at 399.05 eV is found for NbZnSnO films
after MEA adding. Li et al. [14] and Chien et al. [15] reported that a similar binding energy of 399.32 and 399.2 eV associated with the ZneN bonds in the ZnO: N films. From Fig. 2, the N 1s XPS peaks present for all MEA-doped NbZnSnO films in comparison with undoped films, implying the effective nitrogen doping into NbZnSnO films using MEA additives. The O 1s photoelectron spectra of the NbZnSnO channel layers doped with various concentrations of MEA are present in Fig. 3. The XPS O 1s spectra of the NbZnSnO films with different amounts of MEA/Nb were deconvoluted into three components, including the low-energy component is associated with lattice oxygen in NbZnSnO films (530.62eV, OI), the medium-energy component is referred to oxygen nearby oxygen-deficient oxide of NbZnSnO films (532.2eV, OII), and the high-energy component is the oxygen in water molecules adsorbed on the NbZnSnO film surface (533.62eV, OIII) [16]. Fig. 4(a) shows the oxygen vacancy ratios [i.e., OII/(OI þ OII þ OIII)] of the NbZnSnO channel layers versus the MEA/Nb ratios. One can see that the oxygen vacancy ratios of the NbZnSnO films decrease from 0.17 to 0.1 for the MEA/Nb ratio increases from 0 to 0.2, and then the ratios increase to 0.175 and
Fig. 3. O 1s XPS spectra of the NbZnSnO thin films with MEA/Nb ratio of (a) 0, (b) 0.1, (c) 0.2, (d) 0.3, and (e) 0.5.
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
Fig. 4. (a) The oxygen vacancy ratios of NbZnSnO channel layers as a function of MEA/ Nb ratios. (b) FTIR transmission spectra of NbZnSnO films with various ratios of MEA/ Nb.
0.19 as the MEA/Nb ratio reaches 0.3 and 0.5, respectively. A decrease in oxygen vacancy ratios of the NbZnSnO films with adding small ratios of MEA/Nb (0.2) is due to the effective substitution of O by N. It is noted that the oxygen vacancy ratios of the NbZnSnO films increase with adding MEA content for the larger MEA/Nb ratios (>0.2). In order to understand the reason that a rise in oxygen vacancy ratios of NbZnSnO films with MEA/Nb of 0.3 and 0.5, the binding environment of the NbZnSnO films with different MEA/Nb ratios, prepared with 20 h aged precursor solution, was measured by Fourier transform infrared spectroscopy (FTIR). To obtain the information of the CeH related bonds, all NbZnSnO films with different MEA/Nb ratios were spun on KBr substrates and further baking at 110 C for 3 min to proceed the FTIR analysis. Fig. 4(b) present the FTIR transmission spectra of NbZnSnO films with different amounts of MEA. Two strong bands are observed in the IR spectrum of NbZnSnO films without adding MEA, one at 1336 cm1 and the second band at 1375 cm1, which are due to an interaction of nitrate with metallic cation [17]. In addition, the nitrate related bands (at 1336 and 1375 cm1) eliminated in NbZnSnO films with adding MEA. The OeH stretching vibration (at 3000e3500 cm1) and eOH group (at around 1620 cm1) are related to the hydroxyl group in NbZnSnO film. OH-related bands in IR spectrum of undoped NbZnSnO films may be attributed to the vibrations of MOH (where M denotes metals, including Nb, Zn, and Sn in this paper) or MeOHeM groups. By contrast, asymmetric CeH related bands at around 2850 and 2924 cm1 are found in NbZnSnO films with MEA/Nb ratios of 0.3 and 0.5, which can be assigned to the CeH related bands of the hydrocarbon byproducts. This result indicates that there is a suitable MEA-doping concentration in NbZnSnO films. When MEA-doping concentration is larger than the suitable value, the CeH related bonds are present. After annealing at 430 C, the hydrocarbon byproducts of NbZnSnO films with MEA/ Nb ratios of 0.3 and 0.5 decompose and result in the formation of
373
Fig. 5. (a) UVevis transmittance spectra of NbZnSnO films with different ratios of MEA/Nb (b) Absorption coefficient of NbZnSnO films with different contents of MEA/ Nb varies as a function of the photon energy (hn).
oxygen vacancy, leading to the higher amounts of oxygen vacancy ratios of NbZnSnO films with MEA/Nb ratios of 0.3 and 0.5, as shown in Fig. 4(a). The transmittance of NbZnSnO films with various MEA concentrations, prepared from solution aged for 20 h, was measured by
Fig. 6. (a) IDVG transfer characteristics of NbZnSnO TFTs using NbZnSnO channel layers with various MEA/Nb ratios, prepared from precursor solution aged for 20 h. (b) Mobility, threshold voltage (Vth), and subthreshold swing (S$S.) as functions of MEA/ Nb ratios.
374
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
a spectrophotometer. To obtain reasonable transmittance results, all NbZnSnO films were spun 10 times on quartz glass to achieve sufficient thickness. Fig. 5(a) shows the transmittance spectra of 10layers NbZnSnO films with various MEA/Nb ratios. It can be seen that all NbZnSnO films show good optical transparency (>80%) in the 400 nme700 nm range (visible region) from the transmittance spectra. After adding MEA, the transmittance increases for all MEAadded NbZnSnO films. By contrast, the absorption coefficients of NbZnSnO films with MEA additives are smaller than that of NbZnSnO films without MEA additives (see Fig. 5(b)), implying that the MEA dopants can eliminate the subgap absorption of NbZnSnO films. In order to realize the electrical characteristics of NbZnSnO TFTs, NbZnSnO films with different MEA/Nb ratios, prepared from precursor solution under the aging time of 20 h, were used as channel layers. We have fabricated all devices simultaneously at the same batch in order to avoid the susceptibility to the humidity, temperature and other environmental factors that will affect the device performance. Fig. 6(a) presents the transfer characteristics of NbZnSnO TFTs with and without adding MEA, prepared from precursor solution under the aging time of 20 h. The on/off current ratio of all the TFTs is larger than 107. Fig. 6(b) shows the carrier mobility (m), the threshold voltage (Vth), and
subthreshold swing (S$S.) as functions of MEA/Nb ratios. The carrier mobility and the threshold voltage in the saturation region of device operation for the NbZnSnO TFTs with and without adding MEA were extracted from the slope and intercept of an extrapolated line for the (ID)1/2-VG curve, respectively. The mobility can be overestimated in TFTs with the non-patterned channel layers because of the fringing currents around the peripheral region of non-precise active layer [18]. The subthreshold swing (S.S.) is defined by the inverse slope of the log (ID) vs. VG characteristic in the subthreshold region. The mobility of the NbZnSnO device increases with increasing the MEA/Nb ratio to 0.2 (m~7.4 cm2V1s1) and then decreases with increasing the MEA/Nb ratio to 0.5 (m~1.3 cm2V1s1), as shown in Fig. 6(b). The S.S. curve is a reversed trend of that of the mobility trend for the NbZnSnO TFTs with different MEA/Nb ratios. The lowest value of S.S. was observed to be 0.3 V/decade for the device with MEA/Nb ratio of 0.2. In addition, the Vth of the TFTs increases with increasing the MEA concentrations, as shown in Fig. 6(b). N ions were considered as p-type dopants for ZnO-based films [19,20]. Also, literature [21,22] suggested that nitrogen additives in metal oxide can reduce the carrier concentration, resulting in the positive shift of Vth. Fig. 7 shows the drain current versus drain to source voltage (ID-VD) output characteristics of the NbZnSnO TFTs
Fig. 7. The output characteristics (ID-VD) curves of the NbZnSnO TFTs with different MEA/Nb ratios.
