Electrical Performance and Stability Improvements of High-Mobility Indium–Gallium–Tin Oxide Thin-Film Transistors Using an Oxidized Aluminum Capping Layer of Optimal Thickness

Abstract

We examined the effects of aluminum (Al) capping layer thickness on the electrical performance and stability of high-mobility indium–gallium–tin oxide (IGTO) thin-film transistors (TFTs). The Al capping layers with thicknesses (t_Als) of 3, 5, and 8 nm were deposited, respectively, on top of the IGTO thin film by electron beam evaporation, and the IGTO TFTs without and with Al capping layers were subjected to thermal annealing at 200 °C for 1 h in ambient air. Among the IGTO TFTs without and with Al capping layers, the TFT with a 3 nm thick Al capping layer exhibited excellent electrical performance (field-effect mobility: 26.4 cm²/V s, subthreshold swing: 0.20 V/dec, and threshold voltage: −1.7 V) and higher electrical stability under positive and negative bias illumination stresses than other TFTs. To elucidate the physical mechanism responsible for the observed phenomenon, we compared the O1s spectra of the IGTO thin films without and with Al capping layers using X-ray photoelectron spectroscopy analyses. From the characterization results, it was observed that the weakly bonded oxygen-related components decreased from 25.0 to 10.0%, whereas the oxygen-deficient portion was maintained at 24.4% after the formation of the 3 nm thick Al capping layer. In contrast, a significant increase in the oxygen-deficient portion was observed after the formation of the Al capping layers having t_Al values greater than 3 nm. These results imply that the thicker Al capping layer has a stronger gathering power for the oxygen species, and that 3 nm is the optimum thickness of the Al capping layer, which can selectively remove the weakly bonded oxygen species acting as subgap tail states within the IGTO. The results of this study thus demonstrate that the formation of an Al capping layer with the optimal thickness is a practical and useful method to enhance the electrical performance and stability of high-mobility IGTO TFTs.

1. Introduction

Amorphous oxide semiconductor (AOS) thin-film transistors (TFTs) are of significant interest for applications in active matrix displays, sensors, memories, and other electronic systems because of their excellent electrical properties, good uniformities, and low fabrication costs [1,2,3,4]. Among them, the indium–gallium–zinc oxide (IGZO) TFT, which was first introduced by Nomura et al., in 2004 [5], is the most popular AOS TFT that is currently used for commercial products such as large-area active matrix organic light-emitting diode televisions. However, the field-effect mobility (μ_FE = ~10 cm²/V s) of the IGZO TFT is insufficient to meet the requirements of large area, high-frame rate, and ultra-high-resolution next-generation displays [6,7,8]. Furthermore, μ_FE of the TFT should be higher than that of the IGZO TFT to integrate the scan/driver circuits composed of the TFTs into a glass or flexible substrate. To date, a variety of AOS TFTs having higher μ_FE values than those of the IGZO TFTs have been intensively studied for application in next-generation displays [9,10,11]. Among them, the indium–gallium–tin oxide (IGTO) TFT has recently been attracting special attention as a promising high-mobility AOS TFT, because it exhibits excellent electrical properties even at low annealing temperatures (<200 °C) [12,13,14]. Low temperature processing is exceedingly important for flexible display applications because most plastic substrates have low glass transition temperatures (below 200 °C) [15,16]. In IGTO, the cation Sn is used instead of Zn, because a similar electronic configuration of Sn⁴⁺ and In³⁺ can increase the electron mobility within the AOS by facilitating percolation path formation [17,18,19].

