Electrical and Hysteresis Characteristics of Top-Gate InGaZnO Thin-Film Transistors with Oxygen Plasma Treatment Prior to TEOS Oxide Gate Dielectrics

Abstract

We report the impact of oxygen (O₂) plasma time on an amorphous indium–gallium–zinc oxide (a-IGZO) thin-film surface that was carried out before TEOS deposition in order to optimize the performance of thin-film transistors (TFTs). TheO₂ plasma time of 60 s possessed the largest on/off current ratio of >10⁸, with a field-effect mobility (µ_FE) of 8.14 cm² V⁻¹ s⁻¹, and the lowest subthreshold swing (S.S.) of 0.395 V/decade, with a threshold voltage (V_th) of −0.14 V. However, increases in I_off and S.S. and decreases in the µ_FE were observed for the longer O₂ plasma time of 120 s. As the O₂ plasma time increased, the reduction in the carrier concentration in the IGZO channel layer may have resulted in an increase in V_th for the IGZO TFT devices. With an increase in the O₂ plasma time, the surface roughness of the IGZO channel layer was increased, the carbon content in the TEOS oxide film was reduced, and the film stoichiometry was improved. The SIMS depth profile results showed that the O/Si ratio of TEOS oxide for the sample with the O₂ plasma time of 60 s was 2.64, and its IGZO TFT device had the best electrical characteristics. In addition, in comparison to the IGZO TFT device without O₂ annealing, larger clockwise hysteresis in the transfer characteristics revealed that a greater number of electrons were trapped at the interface between TEOS oxide and the a-IGZO channel layer. However, hysteresis curves of the O₂-annealed IGZO TFTs with various O₂ plasma times were greatly reduced, meaning that the electron traps were reduced by O₂ annealing.

1. Introduction

Due to fast advances in optoelectronic technology, the next generation of displays will have thinner, lighter, and bigger screens. High-performance thin-film transistors (TFTs) have received a lot of interest [1,2,3,4,5,6]. Because of their great field-effect mobility (μ_FE) (>10 cm² V⁻¹ s⁻¹) and lower processing temperatures, amorphous oxide semiconductors (AOSs) are suitable for TFT channel materials [7,8] and make it possible to drive TFTs in active-matrix organic light-emitting diode (AMOLED) displays. For amorphous indium–gallium–zinc oxide (a-IGZO) TFTs, the traditional bottom gate structure has been researched extensively [9,10] in our previous studies. However, due to large parasitic capacitance and low scalability, this structure is inappropriate to integrate into the system peripheral circuits on the glass and implement superior AMOLED displays. As a result, self-aligned top-gate a-IGZO TFTs devices with high performance and good stability are required for AMOLED display applications [11,12].

In general, the electrical properties of a-IGZO TFTs, including V_th, subthreshold swing (S.S.), μ_FE, and I_on/I_off current ratio, can clearly be controlled by varying the partial pressure of oxygen (O₂) during a-IGZO channel layer deposition with radio frequency sputtering [9,13]. Jeong et al. [14] reported that O₂ plasma treatment on the a-IGZO channel layer decreases the oxygen vacancy density of the IGZO films at low processing temperatures, improving the IGZO channel layer material and the performance of IGZO TFT devices. Pu et al. reported [15] the effects of O₂ plasma treatment on solution-processed a-IGZO TFTs. The plasma-treated devices not only showed greater mobility and a greater on/off current ratio, but also a monotonic increase in the SS. with the plasma time. The effects of several plasma treatments on a-IGZO TFTs, including O₂, nitrogen (N₂), and argon (Ar), were observed [16] by Kim et al. They used X-ray photoelectron spectroscopy (XPS) to investigate oxygen stoichiometry in a-IGZO TFTs in diverse plasma conditions. The results indicated that O₂ and N₂ plasmas decreased oxygen vacancies, but Ar plasma treatment increased them. Liu et al. recently reported [17] on IGZO thin films that were treated with an Ar/O₂ plasma combination with various oxygen content ratios, resulting in the bottom-gate IGZO TFTs have improved device operation characteristics. Wang et al. reported [18] the O₂ plasma treatment on gate dielectrics (Al₂O_x) for a-IGZO TFTs and demonstrated that O₂ plasma treatment can improve the gate dielectric performance. The results were ascribed to passivating the interfacial and bulk traps by reducing the oxygen vacancies. Most notably, this was the first study to our knowledge that focused on the effects of the O₂ plasma treatment prior to the tetraethyl–orthosilicate (TEOS) oxide gate dielectric deposition and O₂ annealing on the electrical characteristics of top-gate IGZO TFTs, wherein the compositions of TEOS oxide interfaces with a-IGZO films were reported.