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
Fig. 8. The transfer characteristics of TFTs using NbZnSnO channel layers with different MEA/Nb ratios, prepared from precursor solution aged for (a) 2 and (b) 6 h.
with and without adding MEA. All TFT devices exhibit typical normally-off type n-channel behavior. We have also studied the effects of stirring time (aging time) on the electrical performance of solution processed NbZnSnO TFTs with MEA/Nb ratios of 0, 0.2, and 0.5. The NbZnSnO-based presolutions were stirred with a magnetic stirrer for 2, 6, and 20 h at room temperature and then spun on Si substrate with SiO2 to be fabricated to form TFTs (as stated in the experimental section). Fig. 8 (a) and 8 (b) present the transfer characteristics of TFTs using NbZnSnO channel layers with different MEA/Nb concentrations, prepared from precursor solution aged for 2 and 6 h, respectively. As can be seen, the transfer characteristics of the same devices using channels prepared with 20 h aged precursor solution are shown in Fig. 6(a). The extracted TFT device parameters were listed in Table 1. From Fig. 8(a), the device using
375
channels without adding MEA, prepared with 2 h aged precursor, can not operate whereas the devices with MEA/Nb of 0.2 and 0.5 under the same aging time can successfully operate. Compared with the device without any MEA additive, an improvement of the TFT performance using NbZnSnO channels prepared with 2 h aged precursor solution is observed when the MEA/Nb ratio increases from 0.2 to 0.5. On the other hand, after 20 h aging of the precursors, 280 C and 380 C annealed NbZnSnO devices with MEA/Nb ratio of 0.2 show better performance such as higher mobility and lower subthreshold swing than that of device without and with adding MEA/Nb ratio of 0.5 under the same annealing temperature, as shown in Table 1. Literature [23,24] indicated MEA can improve the bonding ability of metal ion in precursor and decrease the formation energy barrier of metal oxide. From FTIR spectra in Fig. 4(b), as compared with undoped NbZnSnO film prepared with 20 h aged precursor solution, the hydroxyl groups (at around 1620 and 3400 cm1) in NbZnSnO film disappear for NbZnSnO film after adding MEA, implying the MEA can enhance the reaction of the presolutions and reduce the annealing temperature of NbZnSnO channels. As the aging time of the precursor increases to 6 h, a device without adding MEA shows the typical transfer characteristics of the TFTs measured at VD ¼ 40 V, as shown in Fig. 8(b). After increasing the aging time of the precursor from 6 to 20 h, the mobility of 430 C annealed undoped device improves from 0.61±0.09 to 4.81±0.17 cm2V1s1, as shown in Table 1. The mobility of a device with MEA/Nb ratio of 0.2 also increases with increasing the aging time of the presolutions. It is noted that TFTs with MEA/Nb ratio of 0.5 possesses a mobility of 6.79±0.55 cm2V1s1 after 2 h aging of the precursors and then deteriorates to 2.32±0.21 cm2V1s1 after aging of precursor solution for 6 h. After 20 h aging of the precursor solution, a continuous decrease in mobility of device with MEA/Nb ratio of 0.5 is observed (see Table 1). In order to realize the relation between the aging time of the precursors and electrical characteristics of TFTs, FTIR analysis was proceeded to understand the bonding environment of the NbZnSnO channels prepared from precursor solution under different aging time. Fig. 9 presents the FTIR spectra of NbZnSnO channel layers with different MEA concentrations, prepared from precursor solution under various aging time. From Fig. 9(a), the transmittance band found at 623 cm1 is assigned as stretching vibration of SneO [25]. In addition, the transmittance bands for undoped NbZnSnO channel layer at 1055 and 1375 cm1 belong to the NO stretch band of a free nitrate group and the NO2 stretching mode of a free nitrate ion, respectively [17]. After increasing the
Table 1 Comparison of the electrical characteristics measured in the air including mFE, Vth, S.S., and Ion/Ioff ratio for solution-processed NbZnSnO TFTs with different MEA/Nb additives under various fabrication conditions. Errors indicate the standard deviation of multiple independent devices. Devices MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb
Aging time/annealing tem. of of of of of of of of of of of of of of of
0 0.2 0.5 0 0.2 0.5 0 0.2 0.5 0 0.2 0.5 0 0.2 0.5
aging aging aging aging aging aging aging aging aging aging aging aging aging aging aging
2 h/430 C 2 h/430 C 2 h/430 C 6 h/430 C 6 h/430 C 6 h/430 C 20 h/430 C 20 h/430 C 20 h/430 C 20 h/380 C 20 h/380 C 20 h/380 C 20 h/280 C 20 h/280 C 20 h/280 C
mFE (cm2V1s1)
Vth (V)
S.S.(V/dec.)
Ion/Ioff ratio
e 0.91 ± 0.02 6.79 ± 0.55 0.61 ± 0.09 1.16 ± 0.16 2.32 ± 0.21 4.81 ± 0.17 7.41 ± 0.51 1.32 ± 0.12 1.35 ± 0.11 1.46 ± 0.24 0.056 ± 0.012 0.003 ± 0.001 0.08 ± 0.01 0.02 ± 0.01
e 9.0 8.5 7.0 9.0 9.0 6.0 7.0 8.5 3.5 6.0 6.0 3.0 5.0 8.0
e 0.94 1.25 2.34 0.17 0.69 0.88 0.30 0.90 0.96 0.38 0.40 0.37 0.20 0.30
e ~107 ~108 ~106 ~107 ~108 ~107 ~108 ~107 ~106 ~108 ~106 ~106 ~106 ~105
± ± ± ± ± ± ± ± ± ± ± ± ± ±
0.8 0.5 0.7 0.8 0.5 0.7 0.4 0.5 0.7 0.9 0.8 0.7 0.7 0.7
± ± ± ± ± ± ± ± ± ± ± ± ± ±
0.31 0.11 0.30 0.31 0.11 0.12 0.08 0.20 0.10 0.10 0.10 0.10 0.08 0.25
376
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
Fig. 9. FTIR transmission spectra of TFTs using NbZnSnO channel layers with (a) MEA/ Nb of 0, (b) MEA/Nb of 0.2, and (c) MEA/Nb of 0.5, prepared from precursor solution under different aging time.