To date, several previous studies have investigated the effects of metal capping layers on the electrical characteristics of AOS TFTs [20,21,22]. Ji et al. examined the effects of titanium (Ti) capping layer formation and subsequent thermal annealing on the electrical properties of indium–zinc oxide (IZO) TFTs [23]. They showed that the Ti capping layer could enhance the μ_FE and positive bias stress (PBS) stability of IZO TFTs. Kim et al. compared the transfer characteristics of zinc oxynitride (ZnON) TFTs with and without the tantalum (Ta) capping layer and observed that the Ta capping layer could increase μ_FE of the ZnON TFTs [24]. Recently, Park et al. showed that the aluminum (Al) capping layer could enhance μ_FE of IGZO TFTs even without the subsequent post-deposition thermal annealing process [25]. However, there has been no study on the optimal thickness of the metal capping layer for AOS TFTs considering both the electrical performance and stability of the devices. In this study, we investigated the effects of Al capping layer thickness (t_Al) on the electrical characteristics, PBS stability, and negative bias illumination stress (NBIS) stability of high-mobility IGTO TFTs. Al capping layers with t_Al values of 3, 5, and 8 nm were deposited on top of the IGTO thin films using the electron beam evaporation technique and the IGTO TFTs without and with Al capping layers were thermally annealed at 200 °C for 1 h in ambient air. From the experimental results, it was concluded that the 3 nm thick Al capping layer was most advantageous for improving the electrical performance and stability of IGTO TFTs. A systematic study was performed to analyze the effects of the Al capping layer with each thickness on the electrical characteristics and stabilities of IGTO TFTs.

2. Experimental Details

The IGTO TFTs were fabricated on a heavily doped p-type silicon wafer (resistivity < 0.005 Ω cm) covered by 100 nm thick thermally grown SiO₂ with the bottom-gate configuration. A 15 nm thick IGTO thin film was deposited via direct current (DC) magnetron sputtering using a 3 inch IGTO target. The working pressure and Ar/O₂ ratio were maintained at 3 mTorr and 35/15 (sccm/sccm), respectively, during the deposition process. The DC power was fixed at 150 W, and the substrate was not intentionally heated. A 100 nm thick indium–tin oxide (ITO) was deposited on the IGTO/SiO₂/Si by DC magnetron sputtering for the source and drain electrodes of the TFTs. During the sputtering of the ITO, the DC power was fixed at 150 W and the working pressure was maintained at 3 mTorr under an Ar atmosphere. Subsequently, Al capping layers with t_Al values of 3, 5, and 8 nm were deposited on top of the IGTO channel layer by electron beam evaporation, respectively. Finally, the fabricated IGTO TFTs without and with Al capping layers were thermally annealed at 200 °C for 1 h in ambient air. All layers were patterned using photolithography and lift-off processes. Figure 1a and b display the schematic view and optical image of the fabricated Al-capped IGTO TFTs, respectively. All IGTO TFTs in this study were designed to have a channel width/length (W/L) of 500/500 (μm/μm). The chemical compositions of the IGTO thin films covered by Al capping layers of different thicknesses were examined by X-ray photoelectron spectroscopy (XPS, K-alpha+, Thermo Scientific). The work-function (Φ) of the deposited IGTO thin film was determined by Kelvin probe force microscopy (SKP5050, KP Technology), and the optical bandgap (E_g) of the IGTO thin film was calculated via ultraviolet visible-near infrared (UV-vis-NIR, wavelength: 300–1400 nm) spectroscopy (V-670, JASCO) using a Tauc plot. The electrical characteristics of the IGTO TFTs were measured using a semiconductor parameter analyzer (4156C, Agilent Technologies) in the dark at room temperature in a vacuum environment to avoid environmental ambient effects on the electrical characteristics and stabilities of the IGTO TFTs.

3. Results and Discussion

Figure 2a and b depict the representative transfer curves of the IGTO TFTs without and with 3, 5, and 8 nm thick Al capping layers on a semi-logarithmic (Figure 2a) and linear scale (Figure 2b). In Figure 2, I_D, V_GS, and V_DS represent the drain current, gate-source voltage, and drain-source voltage, respectively. Electrical characterization was conducted by sweeping V_GS from −20 to 20 V at V_DS = 1.0 V for all IGTO TFTs.

