Enhancement in Performance and Reliability of Fully Trans-Parent A-IGZO Top Gate Thin-Film Transistors by a Two-Step Annealing TreatmentEnhancement in Performance and Reliability of Fully Trans-Parent A-IGZO Top Gate Thin-Film Transistors by a Two-Step Annealing Treatment

1. Introduction

The top gate (TG) structure amorphous InGaZnO₄ (a-IGZO) thin-film transistor (TFT) was widely used in fields of fingerprint sensors, transparent displays, and flexible electronics [1-6] due to its high gate control capability and excellent stability. The high stability of a-IGZO TG-TFTs requires a certain special treatment on the a-IGZO active layer, such as annealing in different gas atmospheres, plasma gas treatment, or covering a passivation layer [5, 7-12]. The before mentioned various treatments could improve the bonding state of metal ions and oxygen atoms both on the surface, interface, and inside the a-IGZO active layer [9, 13]..

For TG-TFTs, the gate dielectric layer was deposited after the semiconductor active layer. The bonding state in the a-IGZO active layer would degrade during the deposition of Al₂O₃ gate dielectric material in the fabrication of TG-TFTs [5, 14-16]. Especially, the quantity of oxygen vacancy (V_O) and the distribution of O defects in a-IGZO were significantly altered when the Al₂O₃ top gate dielectric was grown through atomic layer deposition (ALD) process, leading to the degeneration of the electrical performance of a-IGZO TG-TFTs [8-10, 14, 17]. The variation of oxygen-related states mainly originated from oxygen outgassing from a-IGZO during the deposition of Al₂O₃ in the ALD high-temperature vacuum environment [18]. Thermal annealing treatments in air or O₂ after gate dielectric deposition have been considered to be an effective method to improve oxygen-related states between the a-IGZO-Al₂O₃ interface and the internal a-IGZO and Al₂O₃, and increase the film density and uniformity [5, 14-16], which mainly relies on oxygen diffusion mechanisms [5, 14, 15, 19]. However, the diffusion of external oxygen into a-IGZO active layer are severely hindered due to the passivation effect of Al₂O₃ top gate dielectric, resulting in only partly eliminating the degradation of a-IGZO. In addition, oxygen vacancies passivated by the hydrogen diffusing from the precursor residue of trimethylaluminum (TMA) in Al₂O₃ and the deficiency of oxygen in a-IGZO lead to the negative degradation of threshold voltage (V_th) of a-IGZO TG-TFTs [16, 19]. Therefore, another annealing treatment of a-IGZO in air or O₂ before the gate dielectric deposition was proposed in some reports to resist the degradation of a-IGZO due to the formation of metal-oxygen (M-O) bonds, reducing the V_O and O defects and densifying a-IGZO films [20, 21]. In previous reports, the effect on the performance of devices and the mechanism of the annealing treatment after gate dielectric deposition have been thoroughly investigated [5, 11, 16, 22]. While the effect and mechanism of the annealing treatment of a-IGZO before the gate dielectric deposition are seldom reported.

In this work, a two-step annealing treatment was applied during the fabrication of the TG-TFT. The annealing treatment on a-IGZO active layer before ALD-Al₂O₃ gate dielectric deposition is defined as the first annealing (1^st-annealing), while the annealing of a-IGZO TG-TFTs after Al₂O₃ coverage is defined as the second annealing (2^nd-annealing). The mechanism on a-IGZO and a-IGZO TFTs of the 2^nd-annealing treatment had been reported in our previous study [16]. Herein, the effect and mechanism of the 1^st-annealing treatment on a-IGZO active layer and the stability of a-IGZO TG-TFTs were systematically investigated based on a fully transparent a-IGZO TG-TFT fabricated under different 1^st-annealing conditions. The UV–Vis optical transmission spectra was used to obtain the transmittance of the a-IGZO TG-TFT. The photoluminescence (PL) spectroscopy and X-ray photoelectron spectroscopy (XPS) were employed to analysis the 1^st-annealing mechanism through investigating the distribution and the content of defects in a-IGZO.

