Facile control of p-type SnO TFT performance with restraining redox reaction by ITO interlayers

By comparing Ni and ITO electrodes of SnO TFT, we find a facile method to control p-type SnO TFT performance. A Ni-electrode TFT has a high field-effect mobility of 3.3 cm2/Vs and a low on/off current ratio of 3.6 × 101. Compared to Ni, ITO-electrode TFT has low field-effect mobility of 1.4 cm2/Vs and a high on/off current ratio of 1.1 × 103. Using various analysis methods, we suggested why the electrical properties of SnO TFT differed depending on the electrode materials. First, a redox reaction occurs at the interface of SnO and Ni during the post-annealing process. Second, Ni has an ohmic-like contact formation with SnO, which lowers the Schottky barrier height of carriers. ITO ILs are adopted to Ni electrode to reduce the off-current by hindering the redox reaction. The off-current of TFTs is effectively reduced with ITO ILs as thickness increases. An ITO IL that is 10-nm thick yields the optimum electrical properties: field-effect mobility of 2.5 cm2/Vs, Ion/Ioff of 1.7 × 103 and Vth shift under NBS of −1.4 V.


Introduction
Tin monoxide (SnO) is a promising material for ptype thin-film transistors (TFTs) due to its high hole mobility [1]. SnO forms a delocalized and isotropic hole conduction path with hybridized spherical Sn 5s orbitals and O 2p orbitals [2]. However, previously reported SnO TFTs exhibit high off-current that is attributed to their ambipolar characteristics. This offcurrent is undesirable because it prevents low power consumption and high CMOS gain. High off-current originates from a redox reaction, oxygen vacancy generation in SnO, and electron injection through the source/drain [3][4][5].
Studies on precise process condition control of SnO fabrication [1] and optimization of source/drain contacts [6] have demonstrated enhancement of the electrical properties of TFTs. Tin has three ionization states in tin oxide: Sn 2+ (SnO), Sn 4+ (SnO 2 ), and Sn 0 (Sn metal). SnO is easily oxidized to SnO 2 or reduced to Sn during film fabrication, post-annealing, and even upon exposure to air because of its metastable phase [7,8]. The mixture phase of SnO 2 resulting from oxidation degrades p-type characteristics with low mobility and high off current because of the conducting n-type aspects of SnO 2 [9]. The reduction of SnO further degrades electrical properties via an oxygen vacancy defect known as electron generation [4].
The source/drain electrode material of TFTs makes direct contact with the active layer. Ni is adopted in the electrode layer because of its low contact resistance along with a similar work function to SnO [10]. However, high off-current and ambipolar characteristics were reported in the Ni-electrode TFTs. L. Y. Liang [5] suggested that the high off-current originates from electron injection via the Ni electrode because of the narrow band gap of SnO. T. Kim et al. deposited a very thin ( ∼ 0.5 nm) Al 2 O 3 interfacial layer onto the electrode using the Atomic Layer Deposition (ALD) method to reduce the off-current [6]. The off-current of TFTs with interfacial layers (ILs) was effectively reduced as the ILs increased the Schottky barrier height of electrons. However, the mobility of these TFTs was degraded as IL thickness increased due to massive contact resistance, owing to the insulating properties of Al 2 O 3 .
In this study, we have investigated why electrical properties such as the field-effect mobility and off-current of SnO TFTs vary depending on the electrode materials. From film and device analysis, we identified contact properties and redox reactions at the interface of the active and electrode affecting the electrical properties of TFTs. Consequently, we optimize p-type SnO TFT performance by adopting the ITO ILs on the Ni electrode.

Experimental design
Coplanar bottom-gate structure TFTs were fabricated as shown in Figure 1(a). Highly doped p-type silicon and 100 nm of ALD-deposited Al 2 O 3 were used as the gate and gate insulators, respectively. The 100-nm-thick electrodes (Ni, ITO, Ni with ITO ILs) were deposited by sputtering and formed using a lift-off process. Subsequently, 10 nm of SnO and Al 2 O 3 were deposited by ALD. The details of the ALD processes used are in previous reports [1,11]. Each layer was defined by conventional photolithography and a wet-etching process. The TFT was 40 μm in width while its length varied at 10, 20, 30, and 60 μm. After TFT fabrication, a post-annealing process was carried out at 300°C in a nitrogen atmosphere for one hour. All the electrical properties of TFTs were measured using a Keithley 4200 semiconductor parameter analyzer. The I on /I off in this paper was defined as the ratio of current at V G = −10V to that at 5V. The five transistors were used to extract the standard deviation of the electrical properties.
X-ray photoelectron spectroscopy (XPS; Versaprobe II), high-resolution transmission electron microscopy (HRTEM, JEM-2100F, JEOL), and energy dispersive Xray microanalysis (EDX) were conducted for the films analysis. The work functions of SnO, Ni, and ITO were measured by Kelvin probe force microscopy (KPFM) using a highly ordered pyrolytic graphite (HOPG) reference sample. The experimental barrier height of the electrode/SnO/P ++ Si diode was calculated using b = kT/q ln(AA * T 2 /I 0 ) [12], where b is the barrier height, k is the Boltzmann constant, T is the temperature, q is the electron charge, A is the diode area, A * is the Richardson constant of SnO (70.8 A/cm 2 K 2 ) [13,14], and I 0 is the reverse saturation current.