Table 1. Electrical parameters extracted from the indium–gallium–tin oxide (IGTO) thin-film transistors (TFTs) without and with 3, 5, and 8 nm thick Al capping layers.

tAl [nm]	VTH [V]	μFE [cm2·V−1·S−1]	SS [V/decade]	ION/OFF
0 (w/o Al)	−0.1	25.9	0.42	3.51 $\times {10}^7$
3	−1.7	26.4	0.20	5.52 $\times {10}^8$
5	−3.7	27.6	0.37	3.90 $\times {10}^7$
8	−14.0	-	1.31	2.50 $\times {10}^6$

lists the electrical parameters extracted from the IGTO TFTs without and with 3, 5, and 8 nm thick Al capping layers. The threshold voltage (V_TH) was defined as the V_GS value giving I_D = W/L × 10⁻⁸ (A), and the field-effect mobility (µ_FE) was extracted from the maximum transconductance at V_DS = 1.0 V using Equation (1):

$${\mu }_{FE}=\frac{L{g}_m}{W{C}_i{V}_{DS}}$$

(1)

where C_i and g_m are the gate dielectric capacitance per unit area and transconductance, respectively. The subthreshold swing (SS) was calculated from the semi-logarithmic scale transfer curve using Equation (2).

$$SS=\frac{d{V}_{GS}}{d\left(log{I}_D\right)}$$

(2)

The on/off current ratio (I_ON/OFF) was defined as the ratio between the maximum value of the on-current and the minimum value of the off-current for V_GS ranging from −20 to 20 V at a V_DS of 1.0 V. From Figure 2, it is observed that the IGTO TFTs without an Al capping layer and those having Al capping layers with t_Als of 3 and 5 nm exhibit normal transfer characteristics; however, the TFT with t_Al of 8 nm shows nonideal double slope characteristics in the transfer curve. μ_FE of the IGTO TFT increases and V_TH shifts in the negative direction with an increase in t_Al. The IGTO TFT with a 3 nm thick Al capping layer exhibits the minimum value of SS (0.2 V/dec) with satisfactory values of μ_FE (26.4 cm²/ V s) and V_TH (−1.7 V). The obtained results clearly demonstrate that the thickness of the Al capping layer significantly influences the electrical performance of IGTO TFTs.

Figure 3a–d depict the time dependences of transfer curves for IGTO TFTs without and with 3, 5, and 8 nm thick Al capping layers, respectively, under a PBS of V_OV = 20 V, where V_OV = V_GS – V_TH. Figure 3e summarizes the V_TH shift (ΔV_TH) obtained from IGTO TFTs without an Al capping layer and those with Al capping layers having different t_Al values after each PBS time. The transfer curves shift in the positive direction for all IGTO TFTs as the PBS time increases. From Figure 3, it is observed that the formation of the Al capping layer effectively reduces the magnitude of ΔV_TH after PBS in IGTO TFTs. However, no significant difference was observed in the PBS stabilities of the Al-capped IGTO TFTs having t_Al values of 3, 5, and 8 nm. Figure 4a–d and e depict the time dependence of transfer curves and ΔV_TH for IGTO TFTs without and with 3, 5, and 8 nm thick Al capping layers under an NBIS of V_OV = −20 V and a white light (luminance: 2000 lx). Transfer curves shift in the negative direction for all IGTO TFTs as the NBIS time increases. Figure 4 shows that the magnitudes of ΔV_TH after being exposed to NBIS have similar values for the uncapped IGTO TFT and that having a 3 nm thick Al capping layer. However, the transfer curves shift in the negative direction more significantly when t_Al is greater than 3 nm. Figure 3 and Figure 4 show that the thickness of the Al capping layer strongly influences not only the electrical characteristics, but also the PBS and NBIS stabilities of the IGTO TFTs.