2. Materials and Methods

Fig. 1(a) shows a schematic of the fabricating process sequence and major processing steps, respectively, of the top-gate bottom-contact a-IGZO TFT in this work. The fully transparent staggered TG-TFTs were fabricated on a cleaned SiO₂ (100 nm)/Si substrate. Firstly, ITO source and drain (S/D) electrodes were deposited by radio-frequency (RF) magnetron sputtering at 60 W in Ar atmosphere under the pressure of 0.43 mtorr for 840 s. The S/D regions were defined by photolithography, and outside regions were etched by CH₄ and H₂ at 60 °C using inductively coupled plasma reactive-ion-etching (ICP-RIE). Secondly, the a-IGZO channel was deposited through RF magnetron sputtering at 90 W in pure Ar atmosphere from a ceramic target with a molar ratio of In₂O₃ : Ga₂O₃ : ZnO = 1 : 1 : 1 at room temperature (RT). Right after the deposition of a-IGZO layer, the 1^st-annealing treatment was carried out in air (the air humidity is 40%) for 1 hour, which the 1^st-annealing treatment temperatures were controlled between 150°C and 350°C with a step of 50°C. The a-IGZO channel regions were defined by photolithography, and were etched by CH₄ and H₂ at 20°C using ICP-RIE technology. Subsequently, the Al₂O₃ gate insulator (GI) was deposited through thermal-ALD at 150°C with trimethylaluminum (TMA) and H₂O as aluminum and oxygen precursor, respectively, and N₂ as purge gas. The pulse time of TMA, H₂O, and N₂ were 0.03 s, 0.02 s, and 30 s, respectively. Finally, ITO gate electrodes were deposited by RF magnetron sputtering in Ar atmosphere and defined by lift-off processes. The 2^nd- annealing treatment was carried out at 300°C in air for 1 hour after the fabrication of a-IGZO TFTs finished. Additionally, the layer thicknesses of ITO S/D electrodes, a-IGZO channel, Al₂O₃ GI, and ITO gate electrode were measured using atomic force microscope (AFM) and found to be 50 nm, 24 nm, 10 nm, and 50 nm, respectively.

All electrical measurements of the a-IGZO TG-TFTs were performed using an Agilent B1500A parameter analyzer connecting Cascade EPS150TRIAX probe station at RT in a dark ambient atmosphere. The PL spectra was obtained using a Renishaw inVia Raman spectrometer equipped with a 325 nm wavelength laser. XPS spectra were obtained on a Thermo Scientific K-Alpha XPS system with monochromatic Al Kα X-ray (1486.7 eV) radiation and were calibrated by the C 1s peak (∼284.8 eV).

3. Results

3.1. Optical and Electric characteristics of TG-TFTs

The transmittance of the fully transparent staggered a-IGZO TG-TFT was showed in Figure 1(b). The transmittance of the a-IGZO TG-TFT obtained by the UV–Vis optical transmission spectra had achieved over 90% (quartz as substrate) in visible light range. An optical band gap of 3.84 eV was calculated based on the UV-Vis optical transmission spectrum of a 130 nm thick a-IGZO film on quartz glass, which was shown in the insert of Fig. 1(b). The pleasant transmittance of the TG-TFT was mainly attributed to the wide bandgap of the a-IGZO material, which was benefit for the application in transparent flexible electronic smart wearable devices.

Table 1. Electrical characteristic parameters of TG-TFTs with different 1^st-annealing temperature.

Parameters T (°C)	Vth-lin（V）	Vth-sat（V）	μsat （cm2V-1s-1）	Ion/Ioff (VD = 3 V)	SS （mV/dec）	Nt (eV-1cm-2)
150	-1.13	-1.23	9.13	4.03×103	--	--
200	-0.53	-0.83	30.52	4.46×108	203	6.10×1012
250	1.12	0.78	6.87	6.98×107	137	3.28×1012
300	0.52	0.33	13.05	1.73×108	130	2.98×1012
350	-2.05	-3.05	--	7.2	--	--

Table 1. Electrical characteristic parameters of TG-TFTs with different 1^st-annealing temperature.

Parameters T (°C)	Vth-lin（V）	Vth-sat（V）	μsat （cm2V-1s-1）	Ion/Ioff (VD = 3 V)	SS （mV/dec）	Nt (eV-1cm-2)
150	-1.13	-1.23	9.13	4.03×103	--	--
200	-0.53	-0.83	30.52	4.46×108	203	6.10×1012
250	1.12	0.78	6.87	6.98×107	137	3.28×1012
300	0.52	0.33	13.05	1.73×108	130	2.98×1012
350	-2.05	-3.05	--	7.2	--	--