Results and discussion
The transfer and output characteristics of the SnO TFTs are shown in Figures 1(b-f). Table 1 summarizes the device parameters. The V th and subthreshold swing (S.S.) are similar regardless of the electrode material used. However, the field-effect mobility and I on /I off vary depending on the electrode material. The field-effect mobility of Ni-electrode TFTs (3.3 cm 2 /Vs) is three times higher than ITO-electrode TFTs (1.4 cm 2 /Vs), and interestingly, the off-current is 2-steps higher ( Figure 1(b,e). Even though SnO is known to inherently have high off-current induced from back-channel and bulk defect states, which mainly consist of oxygen vacancy (V o ) defect [4], the ITO electrode can successfully suppress the off-current level. However, the lowest current at flat band voltage was similar at Ni (2.2 × 10 −12 A) and ITO (2.9 × 10 −12 A) electrode TFTs. These results indicate electrode materials affect channel formation and carrier injection in electrons. Since the materials and process conditions are the same except the electrode material, a  high off-current results from the contact region of the Ni-electrode and SnO layer. There are two potential reasons for this high off-current: (1) V O defects that could generate electrons at the interface by SnO phase transition during a post-annealing process and (2) contact characteristics, such as Schottky barrier height of electrons [6]. The drain current was saturated at high drain voltage regardless of the electrode materials, as shown in Figure 1(c,f). However, a certain degree of current crowding effect is observed in the ITO electrode TFTs at the low gate and drain voltage region, whereas Ni electrode TFTs did not exhibit such an effect in the identical region ( Figure 1(d,g). These results indicate that electrode materials affect the contact properties and hole carrier injection.
The XPS depth profile method is adopted to determine whether redox reactions occur. As shown in Figure 2(a-c), the atomic ratio of metal cations during etching can be divided into four regions: the (1) surface, (2) bulk, (3) interface and (4) electrode. The binding state at the interface region, Sn 3d 5/2 spectra, are shown in Figure 2(d,e) by electrode and depth region. The XPS Sn 3d 5/2 spectra can be deconvoluted into three different tin oxidized states: 486.8 (Sn 4+ ) [14], 486.1 (Sn 2+ ) [15], and 493.0 eV (Sn 0 ) [16]. The binding state of Sn is distinguishably different at (3), the SnO/Ni interface, with a sharply decreased Sn 2+ peak, whereas Sn 0 peak is increased. This result indicates that 300°C-annealed SnO (Sn 2+ ) is partially reduced to Sn metal (Sn 0 ), with electrons generated by oxygen vacancy formation at the Ni electrode. To further study whether redox reaction occurred at the interface of SnO/Ni, XPS Ni 2p 3/2 spectra analysis is conducted as shown in Figure 2(f). The Ni 2p 3/2 spectra can be deconvoluted into two different Ni oxidized states: 853.4 (Ni 2+ ) [17] and 852.8 eV (Ni 0 ) [18]. The binding state of Ni is a slightly positive shift from Ni 0 to Ni 2+ at (3), whereas the binding energy peak of (4) was Ni 0 state. These results indicate that redox reaction occurred at the interface of SnO/Ni. The occurrence of the redox reaction at the interface region is reasonable in terms of the standard reduction potential (Sn: + 0.15, Ni: −0.25 E°at 25°C) and the reported oxidation temperature (300°C) of Ni [19,20]. However, appreciable binding state shift for the depth region of Sn does not happen with the ITO electrode. As shown in Figure (  the binding energy of Sn 2+ and Sn 0+ slightly increased at the interface region of SnO/ITO. This may result from the overlapped Sn binding energy of SnO (Sn 2+ ) and ITO (Sn 4+ and Sn 0+ ).
This chemical reaction at the electrode interface could influence electrical property differences, such as the direction of the effective channel length difference. A decrease in effective channel length ( L has positive values) in the ITO-TFT occurred, which is attributed to electron or oxygen diffusion at the interface, and is generally reported in previous studies [21,22]. In contrast, the increase in effective channel length ( L has negative values) in Ni-TFT is attributed to the redox reaction between the channel and electrode. As shown in Figures 3(a,b), the contact resistivity of Ni electrode TFT (1.58 k ·cm) is lower than the ITO electrode TFT (7.24 k ·cm). The increased effective channel length (2 L = −3.9 μm) in the Ni electrode is not observed in the ITO electrode. The effective channel length increase originated from the SnO/Ni interface redox reaction because the Ni electrode loses conductivity by oxidation to NiO, which has p-type semiconductor properties (Figure 3(c)) [19]. The XPS and TLM results indicated that redox reaction occurred at SnO/Ni interface, but was restrained at SnO/ITO.
We fabricated a diode for measuring contact properties and effective Schottky barrier height (eSBH) [23] ( Figure 4). SnO forms an Ohmic-like contact with the Ni electrode, but it forms a typical Schottky contact with the ITO electrode (Figure 4(a)). To elucidate why the contact types are different, work functions were measured by KPFM. The measured work functions of ITO, SnO, and Ni are 4.42, 4.56, and 4.60 eV, respectively. Figure 4(b) compares the theoretical and experimental SBHs. The experimental SBH is the eSBH value at the diode, while the theoretical SBH value is calculated by band alignment of 300°C annealed SnO as previously reported [1]. The calculated and experimental SBH values of ITO are 0.70 and 0.63 eV, respectively, while those of Ni are 0.52 and 0.32 eV, respectively. The values of the two SBHs are similar for ITO, but the experimental SBH is about 38% lower than the calculated SBH for Ni. This result indicates that holes can be easily injected into the channel layer when an ohmic contact is formed. Overall, lower field-effect mobility with higher contact resistance and a higher energy barrier formation are observed in an ITO electrode TFT compared to Ni. Figure 5 shows a schematic band diagram and carrier conduction mechanism of the SnO TFT at different V G and electrodes. Under negative gate voltage   (−10 V), holes accumulated, and the P-type channel was formed. As shown in Figure 5(a,b), the Ni electrode TFT has lower hole contact resistance with higher field-effect mobility than ITO electrode TFT because of the contact properties and Schottky barrier height (Ni: Ohmic-like, ITO: Schottky). At the positive gate voltage (+5 V), electrons accumulated, and the N-type channel formed ( Figure 5(c,d)). The Ni electrode TFT N-type channel forms more easily because of electrons generated by redox reaction at SnO/Ni interface. Because ITO electrodes restrain the redox reaction at the interface, the off-current at V G = 5V is much higher at the Ni electrode than at ITO electrode ITO electrode TFTs.
Finally, the ITO interfacial layers (ILs) are adopted for lowering the off-current of Ni electrode TFTs by restraining the redox reaction. Figure 6(a) exhibits uniform  Table 2. Interestingly, there were off-current decreases with increased IL thickness, as shown in Figure 6(b). The off-current with the 10 and 15 nm IL was comparable to that of the ITO S/D. As shown in Figures 6(d,e), the redox reaction at the interface of SnO/Ni is effectively restrained by 10 nm ITO IL and lowers the off-current of TFT. The mobility was also slightly decreased as IL thickness increased. The TLM results of 10 nm IL TFT (Figure 6(f)) indicate that ITO IL increases the contact resistance. We agree that 10 nm IL restrain the redox reaction because the effective channel length difference has not occurred. The mobility of 10 nm IL TFT is slightly low compared to that of the Ni S/D ( ∼ 20% lower), but it is still two times higher than that of the ITO electrode. As a result, the I on /I off values were maximal with 10nm-ITO-IL TFTs due to the off-current reduction, as the on-current of these devices resemble those made with Ni ( Figure 6(c)).
As reported earlier, the oxygen vacancy defects of SnO degrade the reliability properties of TFTs [24]. We conducted the negative bias stress test under −10 V gate bias