aging time of the precursor to 6 h, the NO2 stretching band is split into two bands at 1336 and 1375 cm1, which is due to an interaction of nitrate with metallic cation [17] (i.e., the coordinated nitrate groups), as shown in Fig. 9(a). The coordinated nitrate band tends to become more apparent after aging for 20 h, implying that the hydrolysis and condensation reaction becomes more complete (i.e., more metaloxide formation) as the aging time of the precursor solution increases. The broad concave at 3400 cm1 and a transmittance peak at 1620 cm1 in NbZnSnO film, which are related to the hydroxyl group in NbZnSnO film. The hydroxyl group is still present in the NbZnSnO film without adding MEA even though precursors persist for a long-term aging of 20 h, as shown in Fig. 9(a). After aging for 20 h, the three nitrate-related bands (i.e., 1055, 1336, and 1375 cm1) of NbZnSnO films with MEA/Nb ratio of 0.2 disappear as compared to undoped NbZnSnO films, as shown in Fig. 9(b). This result indicates the hydrolysis and condensation reaction of NZTO films with an MEA/Nb ratio of 0.2
is complete under longer aging time. In the IR spectral range less than 1000 cm1 (see Fig. 9(b) and (c)), NbZnSnO samples reveal several transmittance bands. A band at around 806 cm1 is assigned to CeO out of plane bending mode of MEA. In addition, the vibrational bands at around 955 cm1 and 1018 cm1 are due to the CeNeH out-of-plane wagging and CeNH2 twisting of MEA [26] and alcohol CeO stretching vibrations [27], respectively. The FTIR spectra of NbZnSnO films with an MEA/Nb ratio of 0.5 do not show any nitrate-related groups after 2 h aging of the precursor solution, indicating an aging time of 2 h can lead to a complete hydrolysis and condensation reaction, as seen in Fig. 9(c). After increasing the aging time of the precursor solution, symmetric CeH stretching vibration (2924 cm1) was observed in the FTIR spectra of NbZnSnO films with an MEA/Nb ratio of 0.5. Adding MEA in NbZnSnO solution can increase the pH of the presolution due to the amine [28], which should promote the formation of NbZnSnO films. After increasing MEA additives up to MEA/Nb of 0.5, the enhanced formation of NbZnSnO films occurs under the short aging time of 2 h. With increasing the aging time of the precursor solution, hydrocarbon byproducts produced from the condensation and hydrolysis reactions of the precursors, which is accompanied by a main M-O-M product. This result leads to the occurrence of CeH stretching vibration in FTIR spectra of NbZnSnO TFT with an MEA/Nb ratio of 0.5. To investigate the electrical stability of the TFTs, a positive gate bias (PBS) of þ20 V was applied to the gate electrode of the devices and the source and drain electrodes were grounded for time periods ranging from 0 to 3600 s. The transfer curves of the TFTs using NbZnSnO channel layers with different MEA/Nb concentrations, prepared from precursor solution aged for 20 h, were measured under the dark in the air and stress duration of 3600 s at VD ¼ 40 V for PBS test. The measured results were shown in Fig. 10. All ID-VG curves of the NbZnSnO devices shift positively after PBS. The shift values of the threshold voltage (DVth) are relative to the pristine device. Fig. 11(a) presents the change in DVth as a function of PBS time for the NbZnSnO devices with various MEA/Nb ratios. As compared to other devices, a device with MEA/Nb ratio of 0.2 exhibits a smaller positive change in DVth after PBS for 3600 s. Additionally, a possible cause of electrical instability could be related to the presence of oxygen vacancies in the channel layer. Furthermore, the inset in Fig. 10 shows the transfer ID-VG curve of NbZnSnO films with and without adding MEA under negative gate bias (NBS, VG ¼ 20 V) and stress duration of 3600 s for VD ¼ 40 V (the source and drain electrodes were grounded). From Fig. 11(b), a much smaller negative change in DVth is found in our MEA doped NbZnSnO devices regardless of the MEA concentrations in NbZnSnO channel layers. A TFT with an MEA/Nb ratio of 0.2 also present the smallest negative shift of DVTH. The variation of DVth under bias stresses is related to the defects at the interface between the channel and dielectric layers or in the channel and dielectric layers [29]. From XPS and transmittance results, a device with MEA/Nb ratio of 0.2 possesses the lowest number of oxygen vacancies and lower absorption coefficient as compared with other devices, resulting in good bias stress stability. Overall, TFTs using NbZnSnO channels with a suitable amount of MEA can exhibit good electrical performance as well as good device stability, which are suitable for optoelectronics applications such as sensor technology. 4. Conclusions In this study, monoethanolamine (MEA) dopants are not only as a precursor stabilizer but also as nitrogen additives to replace oxygen ions in NbZnSnO channel layers fabricated by a solution
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
377
Fig. 10. ID-VG transfer characteristics for NbZnSnO TFTs with (a) MEA/Nb ¼ 0, (b) MEA/Nb ¼ 0.1, (c) MEA/Nb ¼ 0.2, (d) MEA/Nb ¼ 0.3, and (e) MEA/Nb ¼ 0.5 under positive bias stresses (VG ¼ þ20 V) and stress duration of 3600 s. The inset presents the ID-VG transfer characteristics for NbZnSnO TFTs with MEA/Nb ratios of 0e0.5 under negative bias stresses (VG ¼ 20 V) and stress duration of 3600 s.
378
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
Acknowledgements We would like to appreciate the financial support from the Ministry of Science and Technology of Taiwan (Grant Nos. MOST 106-2221-E-024-006-). References
Fig. 11. (a) The threshold voltage shift (△Vth) versus the PBS (VG ¼ þ20 V) time for the NbZnSnO devices with different MEA/Nb ratios. (b) The threshold voltage shift (△Vth) versus the NBS (VG ¼ 20 V) time for the NbZnSnO devices with various MEA/Nb ratios.
process. The oxygen vacancy ratios of NbZnSnO channel layers, prepared from precursor solution with aging time of 20 h, decrease with increasing the MEA/Nb ratio up to 0.2, relating to the replacement of O ions by N ions. A considerable improvement in the electrical properties and bias stability of TFT at the optimum MEA/Nb ratio of 0.2 was also achieved. However, the oxygen vacancy ratios increase when the MEA/Nb ratio increases from 0.2 to 0.3 or 0.5, which are related to the formation of hydrocarbon byproducts. In addition, MEA can enhance the hydrolysis and condensation reaction of NbZnSnO channel layers and result in improved electrical characteristics of TFTs produced under lowtemperature annealing and at a short aging time of the precursor solution.