Figure 2(a) and (b) displayed transfer characteristics of TG-TFTs with different 1^st-annealing treatments. As shown in Fig. 2(a) and (b), the TG-TFT without the 1^st-annealing treatment did not show the switching characteristic, which indicated that the degradation of a-IGZO without 1^st-annealing was most severe after the deposition of Al₂O₃. There was a significant change in transfer characteristics of the transistor after the introduction of the 1^st-annealling treatment. The TFTs with 1^st-annealling at 150°C and 350°C did not exhibit obvious switching characteristics, which were attributed to the extensive shallow donor states and the creation of new trap states in a-IGZO [9], respectively. The TFTs with 1^st-annealling treatment at 200°C-300°C exhibited good switching characteristics, which also showed a variation with 1^st-annealling temperature. The electrical parameters of TG-TFTs, such as threshold voltage (V_th), subthreshold swing (SS), field-effect mobility (μ_sat) and on/off current ratio (I_on/I_off) were all extracted from the saturated transfer curve, as shown in Table I. The V_th-sat was shifting positively with increasing 1^st-annealing temperature up to 250°C and then switching to a negative shift with further increasing 1^st-annealing temperature. The turn-around V_th shift was associated with the changes in the quantity and distribution of V_O and O defects, and the bonding energy between metal ions and oxygen atom within the internal composition of a-IGZO films [9, 17, 23]. From Fig. 2(b) and Table I, the values of SS were 203 mV (200°C), 137 mV (250°C), and 130 mV (300°C), and showed a declining trend under the increasing 1^st-annealing temperatures. It was reported that the SS was mainly influenced by the interfacial quality and defect states between Al₂O₃ and the a-IGZO channel [24]. Excellent interfaces between a-IGZO and Al₂O₃ had been proved by the negligible hysteresis of transfer curves as shown in Fig. 2(a) and (b). And the density of trapping states (N_t) at the interface was extracted from SS by the following formula：

$$SS={\displaystyle \frac{{\kappa }_{B}T\ln 10}{q}}\left[1+{\displaystyle \frac{q^2}{C_{ax}}}{N}_t\right]$$

(1)

where κ_B is Boltzmann’s constant, C_ox is unit area capacitance of 10 nm Al₂O₃, T was the temperature in Kelvin, and q is the electron charge. In this work, the C_ox was 408 nF/cm² obtained from the capacitance-voltage measurement. The calculated N_t value was within the range of 2.98×10¹² ̶ 6.10×10¹² eV^-1cm^-2 as shown in Table I, and this result was also agreement with the reported value 3.0×10¹² eV^-1cm^-2 [24]. The improvement of SS performance with the 1^st-annealling temperature might be mainly attributed to the reduction of O defects and V_O in a-IGZO film [25]. Therefore, we further analyzed the a-IGZO films with 1^st- annealing treatments at 200°C, 250°C, and 300°C by PL spectra and XPS technology, based on the good switching transfer curves, as shown in Fig. 5 and Fig. 6.

3.2. Electrical stability test on a-IGZO TG-TFTs

The electrical stability of a-IGZO TG-TFTs was further investigated. As depicted in Figure 3 and Figure 4, positive bias stress (PBS) and negative bias stress (NBS) measurements were performed on TG-TFTs under the test conditions of V_stress (gate bias voltage) = ± 3 V at V_D = 0.1 V. As displayed in Figure 3 (a), (b) and (c), transfer curves of all TG-TFTs shifted very slightly in negative direction after a positive stress duration of 7200 s, demonstrating an excellent resistance to PBS. And this extraordinary stability was mainly due to the good interface between a-IGZO channel and Al₂O₃ insulator [26-28]. Moreover, the negative shift in transfer curves was likely caused by the H-doping effect, which was resulting from the ALD-Al₂O₃ insulator due to the 2^nd-annealing treatment[16]. The deviations of V_th (ΔV_th) between the initial and the 7200 s positive biased transfer curves at V_D = 0.1 V were -0.11 V (200°C), -0.035 V (250°C), and -0.032 V (300°C), respectively. This result showed PBS stabilities of TG-TFTs had an obvious improvement with increasing 1^st-annealing temperatures, and the TG-TFT with 1^st-annealing treatment at 300°C had the best PBS stability.

The NBS results of a-IGZO TG-TFTs were presented in Fig. 4. The transfer characteristics of all TG-TFTs exhibited negative shifts after a 7200 s stress duration under the V_stress = - 3 V condition, as shown in Fig. 4(a), (b), and (c). The negative shifts in V_th attribute to that holes were difficult to be generated in n-type oxide semiconductors[16]. The ΔV_th values between the initial and the 7200 s negative biased transfer curves at V_D = 0.1 V were -0.573 V (200°C), -0.335 V (250°C), and -0.044 V (300°C). It was evident that the NBS stabilities of the TG-TFTs were improved with increasing 1^st-annealing temperature, and TG-TFT with 300°C 1^st-annealing treatment had the best NBS stability. The excellent NBS stability would make a-IGZO TG-TFTs very advantageous for applications in the field of transparent displays.