To elucidate the physical mechanism responsible for the observed phenomena, we compared the O1s spectra of the IGTO thin films without and with 3, 5, and 8 nm thick Al capping layers using XPS. Figure 5a–d depict the XPS O1s spectra of the IGTO thin films without and with 3, 5, and 8 nm thick Al capping layers, respectively. The XPS spectra were obtained for the middle sections of each IGTO thin film. The three sub-peaks at 530.0, 531.0, and 532.5 eV deconvoluted from the XPS spectra were assigned to the fully-coordinated metal ions (metal–oxygen lattice) (O_I), oxygen deficiencies (O_II), and weakly bonded oxygen components such as an interstitial oxygen and a hydroxyl group (O_III), respectively [19]. Figure 5e displays the relative areas of each sub-peak corresponding to O_I, O_II, and O_III, obtained from the IGTO thin films without and with Al capping layers. The XPS results in Figure 5 show that the weakly bonded oxygen-related components decreased from 25.0 to 10.0%, whereas the oxygen-deficient portion was maintained at 24.4% after the formation of the 3 nm thick Al capping layer. In contrast, a significant increase in the oxygen-deficient portion was observed after the formation of Al capping layers with t_Al values greater than 3 nm (24.4% for uncapped IGTO thin film and that having a 3 nm thick Al capping layer, 26.5% for IGTO thin film with t_Al value of 5 nm, and 34.1% for IGTO thin film with t_Al value of 8 nm). These results imply that the thickness of the Al capping layer strongly affects the chemical composition of the underlying IGTO channel layer. The Gibbs free energies of formation (ΔG_f) for In₂O₃, Ga₂O₃, SnO₂, and Al₂O₃ are −830.7, −998.3, −516.0, and −1582.3 kJ/mol, respectively [25,26], and the lowest ΔG_f value corresponding to that of Al₂O₃ implies that Al has a stronger oxidation power than In, Ga, and Sn. Therefore, the Al capping layer causes a reduction in the IGTO channel and becomes aluminum oxide (AlO_X) after thermal annealing, which eliminates the weakly bonded oxygen such as an interstitial oxygen and a hydroxyl group and the lattice oxygen bonded to In, Ga, and Sn from the IGTO. The results in Figure 5 show that the gathering power of the oxygen species strongly depends on the thickness of the Al capping layer. The breaking of the cation–oxygen bonds requires a high activation energy; therefore, the 3 nm thick, thin Al layer eliminates only the weakly bonded oxygen such as an interstitial oxygen and a hydroxyl group. However, the Al capping layers thicker than 3 nm have a stronger gathering power of the oxygen species; therefore, the lattice oxygen is also eliminated from the IGTO after thermal annealing. The weakly bonded oxygen such as an interstitial oxygen and a hydroxyl group generates acceptor-like trap states near the conduction band (CB) edge and enhances electron trapping during the application of PBS in IGTO [27]. Therefore, a higher value of μ_FE, smaller value of SS, and better PBS stability of the IGTO TFT with a 3 nm thick Al capping layer can probably be attributed to the lower concentration of the weakly bonded oxygen species within the IGTO channel layer than the IGTO TFT without a capping layer. Oxygen vacancies (V_O) generate shallow and deep donor states within the IGTO channel layer [28,29]. Shallow donor states provide electrons to the CB; therefore, the electron concentration increases as the number of V_O increases within the IGTO channel layer. The increase in the electron concentration enhances the formation of the percolation conduction path in IGTO and makes it difficult to turn off the TFT [30,31,32]. In addition, a higher carrier concentration within the channel increases the SS value of the TFT [33,34]. Therefore, a lower value of V_TH and higher values of μ_FE and SS in the 5 and 8 nm thick Al-capped IGTO TFTs than in the uncapped and 3 nm thick Al-capped IGTO TFTs can be attributed to the larger concentration of V_O within the IGTO channel layer caused by the stronger oxidation power of the thicker Al capping layer. The poorest NBIS stability exhibited by TFTs with t_Al values of 5 and 8 nm can also be attributed to the high density of V_O within the IGTO channel, because the doubly ionized V_Os are generated from V_O and diffuse toward the gate dielectric/channel interface under NBIS [35,36].