[1] H.Q. Chiang, J.F. Wager, R.L. Hoffman, J. Jeong, D.A. Keszler, Appl. Phys. Lett. 86 (2005), 013503-13511e013503-3. [2] D.L. Young, H. Moutinho, Y. Yan, T.J. Coutts, J. Appl. Phys. 92 (2002) 310e319. [3] R.D. Chandra, M. Rao, K. Zhang, R.R. Prabhakar, C. Shi, J. Zhang, S.G. Mhaisalkar, N. Mathews, ACS Appl. Mater. Interfac. 6 (2014) 773e777. [4] Y. Jeong, C. Bae, D. Kim, K. Song, K. Woo, H. Shin, G. Cao, J. Moon, ACS Appl. Mater. Interfac. 2 (2010) 611e615. [5] Y.S. Rim, D.L. Kim, W.H. Jeong, H.J. Kim, Electrochem. Solid State Lett. 15 (2012) H37eH40. [6] H. Kim, W.-S. Choi, Ceram. Int. 43 (2017) 4775e4779. [7] J.S. Jeng, J. Alloys Compd. 676 (2016) 86e90. [8] B.D. Ahna, K.H. Leeb, J. Parkc, J.-S. Park, Appl. Surf. Sci. 355 (2015) 1267e1271. [9] C.B. Carter, M.G. Norton, Ceramic Materials: Science and Engineering, first ed., Springer Science & Business Media, Berlin, 2007, p. p126. [10] J. Li, Y.-Z. Fu, C.-X. Huang, J.-H. Zhang, X.-Y. Jiang, Z.-L. Zhang, Appl. Phys. Lett. 108 (2016), 143505-1e143505e5. [11] J. Raja, K. Jang, N. Balaji, W. Choi, T.T. Trinh, J. Yi, Appl. Phys. Lett. 102 (2013), 083505-1e083505e4. [12] J. Zhang, P. Dong, Y. Gao, C. Sheng, X. Li, ACS Appl. Mater. Interfac. 7 (2015) 24103e24109. [13] Y.N. Gao, Y.L. Xu, J.G. Lu, J.H. Zhang, X.F. Li, J. Mater. Chem. C 3 (2015) 11497e11504. [14] J. Li, Y.-Z. Fu, C.-X. Huang, J.-H. Zhang, X.-Y. Jiang, Z.-L. Zhang, Appl. Phys. Lett. 108 (2016), 143505-1e14505e5. [15] J.-F. Chien, C.-H. Chen, J.-J. Shyue, M.-J. Chen, ACS Appl. Mater. Interfac. 4 (2012) 3471e3475. [16] H.S. Lim, Y.S. Rim, D.L. Kim, W.H. Jeong, H.J. Kim, Electrochem. Solid State Lett. 15 (2012) H78eH80. [17] C. Chouillet, J.-M. Krafft, C. Louis, H. Lauron-Pernot, Spectrochim. Acta, Part A 60 (2004) 505e511. [18] J. Park, X. Zhang, M.-H. Bae, G.-T. Park, J.-H. Bae, Jpn. J. Appl. Phys. 52 (Part 1) (2013), 111602-1e111602-6. [19] C.H. Park, S.B. Zhang, S.-H. Wei, Phys. Rev. B 66 (2002), 073202-1e073202-3. [20] A. Kobayashi, O.F. Sankey, J.D. Dow, Phys. Rev. B 28 (1983) 946e956. [21] C.E. Kim, I. Yun, Appl. Phys. Lett. 100 (2012), 013501-1e013501-3. [22] J. Yang, Y. Han, R. Fu, Q.J. Zhang, Disp. Technol 12 (2016) 1560e1564. [23] L. Znaidi, Mater. Sci. Eng. B 174 (2010) 18e30. [24] S.H. Yoon, D. Liu, D. Shen, M. Park, D.-J. Kim, J. Mater. Sci. 43 (2008) 6177e6181. [25] Pankaj Kr, Baitha, J. Manam, J. Rare Earths 33 (8) (2015) 805e812. [26] B.C. Smith, Infrared Spectral Interpretation: a Systematic Approach, CRC press, 1998. [27] G. Socrates, Infrared Characteristic Group Frequencies, John Wiley & Sons, New York, 1994. [28] L. Znaidi, G.J.A.A. Soler Illia, S. Benyahia, C. Sanchez, A.V. Kanaev, Thin Solid Films 428 (2003) 257e262. [29] B. Ryu, H.-K. Noh, E.-A. Choi, K.J. Chang, Appl. Phys. Lett. 97 (2010), 0221081e022108-3.
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: http://www.elsevier.com/locate/jalcom
Fabrication of solution-processed nitrogen-doped niobium zinc tin oxide thin film transistors using ethanolamine additives Jiann-Shing Jeng*, Chi-Min Wu Department of Materials Science, National University of Tainan, Tainan 700, Taiwan
a r t i c l e i n f o
a b s t r a c t
Article history: Received 7 June 2017 Received in revised form 5 September 2017 Accepted 14 September 2017 Available online 18 September 2017
In this study, we fabricated the nitrogen-doped NbZnSnO channel layers by using the sol-gel method. Monoethanolamine (MEA) was used as a nitrogen additive. From the XPS results, the concentration of oxygen vacancies changes as a function of MEA/Nb ratio. A NbZnSnO film with MEA/Nb of 0.2 shows the lowest amount of oxygen vacancies. TFT electrical performance also shows a device with an MEA/Nb ratio of 0.2 possesses a high carrier mobility (7.4 cm2 V1s1) and good bias stress stability. In addition, we also investigated the effect of the aging time of precursor solution on the electrical characteristics of the TFT. After adding MEA, the annealing temperature of the NbZnSnO channels can be reduced, pertaining to the acceleration of the hydrolysis and condensation reaction. © 2017 Elsevier B.V. All rights reserved.
Keywords: Solution process Monoethanolamine Thin film transistor Nitrogen Oxide
1. Introduction In recent years, ZnSnO films have received much attention in the application for channels of thin film transistors (TFTs) [1]. The devices show a high carrier mobility in ambient air, a large on-to-off current ratio, and low production cost in comparison with Sibased TFTs [2,3]. In addition, solution-derived thin films have been widely investigated, which possesses many merits such as cost reduction, low-temperature fabrication, the simplicity of the process, the precise control of the stoichiometry, and high yield as compared with vapor deposition techniques. Therefore, solution processed ZnSnO has been utilized as a channel material in microelectronic. It is well known that the bias stress instability of TFTs is related to the amounts of oxygen vacancies in channel layers [4]. To improve the bias stress instability during device operation, some works have investigated ZnSnO channels added with various transition metal cations [5,6]. Our previous study [7] showed that introducing Nb into the ZnSnO channel materials could enhance the electrical performance of TFTs. Some literature [8,10] indicated that nitrogen anions can successfully substitute to O and eliminate the amounts
of oxygen vacancies in metal oxide channel layers because the ionic radii of N2 (0.129 nm) and N3 (0.132 nm) are similar to that of O2 (0.126 nm). The effective substitution of oxygen ions by nitrogen ions is based on the Hume-Rothery rules [9] where the ionic radii difference between the solute and solvent atoms should be no more than 15%. However, the N-doped metal oxide channel layers in the previous studies were fabricated by vacuum deposition techniques instead of solution processes [10,11]. In addition, monoethanolamine (MEA) was often used as a stabilizing agent for fabrication of solution-processed metal oxide channel layers [12,13]. In this study, we attempt to use MEA as a nitrogen additive for adding N into the host NbZnSnO channel layers to enhance the electrical performance of solution-processed NbZnSnO TFTs. We demonstrate the production of NbZnSnO device with MEA/Nb of 0.2 shows a mobility of 7.4 cm2 VV1s1, an on-to-off current ratio of 108 and a small subthreshold swing (~0.3 V/decade) by using a solution process technique. At the same time, we try to adjust the oxygen vacancies of the N-doped NbZnSnO TFTs using different amounts of MEA additives. In addition, the effect of MEA on the aging time of NbZnSnO precursor is also investigated. Finally, the correlation among the MEA-doping concentration, aging time, and device performance is discussed. 2. Material and methods
* Corresponding author. E-mail address: [email protected] (J.-S. Jeng). http://dx.doi.org/10.1016/j.jallcom.2017.09.151 0925-8388/© 2017 Elsevier B.V. All rights reserved.