In this study, top-gate IGZO TFT devices with various O₂ plasma times prior to TEOS gate dielectric and O₂ annealing after TEOS oxide deposition were fabricated using the same plasma-enhanced chemical vapor deposition (PECVD) system. Moreover, the electrical and hysteresis characteristics of the IGZO TFT devices using O₂ annealing after TEOS oxide deposition were evaluated. Additionally, atomic force microscope (AFM) and X-ray photoelectron spectroscopy (XPS) tests were performed to investigate the surface morphologies and compositions of the IGZO films. A depth profile analysis of the secondary ion mass spectrometer (SIMS) was conducted to verify the compositions of TEOS oxide interface with a-IGZO films treated by different O₂ plasma times. The best μ_FE of 8.14 cm² V⁻¹s⁻¹, S.S. of 0.395 V/decade, V_th of −0.14 V, and lower ∆V_th of 0.5 V from the hysteresis curve were obtained in the IGZO TFTs that had been subjected to O₂ plasma treatment for 60 s and O₂ annealing.

2. Experimental Methods

The cross-sectional schematic of the top-gate IGZO TFT structure employed in this investigation is shown in Figure 1a. Details of the devices’ fabricating processes can be found elsewhere [19]. The O₂ plasma time was varied in order to verify the effects of O₂ plasma prior to depositing the TEOS gate dielectric on the electrical and hysteresis characteristics of the IGZO TFT devices. The O₂ plasma treatment was performed at various treatment times using an Oxford PECVD system prior to depositing TEOS oxide at 350 °C at a fixed deposition power of 40 W. The system working pressure was 700 m Torr with an O₂ flow of 250 sccm. After TEOS oxide deposition, the samples were annealed at 400 °C at a working pressure of 5 mTorr in an O₂ ambient atmosphere of 40 sccm for 1 h in the same PECVD system. Figure 1b summarizes the process charts for each step of the top-gate IGZO TFTs employed in this investigation. For the I_DS-V_GS transfer characteristics and I_GS-V_GS, V_GS varied from −15 V to +15 V in 0.3 V steps for V_DS = 0.1 V in the tested devices. The Agilent 4156A precision semiconductor parameter analyzer was employed to thoroughly evaluate and compare the electrical and hysteresis characteristics of the tested devices, and the measurement temperature was maintained at 25 °C.

In this experiment, the channel width W = 100 µm and the channel length L = 5 µm were used. Figure 1a shows the source and drain offsets, which were designed at 1 µm. The µ_FE was calculated using the following equation:

$${\mu }_{FE}=\frac{L\times G_{m\left(max\right)}}{W\times C_{ox}\times V_{DS}}$$

(1)

where C_ox is the TEOS oxide capacitance per unit area and the maximum transconductance G_m(max) = ($\partial$ I_DS/$\partial$ V_GS)_max at V_DS = 0.1 V. The S.S. was taken as the value of (dlog (I_DS)/d V_GS)⁻¹ when the I_DS was in the range of 10⁻¹⁰–10⁻⁹ A.

Film thickness was examined and fitted using the spectroscopic ellipsometer (SE) (M-2000U) and CompleteEASE software (J. A. Woollam Co., Inc., Lincoln County, NE, United States), respectively. Surface morphology tests were employed using an atomic force microscope (AFM, Veeco Dimension 3100) at a scan area of 2 × 2 μm. X-ray photoelectron spectroscopy (XPS, Thermo Fisher Scientific Theta Probe (Waltham, MA, USA)) was also employed to characterize the composition of the IGZO films. The depth profile analysis of the secondary ion mass spectrometer (SIMS, CAMECA IMS 7F) was conducted to verify the compositions of the TEOS oxide interface with a-IGZO films treated with O₂ plasma. In order to compare the effect of O₂ annealing, we prepared the TEOS oxide on the a-IGZO film without O₂ annealing with an O₂ plasma time of 60 s to determine the hysteresis characteristics.