The best PBS stability and the best NBS stability were both obtained from TG-TFTs with 1^st-annealing treatment at 300°C. In order to further understand the effect of the 1^st-annealing treatment on the bias stress stability improvement of TG-TFTs, PL and XPS spectra were conducted on a-IGZO films with 1^st-annealing treatments at 200°C, 250°C, and 300°C, as shown in Figure 5 and Figure 6.

3.3. 1^st-annealing effect on a-IGZO films

Fig. 5 and the insert figure showed the PL spectra and separated PL spectra of a-IGZO films with 1^st-annealing at fresh (no 1^st-annealing), 200°C, 250°C, and 300°C. Four broad emission peaks were found in the visible light range of 1.6 ~ 3.2 eV, which is consistent with previous reports [29-32]. From the insert of Fig. 5, every peak was observed with a flat top, of which the maximum energy for all peaks was at 2.16 eV. The result meant that the deepest defect in all a-IGZO films was located at 2.16 eV below the conduction band bottom (VBM) )[33]. In addition, the energy bands occupied by the flat top of the measured peaks were 0.159 eV (fresh), 0.130 eV (200°C), 0.107 eV (250°C), and 0.136 eV (300°C), respectively. The peak intensity and width of the energy band of the flat top in the PL spectra may illustrate the density and the distribution range of defects in the a-IGZO film [30, 31, 34]. The PL curve of the a-IGZO film with 1^st-annealing at 250°C exhibited the highest peak intensity and the narrowest energy band (0.107 eV) of the flat top, while the PL spectra of the a-IGZO film without 1^st-annealing showed the lowest peak intensity and the widest energy band (0.159 eV). These results demonstrated that the O defects in a-IGZO film with 1^st-annealing at 250°C were at the deepest energy level, which might result in the largest positive V_th for the TG-TFT, as shown in Fig. 2(a) and (b). The PL spectra of the a-IGZO film with 300°C 1^st-annealing treatment showed a lower peak intensity but wider energy band (0.136 eV) of the flat top than that of the 250°C 1^st-annealed a-IGZO film. This result might suggest that some shallow defect states were generated in the 300°C 1^st-annealed a-IGZO film as the overall number of defects decreased [9]. And this might be the reason for the higher mobility of the TG-TFT with 300°C 1^st-annealing treatment than that of the TG-TFT with 250°C 1^st-annealed a-IGZO, as shown in Table I. The generation of ionized V_O might lead to the generation of O defects, explaining the wider energy band of the flat top for the a-IGZO film with 1^st-annealing temperature at 300°C than that of the 200°C 1^st-annealed a-IGZO film in the PL spectra, as shown in Fig. 5. Consequently, ionized V_O and electrons would gather on the upper surface of the a-IGZO channel, and this made the TG-TFT with 1^st-annealing temperature at 300°C achieved a higher mobility of 13.05 cm²V^-1s^-1.

The peak intensity decreased with the increase of 1^st-annealing temperature as shown in Fig.5, indicating the defect density decreased with the increase of 1^st-annealing temperature. This result would explain the PBS and NBS stability improvement with the increase of 1^st-annealing temperature.