NbZnSnO channel layers with adding different amounts of
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
371
Fig. 1. Cross-section TEM photographs of the NbZnSnO films with MEA/Nb molar ratios of (a) 0, (b) 0.2, and (3) 0.5 on SiO2/pþeSi. (d) GIAXRD patterns of 10-layers NbZnSnO channel layers doped with various MEA/Nb ratios.
monoethanolamine (MEA) were fabricated via a combined sol-gel and spin-coating method. From our previous study [7], an optimum Nb-doping concentration of solution-processed NbZnSnO TFTs is 3%, which possesses the best characteristics of the device. Therefore, the molar ratio of Nb: Zn: Sn in NbZnSnO precursors was chosen as 0.03:1:0.97. All reagents, anhydrous niobium chloride (NbCl5), hydrated Zinc nitrate (Zn(NO3)2$6H2O) and hydrated tin chloride dihydrate (SnCl2$2H2O) were dissolved in 20 ml ethylene glycol monomethyl ether (EGME) for preparing the NbZnSnO precursors. For MEA dopants, different amounts of MEA used as starting materials were weighed according to the molar ratios of MEA/Nb (i.e., 0, 0.1, 0.2, 0.3, and 0.5). After stirring for 20 h at room temperature, the NbZnSnO precursors with and without adding MEA achieved chemical homogeneity. Additionally, the precursor solutions were stirred for 2, 6, and 20 h, based on the color variation of NbZnSnO precursor, to proceed with the aging experiment by Fourier transform infrared absorption spectroscopy (FTIR) analysis and TFT characteristics. The NbZnSnO precursors were spin-coated onto a heavily doped p-type Si wafer with a 100 nm thick SiO2 layer, which functions as the gate electrode and gate dielectric. The spin-coated NbZnSnO active layers with and without MEA additives were rapidly baked at 110 C for 3 min and subsequently annealed on a heating plate at 430 C for 1 h under an air atmosphere. To verify the center of the plate surface controlled by the temperature controller, the temperature calibration of the heating plate was performed by a thermocouple. Finally, the 300-nm-thick Al source and drain electrodes were deposited on top of the active layer using an e-beam evaporation technique through a shadow mask to define transistor with the channel width (2000 mm) and length (100 mm). The thickness of the NbZnSnO film with and without MEA additives was determined using a transmission electron microscope (TEM, PHILIPS CM-200). To prepare the TEM-specimen, a focused ion beam instrument (FIB) was used to thin the samples for electron transparency. A protective layer of carbon was deposited on the NbZnSnO films with and without MEA additives
to prevent beam-induced damage during the milling process. The chemical compositions and oxygen vacancies were carried out with the x-ray photoelectron spectroscopy (XPS, JEOL JAMP9500F). A Hitachi U-2001 UV/Visible Spectrophotometer was used to measure the optical transmittance of the NbZnSnO composite films with and without adding MEA that were deposited on quartz glass substrates. Fourier transform infrared absorption spectroscopy (FTIR) was used to characterize the bonding configurations of NbZnSnO channels with and without MEA additives that were deposited on a KBr substrate. The currentevoltage characteristics of the NbZnSnO TFTs were measured under air ambient in a dark box using an Agilent 4156 C semiconductor parameter analyzer.
3. Results and discussion To determine the thickness of NbZnSnO films with and without adding monoethanolamine (MEA), the FIB-prepared thin samples were measured by transmission electron microscope (TEM). Fig. 1 presents the cross-section TEM images of solution-processed NbZnSnO films with different amounts of MEA (we designated it as NNZTO in Fig. 1). From Fig. 1(a) ~ 1(c), the thickness of NbZnSnO films without and with adding MEA is around 5~7 nm. The crystalline characteristics of all NbZnSnO films were performed by glancing incidence angles x-ray diffraction (GIAXRD). 10-layers NbZnSnO films were fabricated to obtain sufficient pattern intensity. All NbZnSnO films, no matter with or without adding MEA, are amorphous, as shown in Fig. 1(d). A diffraction peak at around 54 is due to the Si(100) plane of the Si substrate. The main idea of this study is to fabricate nitrogen-doped NbZnSnO films by using the sol-gel method and utilizing MEA additives. Therefore, the chemical composition and the chemical state of the NbZnSnO films with various ratios of MEA/Nb, prepared from precursor solution under the aging time of 20 h, were performed with X-Ray Photoelectron Spectroscopy (XPS). In order to avoid the surface contamination, sputter cleaning with 1 keV Ar
372
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
Fig. 2. XPS N1s spectra of the NbZnSnO thin films with various ratios of MEA/Nb.
ions was made for 4s to clean the NbZnSnO samples. For calibration of peak position, all the binding energies values were referenced to adventitious C1s at 284.8eV. Fig. 2 presents the XPS spectra of the N 1s peak for NbZnSnO films with various MEA/Nb ratios. From Fig. 2, N 1s XPS peak centered at 399.05 eV is found for NbZnSnO films
after MEA adding. Li et al. [14] and Chien et al. [15] reported that a similar binding energy of 399.32 and 399.2 eV associated with the ZneN bonds in the ZnO: N films. From Fig. 2, the N 1s XPS peaks present for all MEA-doped NbZnSnO films in comparison with undoped films, implying the effective nitrogen doping into NbZnSnO films using MEA additives. The O 1s photoelectron spectra of the NbZnSnO channel layers doped with various concentrations of MEA are present in Fig. 3. The XPS O 1s spectra of the NbZnSnO films with different amounts of MEA/Nb were deconvoluted into three components, including the low-energy component is associated with lattice oxygen in NbZnSnO films (530.62eV, OI), the medium-energy component is referred to oxygen nearby oxygen-deficient oxide of NbZnSnO films (532.2eV, OII), and the high-energy component is the oxygen in water molecules adsorbed on the NbZnSnO film surface (533.62eV, OIII) [16]. Fig. 4(a) shows the oxygen vacancy ratios [i.e., OII/(OI þ OII þ OIII)] of the NbZnSnO channel layers versus the MEA/Nb ratios. One can see that the oxygen vacancy ratios of the NbZnSnO films decrease from 0.17 to 0.1 for the MEA/Nb ratio increases from 0 to 0.2, and then the ratios increase to 0.175 and
Fig. 3. O 1s XPS spectra of the NbZnSnO thin films with MEA/Nb ratio of (a) 0, (b) 0.1, (c) 0.2, (d) 0.3, and (e) 0.5.
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
Fig. 4. (a) The oxygen vacancy ratios of NbZnSnO channel layers as a function of MEA/ Nb ratios. (b) FTIR transmission spectra of NbZnSnO films with various ratios of MEA/ Nb.
0.19 as the MEA/Nb ratio reaches 0.3 and 0.5, respectively. A decrease in oxygen vacancy ratios of the NbZnSnO films with adding small ratios of MEA/Nb (0.2) is due to the effective substitution of O by N. It is noted that the oxygen vacancy ratios of the NbZnSnO films increase with adding MEA content for the larger MEA/Nb ratios (>0.2). In order to understand the reason that a rise in oxygen vacancy ratios of NbZnSnO films with MEA/Nb of 0.3 and 0.5, the binding environment of the NbZnSnO films with different MEA/Nb ratios, prepared with 20 h aged precursor solution, was measured by Fourier transform infrared spectroscopy (FTIR). To obtain the information of the CeH related bonds, all NbZnSnO films with different MEA/Nb ratios were spun on KBr substrates and further baking at 110 C for 3 min to proceed the FTIR analysis. Fig. 4(b) present the FTIR transmission spectra of NbZnSnO films with different amounts of MEA. Two strong bands are observed in the IR spectrum of NbZnSnO films without adding MEA, one at 1336 cm1 and the second band at 1375 cm1, which are due to an interaction of nitrate with metallic cation [17]. In addition, the nitrate related bands (at 1336 and 1375 cm1) eliminated in NbZnSnO films with adding MEA. The OeH stretching vibration (at 3000e3500 cm1) and eOH group (at around 1620 cm1) are related to the hydroxyl group in NbZnSnO film. OH-related bands in IR spectrum of undoped NbZnSnO films may be attributed to the vibrations of MOH (where M denotes metals, including Nb, Zn, and Sn in this paper) or MeOHeM groups. By contrast, asymmetric CeH related bands at around 2850 and 2924 cm1 are found in NbZnSnO films with MEA/Nb ratios of 0.3 and 0.5, which can be assigned to the CeH related bands of the hydrocarbon byproducts. This result indicates that there is a suitable MEA-doping concentration in NbZnSnO films. When MEA-doping concentration is larger than the suitable value, the CeH related bonds are present. After annealing at 430 C, the hydrocarbon byproducts of NbZnSnO films with MEA/ Nb ratios of 0.3 and 0.5 decompose and result in the formation of
373
Fig. 5. (a) UVevis transmittance spectra of NbZnSnO films with different ratios of MEA/Nb (b) Absorption coefficient of NbZnSnO films with different contents of MEA/ Nb varies as a function of the photon energy (hn).