3. Results and Discussion

3.1. Surface Roughness Analysis of a-IGZO Films at Various O₂ Plasma Times

Figure 2a–c compares the AFM surface morphology characterizations of the a-IGZO channel layers with different O₂ plasma times, and their z range, root mean square (RMS), and average roughness (Ra) are summarized in

Table 1. Z range, root mean square (RMS) roughness, and average roughness (Ra) of the a-IGZO films at various O₂ plasma times.

O2 Plasma Time (s)	Z Range (nm)	RMS (nm)	Ra (nm)
0	8.329	0.671	0.458
60	7.877	1.049	0.817
120	11.14	1.857	1.531

. Plasma bombardment of the channel surface is generally believed to increase surface roughness, as shown in Figure 2. Although the RMS and Ra roughness monotonically increased with the increase in the O₂ plasma time, the z range of the a-IGZO films with the O₂ plasma times of 0 and 60 s was lower than that of the O₂ plasma time of 120 s. The deteriorated electrical properties of the IGZO TFTs were found to be reasonably useful for ascribing the cause of the longer O₂ plasma time to the surface roughness in a later section following the presentation of results of the device characteristic analysis.

3.2. Electrical Characteristics of IGZO TFTs at Various O₂ Plasma Times

Figure 3 shows the typical I_DS-V_GS transfer characteristics and I_GS-V_GS curves of the IGZO TFTs at various O₂ plasma times prior to the deposition of the TEOS oxide gate dielectrics. V_th is set as the gate voltage (V_GS) when the drain current (I_DS) is at a constant value (10⁻⁹ × (W/L) A). Measurements were performed with more than three samples per device. The measured capacitance per unit area (C_ox) of the gate dielectric was 4.98 × 10⁻⁸ F/cm² in this study.

Table 2. Electrical parameters extracted from the IGZO TFTs at various O₂ plasma times.

O2 Plasma Time (s)	Vth (V)	S.S. (V/Decade)	μFE (cm2 V−1s−1)	Ion (A)	Ioff (A)	ρ2.5V (Ω·cm)
0	−5.74 ± 0.31	1.045 ± 0.036	1.95 ± 0.31	2.4 × 10−6	3.0 × 10−14	6.26
60	−0.14 ± 0.08	0.395 ± 0.031	8.14 ± 0.86	1.3 × 10−6	1.1 × 10−14	7.64
120	5.14 ± 0.51	2.210 ± 0.204	2.98 ± 0.31	1.3 × 10−6	4.4 × 10−12	8.26

shows the average values and standard variations in the electrical parameters extracted from the IGZO TFTs at various O₂ plasma times. It is clearly confirmed in Figure 3 that the transfer curves shifted positively as the O₂ plasma time increased as a result of the lower on-current (I_on) and higher off-current (I_off) within the IGZO TFT channel. In addition, a compliance cut-off phenomenon that was clearly visible in the transfer curves at the O₂ plasma times of 60 and 120 s was attributed to a higher total resistance between the source and drain electrodes as the O₂ plasma time increased. Moreover, the gate leakage current (I_GS) gradually increased from 10⁻¹³ to 10⁻¹² A, which occurred as the O₂ plasma time increased. As shown in Table 1, the a-IGZO film’s roughness monotonically increased as the O₂ plasma time increased; therefore, the gate leakage current (I_GS) of the IGZO TFTs gradually increased by one order of magnitude from when the TEOS gate oxide was deposited after the O₂ plasma treatment. From the summary in Table 2, we determined that the O₂ plasma time of 60 s possessed the largest on/off current ratio of >10⁸, a µ_FE of 8.14 cm² V⁻¹ s⁻¹, and the lowest S.S. slope of 0.395 V/decade and a V_th of −0.14 V. However, increases in I_off and S.S. and degradation in µ_FE were observed at the longer O₂ plasma time of 120 s. It is reasonable to ascribe higher I_off and S.S. deterioration to the higher surface roughness of the a-IGZO films after an O₂ plasma time of 120 s, as shown in Figure 2. The effects of the various O₂ plasma times and O₂ annealing on the hysteresis characteristics are discussed below.