Conclusions
The origin of high off-current in p-type SnO TFTs with Ni electrodes was examined and then suppressed with ITO interfacial layers. A SnO TFT with Ni electrodes showed a high mobility of 3.3 cm 2 /Vs, high off-current, and a low I on /I off of 3.6 × 10 1 . Compared to the TFT with a Ni electrode, a TFT with an ITO electrode exhibited a lower mobility of 1.4 cm 2 /Vs and a high I on /I off of 1.1 × 10 3 . The difference in field-effect mobility originated from the contact properties and Schottky barrier height (Ni: Ohmic-like, ITO: Schottky). The high offcurrent of the Ni-electrode TFT originates from the redox reaction generated at the interface of the active layer and electrode. The oxygen vacancy generated by redox reactions degrades the NBS reliability in terms of V th shift (−4.0 V), S.S (from 0.43 to 0.95), lowest drain current increase and hump. Controlling the thickness of ITO ILs (optimal at 10 nm) between the active layer and the Ni electrode provided a better I on /I off ratio of 1.1 × 10 3 and a reasonable mobility of 2.5 cm 2 /Vs without degrading other device parameters. ITO ILs could improve the NBS properties such as Vth shift (−1.4 V) without S.S degradation, lowest drain current increase, and hump.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by the National Research Foundation of Korea (NRF) under grant NRF-2020M3H4A3081867.