oxygen vacancy, leading to the higher amounts of oxygen vacancy ratios of NbZnSnO films with MEA/Nb ratios of 0.3 and 0.5, as shown in Fig. 4(a). The transmittance of NbZnSnO films with various MEA concentrations, prepared from solution aged for 20 h, was measured by
Fig. 6. (a) IDVG transfer characteristics of NbZnSnO TFTs using NbZnSnO channel layers with various MEA/Nb ratios, prepared from precursor solution aged for 20 h. (b) Mobility, threshold voltage (Vth), and subthreshold swing (S$S.) as functions of MEA/ Nb ratios.
374
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
a spectrophotometer. To obtain reasonable transmittance results, all NbZnSnO films were spun 10 times on quartz glass to achieve sufficient thickness. Fig. 5(a) shows the transmittance spectra of 10layers NbZnSnO films with various MEA/Nb ratios. It can be seen that all NbZnSnO films show good optical transparency (>80%) in the 400 nme700 nm range (visible region) from the transmittance spectra. After adding MEA, the transmittance increases for all MEAadded NbZnSnO films. By contrast, the absorption coefficients of NbZnSnO films with MEA additives are smaller than that of NbZnSnO films without MEA additives (see Fig. 5(b)), implying that the MEA dopants can eliminate the subgap absorption of NbZnSnO films. In order to realize the electrical characteristics of NbZnSnO TFTs, NbZnSnO films with different MEA/Nb ratios, prepared from precursor solution under the aging time of 20 h, were used as channel layers. We have fabricated all devices simultaneously at the same batch in order to avoid the susceptibility to the humidity, temperature and other environmental factors that will affect the device performance. Fig. 6(a) presents the transfer characteristics of NbZnSnO TFTs with and without adding MEA, prepared from precursor solution under the aging time of 20 h. The on/off current ratio of all the TFTs is larger than 107. Fig. 6(b) shows the carrier mobility (m), the threshold voltage (Vth), and
subthreshold swing (S$S.) as functions of MEA/Nb ratios. The carrier mobility and the threshold voltage in the saturation region of device operation for the NbZnSnO TFTs with and without adding MEA were extracted from the slope and intercept of an extrapolated line for the (ID)1/2-VG curve, respectively. The mobility can be overestimated in TFTs with the non-patterned channel layers because of the fringing currents around the peripheral region of non-precise active layer [18]. The subthreshold swing (S.S.) is defined by the inverse slope of the log (ID) vs. VG characteristic in the subthreshold region. The mobility of the NbZnSnO device increases with increasing the MEA/Nb ratio to 0.2 (m~7.4 cm2V1s1) and then decreases with increasing the MEA/Nb ratio to 0.5 (m~1.3 cm2V1s1), as shown in Fig. 6(b). The S.S. curve is a reversed trend of that of the mobility trend for the NbZnSnO TFTs with different MEA/Nb ratios. The lowest value of S.S. was observed to be 0.3 V/decade for the device with MEA/Nb ratio of 0.2. In addition, the Vth of the TFTs increases with increasing the MEA concentrations, as shown in Fig. 6(b). N ions were considered as p-type dopants for ZnO-based films [19,20]. Also, literature [21,22] suggested that nitrogen additives in metal oxide can reduce the carrier concentration, resulting in the positive shift of Vth. Fig. 7 shows the drain current versus drain to source voltage (ID-VD) output characteristics of the NbZnSnO TFTs
Fig. 7. The output characteristics (ID-VD) curves of the NbZnSnO TFTs with different MEA/Nb ratios.
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
Fig. 8. The transfer characteristics of TFTs using NbZnSnO channel layers with different MEA/Nb ratios, prepared from precursor solution aged for (a) 2 and (b) 6 h.
with and without adding MEA. All TFT devices exhibit typical normally-off type n-channel behavior. We have also studied the effects of stirring time (aging time) on the electrical performance of solution processed NbZnSnO TFTs with MEA/Nb ratios of 0, 0.2, and 0.5. The NbZnSnO-based presolutions were stirred with a magnetic stirrer for 2, 6, and 20 h at room temperature and then spun on Si substrate with SiO2 to be fabricated to form TFTs (as stated in the experimental section). Fig. 8 (a) and 8 (b) present the transfer characteristics of TFTs using NbZnSnO channel layers with different MEA/Nb concentrations, prepared from precursor solution aged for 2 and 6 h, respectively. As can be seen, the transfer characteristics of the same devices using channels prepared with 20 h aged precursor solution are shown in Fig. 6(a). The extracted TFT device parameters were listed in Table 1. From Fig. 8(a), the device using
375
channels without adding MEA, prepared with 2 h aged precursor, can not operate whereas the devices with MEA/Nb of 0.2 and 0.5 under the same aging time can successfully operate. Compared with the device without any MEA additive, an improvement of the TFT performance using NbZnSnO channels prepared with 2 h aged precursor solution is observed when the MEA/Nb ratio increases from 0.2 to 0.5. On the other hand, after 20 h aging of the precursors, 280 C and 380 C annealed NbZnSnO devices with MEA/Nb ratio of 0.2 show better performance such as higher mobility and lower subthreshold swing than that of device without and with adding MEA/Nb ratio of 0.5 under the same annealing temperature, as shown in Table 1. Literature [23,24] indicated MEA can improve the bonding ability of metal ion in precursor and decrease the formation energy barrier of metal oxide. From FTIR spectra in Fig. 4(b), as compared with undoped NbZnSnO film prepared with 20 h aged precursor solution, the hydroxyl groups (at around 1620 and 3400 cm1) in NbZnSnO film disappear for NbZnSnO film after adding MEA, implying the MEA can enhance the reaction of the presolutions and reduce the annealing temperature of NbZnSnO channels. As the aging time of the precursor increases to 6 h, a device without adding MEA shows the typical transfer characteristics of the TFTs measured at VD ¼ 40 V, as shown in Fig. 8(b). After increasing the aging time of the precursor from 6 to 20 h, the mobility of 430 C annealed undoped device improves from 0.61±0.09 to 4.81±0.17 cm2V1s1, as shown in Table 1. The mobility of a device with MEA/Nb ratio of 0.2 also increases with increasing the aging time of the presolutions. It is noted that TFTs with MEA/Nb ratio of 0.5 possesses a mobility of 6.79±0.55 cm2V1s1 after 2 h aging of the precursors and then deteriorates to 2.32±0.21 cm2V1s1 after aging of precursor solution for 6 h. After 20 h aging of the precursor solution, a continuous decrease in mobility of device with MEA/Nb ratio of 0.5 is observed (see Table 1). In order to realize the relation between the aging time of the precursors and electrical characteristics of TFTs, FTIR analysis was proceeded to understand the bonding environment of the NbZnSnO channels prepared from precursor solution under different aging time. Fig. 9 presents the FTIR spectra of NbZnSnO channel layers with different MEA concentrations, prepared from precursor solution under various aging time. From Fig. 9(a), the transmittance band found at 623 cm1 is assigned as stretching vibration of SneO [25]. In addition, the transmittance bands for undoped NbZnSnO channel layer at 1055 and 1375 cm1 belong to the NO stretch band of a free nitrate group and the NO2 stretching mode of a free nitrate ion, respectively [17]. After increasing the
Table 1 Comparison of the electrical characteristics measured in the air including mFE, Vth, S.S., and Ion/Ioff ratio for solution-processed NbZnSnO TFTs with different MEA/Nb additives under various fabrication conditions. Errors indicate the standard deviation of multiple independent devices. Devices MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb MEA/Nb
Aging time/annealing tem. of of of of of of of of of of of of of of of
0 0.2 0.5 0 0.2 0.5 0 0.2 0.5 0 0.2 0.5 0 0.2 0.5
aging aging aging aging aging aging aging aging aging aging aging aging aging aging aging
2 h/430 C 2 h/430 C 2 h/430 C 6 h/430 C 6 h/430 C 6 h/430 C 20 h/430 C 20 h/430 C 20 h/430 C 20 h/380 C 20 h/380 C 20 h/380 C 20 h/280 C 20 h/280 C 20 h/280 C
mFE (cm2V1s1)
Vth (V)
S.S.(V/dec.)