The non-ideal double slope transfer characteristics of the Al-capped IGTO TFT with t_Al of 8 nm are considered to be a result of the partial oxidation of the Al capping layer during thermal annealing. Figure 6 displays the time-of-flight secondary ion mass spectrometry (ToF-SIMS) negative ion depth profile for AlO₂- ions obtained from the 8 nm thick Al capping layer in the fabricated Al-capped IGTO TFT. Figure 6 shows that the AlO₂-signal exhibits high values at the surface and near the IGTO channel layer; however, relatively lower values of the AlO₂-signal are observed between these two regions. Considering that the AlO₂-ion is a representative of AlO_X, this implies that the 8 nm thick capping layer is composed of three regions: AlO_X (near the surface), Al, and AlO_X (near IGTO), as depicted in Figure 6. The AlO_X layers at the surface and near the IGTO layer are likely to be formed as a result of the oxidation of the Al capping layer by oxygen in the environmental air and that in the IGTO channel layer, respectively. Figure 7 illustrates the schematic energy band diagram for the Al-AlO_X-Si MOS capacitor, where the Φ and E_g values of the IGTO were determined to be 4.95 and 3.88 eV, respectively. Considering that the Φ value of Al (4.0 ~ 4.28 eV) [37,38] is smaller than that of the IGTO, the accumulation layer can be formed at the interface between the AlO_X and IGTO, as observed in Figure 7, which can act as the parasitic channel causing the non-ideal double slope transfer characteristics of the Al-capped IGTO TFT with t_Al of 8 nm.

Figure 8a displays the technology computer-aided design (TCAD) simulation results for the transfer characteristics of the IGTO TFT with and without an Al/AlO_X stacked layer on top of the IGTO channel layer. The simulation results show that the transfer characteristics of the IGTO TFT with an Al/AlO_X stacked layer exhibit non-ideal double slope transfer characteristics. Figure 8b depicts the TCAD simulation results for the conduction band energy in the IGTO TFT with an Al/AlO_X stacked layer along A-A’ in the inset figure. The TCAD simulation results in Figure 8b show that the main channel is formed near the top AlO_X layer at low V_GS’s, but is formed near the bottom SiO₂ layer at high V_GS’s. These results show that the parasitic channel formed between the AlO_X and IGTO is the reason for the non-ideal double slope transfer characteristics of the Al-capped IGTO TFT with t_Al of 8 nm.

4. Conclusions

In this work, we studied the effect of Al capping layer thickness on the electrical performance and stability of high-mobility IGTO TFTs. An Al capping layer with t_Al values of 3, 5, and 8 nm were deposited on top of the IGTO thin film and the fabricated IGTO TFTs without and with Al capping layers were thermally annealed at 200 °C for 1 h in air. Among the IGTO TFTs, the TFT with a 3 nm thick Al capping layer exhibited excellent electrical properties and better electrical stability under PBS and NBIS than the other TFTs. The IGTO TFT without an Al capping layer exhibited a lower value of μ_FE, higher value of SS, and inferior PBS stability compared with the IGTO TFT with a 3 nm thick Al capping layer. The IGTO TFTs with 5 and 8 nm thick Al capping layers exhibited significantly lower values of V_TH, higher values of SS, and poorer NBIS stabilities compared with the IGTO TFT with a 3 nm thick Al capping layer. The observed phenomenon was attributed to the thickness dependent oxidation power of the Al capping layer. The oxidation power of the Al capping layer increased with increasing t_Al and 3 nm was found to be the optimum thickness of the Al capping layer, which can selectively eliminate the weakly bonded oxygen species from the IGTO channel layer without significantly generating V_O within the IGTO. Our experimental results show that the electrical performance and stability of IGTO TFTs can be effectively improved by forming an Al capping layer with optimal thickness.