Figure 4 shows the typical output characteristics of the IGZO TFTs at various O₂ plasma times under different gate overdrives: V_GS-V_th = 3–11 V. It can be seen that the saturation current at the same gate overdrive clearly decreased as the O₂ plasma time increased. All of the devices demonstrated ohmic behavior, indicating good contact between the a-IGZO channels and Mo electrodes. In a linear current model of a conventional metal oxide semiconductor field-effect transistor (MOSFET), the resistivity (ρ) is expressed by the following equation [16,20]:

$$\rho =\frac{V_{DS}}{I_{DS}}\frac{Wt}{L} \approx \frac{1}{{\mu }_{FE}{C}_{ox}\left(V_{GS}-V_{th}\right)} \approx \frac{1}{nq{\mu }_{FE}}$$

(2)

where C_ox is the oxide capacitance per unit area; n is the charge carrier concentration (cm⁻³); q is the electron unit charge; W, L, and t are the width, length, and thickness of the channel, respectively; and V_DS is the drain-source voltage. It should be noted that ρ is primarily associated with V_th, and its corresponding n is determined according to Equation (2). Table 2 indicates that ρ gradually increased from 6.26 to 8.26 Ω·cm as the O₂ plasma treatment time increased at V_DS = 2.5 V and V_GS-V_th = 11 V. Accordingly, the O₂ plasma treatment mainly affected the changes in n in the channel, and n was reduced by exposure to the O₂ plasma since it has already been demonstrated that oxygen atoms can fill oxygen vacancies after O₂ treatments in metal oxide-based semiconductors [15,16,21]. In addition, it is clearly confirmed in Figure 3 that the transfer curves shifted positively as the O₂ plasma time increased. Details of the changes in the carrier concentrations are presented in a later section and are based on the results of the XPS analysis.

The chemical characteristics of the a-IGZO channel layers treated with different O₂ plasma times were investigted by analyzing the XPS results. The effect of the O₂ plasma time on oxygen scarcity after plasma processing was studied extensively by compaing and analyzing from the O1s XPS data displayed in Figure 5. Figure 5a–c shows the XPS spectra of the O1s peak of the a-IGZO films with the O₂ plasma times of 0, 60, and 120 s, respectively. Generally, for IGZO TFTs, the O1s spectra are composed of double Gaussian distributions, the O_I peak is observed at 530.2 eV and the O_II peak is observed 531.6 eV [15,16,22,23]. The O_I XPS signal peak shows covalent oxygen ion bonds with In, Ga, and Zn cations: and the O_II XPS signal peak represents oxygen scarcities in the IGZO film. As a result, the integrated peak ratio of the O_II/(O_I+O_II) intensity ratio of the O1s core levels of the a-IGZO channel layer was employed to estimate the quality of the channel layer and the number of oxygen scarcities in the a-IGZO channel layer with O₂ plasma treatment. The estimated XPS O_II/(O_I+O_II) intensity ratios for the a-IGZO channel layer with increased O₂ plasma times of 0, 60, and 120 s were 0.44, 0.32, and 0.21, respectively. When O₂ plasma treatment was performed for a longer time, the low O_II/(O_I+O_II) XPS intensity ratios indicated a significant reduction in the oxygen scarcities and greater quality of the IGZO channel layer. Because oxygen vacancies are important for electron donation in the a-IGZO channel layer, the decreased O_II peak intensity shown in the XPS analyses corresponded well with the reduced carrier concentration, which could be ascribed to a higher resistivity as a result of the increased O₂ plasma time, as shown in Table 2. Therefore, as the O₂ plasma time increased, the reduced carrier concentration in the IGZO channel layer may have resulted in an increase in the threshold voltage (V_th) of the IGZO TFT devices [13,17,24].