Ion/Ioff ratio
e 0.91 ± 0.02 6.79 ± 0.55 0.61 ± 0.09 1.16 ± 0.16 2.32 ± 0.21 4.81 ± 0.17 7.41 ± 0.51 1.32 ± 0.12 1.35 ± 0.11 1.46 ± 0.24 0.056 ± 0.012 0.003 ± 0.001 0.08 ± 0.01 0.02 ± 0.01
e 9.0 8.5 7.0 9.0 9.0 6.0 7.0 8.5 3.5 6.0 6.0 3.0 5.0 8.0
e 0.94 1.25 2.34 0.17 0.69 0.88 0.30 0.90 0.96 0.38 0.40 0.37 0.20 0.30
e ~107 ~108 ~106 ~107 ~108 ~107 ~108 ~107 ~106 ~108 ~106 ~106 ~106 ~105
± ± ± ± ± ± ± ± ± ± ± ± ± ±
0.8 0.5 0.7 0.8 0.5 0.7 0.4 0.5 0.7 0.9 0.8 0.7 0.7 0.7
± ± ± ± ± ± ± ± ± ± ± ± ± ±
0.31 0.11 0.30 0.31 0.11 0.12 0.08 0.20 0.10 0.10 0.10 0.10 0.08 0.25
376
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
Fig. 9. FTIR transmission spectra of TFTs using NbZnSnO channel layers with (a) MEA/ Nb of 0, (b) MEA/Nb of 0.2, and (c) MEA/Nb of 0.5, prepared from precursor solution under different aging time.
aging time of the precursor to 6 h, the NO2 stretching band is split into two bands at 1336 and 1375 cm1, which is due to an interaction of nitrate with metallic cation [17] (i.e., the coordinated nitrate groups), as shown in Fig. 9(a). The coordinated nitrate band tends to become more apparent after aging for 20 h, implying that the hydrolysis and condensation reaction becomes more complete (i.e., more metaloxide formation) as the aging time of the precursor solution increases. The broad concave at 3400 cm1 and a transmittance peak at 1620 cm1 in NbZnSnO film, which are related to the hydroxyl group in NbZnSnO film. The hydroxyl group is still present in the NbZnSnO film without adding MEA even though precursors persist for a long-term aging of 20 h, as shown in Fig. 9(a). After aging for 20 h, the three nitrate-related bands (i.e., 1055, 1336, and 1375 cm1) of NbZnSnO films with MEA/Nb ratio of 0.2 disappear as compared to undoped NbZnSnO films, as shown in Fig. 9(b). This result indicates the hydrolysis and condensation reaction of NZTO films with an MEA/Nb ratio of 0.2
is complete under longer aging time. In the IR spectral range less than 1000 cm1 (see Fig. 9(b) and (c)), NbZnSnO samples reveal several transmittance bands. A band at around 806 cm1 is assigned to CeO out of plane bending mode of MEA. In addition, the vibrational bands at around 955 cm1 and 1018 cm1 are due to the CeNeH out-of-plane wagging and CeNH2 twisting of MEA [26] and alcohol CeO stretching vibrations [27], respectively. The FTIR spectra of NbZnSnO films with an MEA/Nb ratio of 0.5 do not show any nitrate-related groups after 2 h aging of the precursor solution, indicating an aging time of 2 h can lead to a complete hydrolysis and condensation reaction, as seen in Fig. 9(c). After increasing the aging time of the precursor solution, symmetric CeH stretching vibration (2924 cm1) was observed in the FTIR spectra of NbZnSnO films with an MEA/Nb ratio of 0.5. Adding MEA in NbZnSnO solution can increase the pH of the presolution due to the amine [28], which should promote the formation of NbZnSnO films. After increasing MEA additives up to MEA/Nb of 0.5, the enhanced formation of NbZnSnO films occurs under the short aging time of 2 h. With increasing the aging time of the precursor solution, hydrocarbon byproducts produced from the condensation and hydrolysis reactions of the precursors, which is accompanied by a main M-O-M product. This result leads to the occurrence of CeH stretching vibration in FTIR spectra of NbZnSnO TFT with an MEA/Nb ratio of 0.5. To investigate the electrical stability of the TFTs, a positive gate bias (PBS) of þ20 V was applied to the gate electrode of the devices and the source and drain electrodes were grounded for time periods ranging from 0 to 3600 s. The transfer curves of the TFTs using NbZnSnO channel layers with different MEA/Nb concentrations, prepared from precursor solution aged for 20 h, were measured under the dark in the air and stress duration of 3600 s at VD ¼ 40 V for PBS test. The measured results were shown in Fig. 10. All ID-VG curves of the NbZnSnO devices shift positively after PBS. The shift values of the threshold voltage (DVth) are relative to the pristine device. Fig. 11(a) presents the change in DVth as a function of PBS time for the NbZnSnO devices with various MEA/Nb ratios. As compared to other devices, a device with MEA/Nb ratio of 0.2 exhibits a smaller positive change in DVth after PBS for 3600 s. Additionally, a possible cause of electrical instability could be related to the presence of oxygen vacancies in the channel layer. Furthermore, the inset in Fig. 10 shows the transfer ID-VG curve of NbZnSnO films with and without adding MEA under negative gate bias (NBS, VG ¼ 20 V) and stress duration of 3600 s for VD ¼ 40 V (the source and drain electrodes were grounded). From Fig. 11(b), a much smaller negative change in DVth is found in our MEA doped NbZnSnO devices regardless of the MEA concentrations in NbZnSnO channel layers. A TFT with an MEA/Nb ratio of 0.2 also present the smallest negative shift of DVTH. The variation of DVth under bias stresses is related to the defects at the interface between the channel and dielectric layers or in the channel and dielectric layers [29]. From XPS and transmittance results, a device with MEA/Nb ratio of 0.2 possesses the lowest number of oxygen vacancies and lower absorption coefficient as compared with other devices, resulting in good bias stress stability. Overall, TFTs using NbZnSnO channels with a suitable amount of MEA can exhibit good electrical performance as well as good device stability, which are suitable for optoelectronics applications such as sensor technology. 4. Conclusions In this study, monoethanolamine (MEA) dopants are not only as a precursor stabilizer but also as nitrogen additives to replace oxygen ions in NbZnSnO channel layers fabricated by a solution
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
377
Fig. 10. ID-VG transfer characteristics for NbZnSnO TFTs with (a) MEA/Nb ¼ 0, (b) MEA/Nb ¼ 0.1, (c) MEA/Nb ¼ 0.2, (d) MEA/Nb ¼ 0.3, and (e) MEA/Nb ¼ 0.5 under positive bias stresses (VG ¼ þ20 V) and stress duration of 3600 s. The inset presents the ID-VG transfer characteristics for NbZnSnO TFTs with MEA/Nb ratios of 0e0.5 under negative bias stresses (VG ¼ 20 V) and stress duration of 3600 s.