For the analysis of the SIMS depth profile results, we fabricated test samples with a 50 nm thick Mo/100 nm thick TEOS oxide/40 nm thick a-IGZO structure on silicon substrates that were the same as the structure of the top-gate IGZO TFT devices. First, the a-IGZO film was deposited on the silicon substrate, and then, it was subjected to O₂ plasma treatment on the a-IGZO film prior to the depositing of the TEOS oxide and O₂ annealing after TEOS oxide deposition in the same PECVD system. The SIMS analysis is able to observe information on the impurity content and compositions of the TEOS oxide interface with a-IGZO films treated with O₂ plasma. The area located 100–150 nm from the sample’s surface was found to be related to the area across the TEOS oxide near the a-IGZO films. Figure 6a–c shows the SIMS depth profiles of the Mo, Si, O, C, In, Ga, and Zn elements for the test samples at various O₂ plasma times. From analyzing the element intensities, we found that the ratios of the elements Mo, In, Ga, and Zn remained essentially unchangeable for these three samples. Figure 6a shows that the TEOS oxide film on the sample without O₂ plasma treatment contained more carbon (3.91%) than the others. Because the TEOS oxide film reaction provided intermediate elements, such as the carbon impurity in the film caused by the inadequate dissociation of TEOS oxide, the carbon dissociation inside the TEOS oxide films of the O₂ plasma-treated samples was enhanced. With the increased O₂ plasma times of 60 and 120 s, the ratio of the carbon content in the TEOS oxide film decreased from 3.91% to the ratios of 2.42% and 2.71%, respectively, as shown in Figure 6a–c. Rhee et al. reported that the use of oxygen gas plasma treatment during TEOS oxide gate dielectric deposition was efficient in removing the impurities and in the densification of the film. The carbon impurity was reduced by cyclic O₂ plasma treatment [25]. Since TEOS oxide is affected by O₂ plasma treatment, the Si and O contents of the TEOS oxide film clearly changed as the O₂ plasma time increased. The O/Si ratios of TEOS oxide gradually increased from 1.15 to the ratios of 2.64 and 3.54 with increased O₂ plasma times of 60 and 120 s, respectively, as shown in

Table 3. SIMS analytical results for the Si, O, C, and O/Si ratio of the test samples at various O₂ plasma times.

O2 Plasma Time (s)	Si (%)	O (%)	C (%)	O/Si
0	44.66	51.41	3.91	1.15
60	26.77	70.80	2.42	2.64
120	21.40	75.88	2.71	3.54

. This is because the impurities on the surface of the a-IGZO channel layer were mainly removed after O₂ plasma time of 60 s and because the low amount of O₂⁺ in the plasma has a tendency to catch electrons from shallow areas [15]. Therefore, applying the O₂ plasma treatment before gate dielectric deposition would affect the composition of the gate dielectric. In addition, increasing the O₂ plasma time to 120 s may allow the O₂⁺ accumulating on the surface to provide more oxygen atoms for TEOS gate dielectrics. In this experiment, we observed that the O/Si ratio of TEOS oxide of the sample with the O₂ plasma time of 60 s was 2.64, and its IGZO TFT device had the best electrical characteristics.

Because the conduction channel layer is mainly formed in a very thin layer near the interface, the electrical properties of a TFT are determined by the interface roughness between the a-IGZO and the TEOS oxide gate dielectric. The existence of defect states that may lead to electron trapping primarily affects the S.S. and µ_FE of the a-IGZO channel layer [26,27].

For the sample with an O₂ plasma time 60 s, a slight increase in the surface roughness of the a-IGZO channel layer was observed based on the AFM morphology, but the impurities on the surface of the a-IGZO channel layer were largely removed. In this case, the reaction of atomic oxygen and oxygen vacancies in the channel layer increased the µ_FE of the IGZO TFTs. Reductions in the oxygen vacancies and carbon content by the XPS and SIMS analyses, respectively, were observed. It has been reported that filling oxygen vacancies can enhance µ_FE because interface traps can act as scattering centers [28]. Similarly, the interfacial roughness of the sample with an O₂ plasma time of 120 s increased three times, leading to enhanced carrier scattering and thus reduced fieled effect mobility. In addition, the accumulation of O₂⁺ on the surface might clog the micropores and prevent the atoms from reaching deep into the channel layer as the O₂ plasma time increases, resulting in large interfacial traps and degraded S.S. [15].

3.3. Effect of O₂ Annealing on Hysteresis Characteristics of IGZO TFTs at Various O₂ Plasma Times

We investigated the hysteresis sweeps of the IGZO TFTs at various O₂ plasma times prior to TEOS oxide deposition, thus studying the impacts of O₂ annealing following TEOS oxide deposition and their interaction with the a-IGZO channel layer. Figure 7a–c shows the hysteresis curves of the IGZO TFTs at the O₂ plasma times of 0, 60, and 120 s, respectively, with O₂ annealing being represented as V_GS ranging from −15 V to +15 V. The hysteresis curve of the IGZO TFT with the O₂ plasma time of 60 s without O₂ annealing and represented as V_GS ranging from −15 V to +15 V is shown in comparison with the effect of O₂ annealing in Figure 7d. Figure 7 shows that the on-current appeared to be rather constant in the saturation region due to a compliance cut-off phenomenon, which was attributed to a higher total resistance between the source and drain electrodes after the O₂ plasma treatment. In the lower threshold voltage shifts (∆V_th = 0.34, 0.5, and 0.51 V) estimated from the individual hysteresis curves shown in Figure 7a–c with O₂ annealing after TEOS oxide deposition, a small number of electrons were trapped at or close to the interface of the TEOS oxide/a-IGZO channel layer [19]. The largest threshold voltage shift (∆V_th = 3.05 V) from the hysteresis curve without O₂ annealing revealed that a greater number of electrons were trapped at the interface between TEOS oxide and the a-IGZO channel layer. However, the lower ∆V_th from the hysteresis curves displayed in Figure 7a–c shows that O₂ annealing after TEOS oxide deposition successfully limited the formation in interface traps and charge trapping between the TEOS oxide gate dielectrics and the IGZO channel layer. It has been already shown that the a-IGZO channel regions were patterned and thermally annealed in the O₂ atmosphere or in air, and the hysteresis disappeared for the annealed bottom-gate IGZO TFTs, which suggests that thermal annealing reduces electron traps [29,30].

4. Conclusions

In this study, O₂ plasma treatment was found to increase the surface roughness of the IGZO channel layer. The O₂ plasma time of 60 s possessed the largest on/off current ratio of >10⁸, with a μ_FE of 8.14 cm² V⁻¹ s⁻¹, and the lowest S.S. slope of 0.395 V/decade, with a V_th of −0.14 V. However, increases in I_off and S.S. and degradation in µ_FE were observed for the longer O₂ plasma time of 120 s. It is reasonable to ascribe the higher I_off and deteriorated S.S. to the higher surface roughness of the a-IGZO channel layer for a longer O₂ plasma time of 120 s. The XPS observations of the O_II peak intensity coincided well with the lower carrier concentration due to larger resistivity with the increasing O₂ plasma time. Therefore, the decrease in the carrier concentration of the IGZO channel layer may lead to an increase in V_th of the IGZO TFT device with the increasing O₂ plasma time. As the O₂ plasma time increased, the carbon content in the TEOS oxide film was reduced, and the film stoichiometry was improved. From the SIMS depth profile results, we found that the O/Si ratio of the TEOS oxide for the sample with the O₂ plasma time of 60 s was 2.64, and its IGZO TFT device had the best electrical characteristics. In addition, compared to the IGZO TFT device without O₂ annealing, more electrons were trapped at the interface between the TEOS oxide and the a-IGZO channel layer, as shown by a significant clockwise hysteresis in the transfer characteristics. However, the hysteresis curves of the O₂-annealed IGZO TFTs with various O₂ plasma times were greatly diminished, meaning that the electron traps were reduced by O₂ annealing.