378
J.-S. Jeng, C.-M. Wu / Journal of Alloys and Compounds 729 (2017) 370e378
Acknowledgements We would like to appreciate the financial support from the Ministry of Science and Technology of Taiwan (Grant Nos. MOST 106-2221-E-024-006-). References
Fig. 11. (a) The threshold voltage shift (△Vth) versus the PBS (VG ¼ þ20 V) time for the NbZnSnO devices with different MEA/Nb ratios. (b) The threshold voltage shift (△Vth) versus the NBS (VG ¼ 20 V) time for the NbZnSnO devices with various MEA/Nb ratios.
process. The oxygen vacancy ratios of NbZnSnO channel layers, prepared from precursor solution with aging time of 20 h, decrease with increasing the MEA/Nb ratio up to 0.2, relating to the replacement of O ions by N ions. A considerable improvement in the electrical properties and bias stability of TFT at the optimum MEA/Nb ratio of 0.2 was also achieved. However, the oxygen vacancy ratios increase when the MEA/Nb ratio increases from 0.2 to 0.3 or 0.5, which are related to the formation of hydrocarbon byproducts. In addition, MEA can enhance the hydrolysis and condensation reaction of NbZnSnO channel layers and result in improved electrical characteristics of TFTs produced under lowtemperature annealing and at a short aging time of the precursor solution.
[1] H.Q. Chiang, J.F. Wager, R.L. Hoffman, J. Jeong, D.A. Keszler, Appl. Phys. Lett. 86 (2005), 013503-13511e013503-3. [2] D.L. Young, H. Moutinho, Y. Yan, T.J. Coutts, J. Appl. Phys. 92 (2002) 310e319. [3] R.D. Chandra, M. Rao, K. Zhang, R.R. Prabhakar, C. Shi, J. Zhang, S.G. Mhaisalkar, N. Mathews, ACS Appl. Mater. Interfac. 6 (2014) 773e777. [4] Y. Jeong, C. Bae, D. Kim, K. Song, K. Woo, H. Shin, G. Cao, J. Moon, ACS Appl. Mater. Interfac. 2 (2010) 611e615. [5] Y.S. Rim, D.L. Kim, W.H. Jeong, H.J. Kim, Electrochem. Solid State Lett. 15 (2012) H37eH40. [6] H. Kim, W.-S. Choi, Ceram. Int. 43 (2017) 4775e4779. [7] J.S. Jeng, J. Alloys Compd. 676 (2016) 86e90. [8] B.D. Ahna, K.H. Leeb, J. Parkc, J.-S. Park, Appl. Surf. Sci. 355 (2015) 1267e1271. [9] C.B. Carter, M.G. Norton, Ceramic Materials: Science and Engineering, first ed., Springer Science & Business Media, Berlin, 2007, p. p126. [10] J. Li, Y.-Z. Fu, C.-X. Huang, J.-H. Zhang, X.-Y. Jiang, Z.-L. Zhang, Appl. Phys. Lett. 108 (2016), 143505-1e143505e5. [11] J. Raja, K. Jang, N. Balaji, W. Choi, T.T. Trinh, J. Yi, Appl. Phys. Lett. 102 (2013), 083505-1e083505e4. [12] J. Zhang, P. Dong, Y. Gao, C. Sheng, X. Li, ACS Appl. Mater. Interfac. 7 (2015) 24103e24109. [13] Y.N. Gao, Y.L. Xu, J.G. Lu, J.H. Zhang, X.F. Li, J. Mater. Chem. C 3 (2015) 11497e11504. [14] J. Li, Y.-Z. Fu, C.-X. Huang, J.-H. Zhang, X.-Y. Jiang, Z.-L. Zhang, Appl. Phys. Lett. 108 (2016), 143505-1e14505e5. [15] J.-F. Chien, C.-H. Chen, J.-J. Shyue, M.-J. Chen, ACS Appl. Mater. Interfac. 4 (2012) 3471e3475. [16] H.S. Lim, Y.S. Rim, D.L. Kim, W.H. Jeong, H.J. Kim, Electrochem. Solid State Lett. 15 (2012) H78eH80. [17] C. Chouillet, J.-M. Krafft, C. Louis, H. Lauron-Pernot, Spectrochim. Acta, Part A 60 (2004) 505e511. [18] J. Park, X. Zhang, M.-H. Bae, G.-T. Park, J.-H. Bae, Jpn. J. Appl. Phys. 52 (Part 1) (2013), 111602-1e111602-6. [19] C.H. Park, S.B. Zhang, S.-H. Wei, Phys. Rev. B 66 (2002), 073202-1e073202-3. [20] A. Kobayashi, O.F. Sankey, J.D. Dow, Phys. Rev. B 28 (1983) 946e956. [21] C.E. Kim, I. Yun, Appl. Phys. Lett. 100 (2012), 013501-1e013501-3. [22] J. Yang, Y. Han, R. Fu, Q.J. Zhang, Disp. Technol 12 (2016) 1560e1564. [23] L. Znaidi, Mater. Sci. Eng. B 174 (2010) 18e30. [24] S.H. Yoon, D. Liu, D. Shen, M. Park, D.-J. Kim, J. Mater. Sci. 43 (2008) 6177e6181. [25] Pankaj Kr, Baitha, J. Manam, J. Rare Earths 33 (8) (2015) 805e812. [26] B.C. Smith, Infrared Spectral Interpretation: a Systematic Approach, CRC press, 1998. [27] G. Socrates, Infrared Characteristic Group Frequencies, John Wiley & Sons, New York, 1994. [28] L. Znaidi, G.J.A.A. Soler Illia, S. Benyahia, C. Sanchez, A.V. Kanaev, Thin Solid Films 428 (2003) 257e262. [29] B. Ryu, H.-K. Noh, E.-A. Choi, K.J. Chang, Appl. Phys. Lett. 97 (2010), 0221081e022108-3.