It is reported that bonding states of gallium atoms in a-IGZO had a great impact on V_th, SS, and μ for a-IGZO TFTs [35]. It is known that Ga-O bond exhibits the shortest bond length and the strongest binding energy compared to In-O bond and Zn-O bond [36]. XPS spectra of oxygen and gallium elements in a-IGZO films with 1^st-annealing treatments at 200°C ̶ 300° were employed to investigate the relationship between transfer characteristics of TG-TFTs and temperatures of 1^st-annealing treatments. The 1^st-annealing temperature of the controlled samples were set at fresh (no 1^st-annealing treatment), 200°C, 250°C, and 300°C, respectively. Fig. 6 (a) and (b) showed the O 1s and Ga 3d XPS spectra of a-IGZO films with 1^st-annealing treatments at fresh, 200°C, 250°C, and 300°C, respectively. O 1s spectra could be deconvoluted into two sub-peaks based on Gaussian fitting, i.e., bond between O atom and metal atom (O-M) (centered around 530.15 eV), V_O (centered around 531.47 eV), and Ga 3d spectra could be deconvoluted into Ga-Ga (centered around 17.8 eV) and Ga-O (centered around 19.60 eV) [36-39]. O-M and V_O core peaks of a-IGZO films with 1^st-annealing at fresh, 200°C, 250°C, 300°C were at 530.02 eV, 530.12 eV, 530.20 eV, and 530.15 eV with error of ± 0.02 eV, and V_O core peaks were at 530.85 eV, 531.45 eV, 531.55 eV, and 531.50 eV with error of ± 0.02 eV, respectively, as shown in Fig.6(a). The core peak energy values of O-M for the a-IGZO films were almost consistent. The V_O core peak energy for the a-IGZO film with 1^st-annealing at 250°C was found to be the biggest (531.55 eV), which indicated that the defect states in the a-IGZO film were located at the deepest energy level compared to that in other a-IGZO films, consistent with the PL spectra results from Fig. 5. The contents of V_O and O-M bonds were calculated from the O 1s spectra, as shown in Table II. The V_O contents in a-IGZO films were 26.60% (fresh), 11.17% (200°C), 12.84% (250°C), and 11.01% (300°C), respectively. The contents of O-M bonds in a-IGZO films were 17.27% (fresh), 32.57% (200°C), 30.45% (250°C), and 32.44% (300°C), respectively. The 1^st-annealing treatment led to a 13.76 ̶ 15.59% decrease in V_O content and 13.18 ̶ 15.3% increase in O-M bonds content for a-IGZO films, as shown in Table II. The change in contents of V_O and O-M bonds demonstrated that the O-related defects in a-IGZO films decreased with the 1^st-annealing treatment temperature. The V_O content in the 250°C 1^st-annealed a-IGZO film was slightly higher than that of a-IGZO films with 1^st-annealing at 200°C and 300°C, indicating that the combination state between O and metal atoms under this condition was different compared to other a-IGZO films.

In this report, the Ga-Ga peaks in all a-IGZO films were centered at 17.80 eV with calculation error of ± 0.02 eV. The Ga-O peak in the 250°C 1^st-annealed a-IGZO film was centered at 19.75 eV, whereas the Ga-O peaks for other a-IGZO films were all centered at 19.60 eV with calculation error of ± 0.02 eV, as shown in Fig. 6(b). The 0.15 eV increase in Ga-O peak indicated that the coordination between Ga and O in the 250°C 1^st-annealed a-IGZO film was better than that in other a-IGZO films. The Ga-O contents in a-IGZO films were 12.82% (fresh), 12.92% (200°C), 11.18% (250°C), and 13.34% (300°C), respectively. The a-IGZO film with 1^st-annealing temperature at 300°C had the highest Ga-O content. This might be the reason why TG TFTs with 1^st-annealing at 300°C exhibited the best PBS and NBS stability as shown in Fig.3(c) and Fig.4(c), based on the fact that Ga-O could suppress the ionization of V_O at gate voltage bias[36].

Usually, an increase of V_O in a-IGZO would lead to V_th shifting to negative direction, while the enhancement of Ga-O bond lead to V_th shifting to positive direction. In comparison to the a-IGZO film with 1^st-annealing temperature at 200°C, the 300°C 1^st-annealed a-IGZO film showed 0.16% decrease of V_O content and a 0.42% increase of Ga-O content. As a result, the V_th of the TG-TFT with 300°C 1^st-annealed a-IGZO exhibited a positive shift. The V_O content in the 250°C 1^st-annealed a-IGZO film was higher than that of a-IGZO films with 1^st-annealing temperature at 200°C and 300°C, and while the Ga-O content in the 250°C 1^st-annealed a-IGZO film was lower than that of a-IGZO films with 1^st-annealing temperature at 200°C and 300°C. Thus, V_th of the TG-TFT with the 250°C 1^st-annealed a-IGZO should be more negative than that of the TG-TFT with other a-IGZO films. However, the largest positive V_th was observed at the TG-TFT with 250°C 1^st-annealed a-IGZO. This strange phenomenon might suggest that the Ga-O content had a greater impact on V_th than V_O content. And the better coordination between Ga and O promoted that the V_th of the TG-TFT with the 250°C 1^st-annealed a-IGZO film was more positive than that of the TG-TFT with 300°C 1^st-annealed a-IGZO.

4. Conclusions

In this work, the electronic properties and stabilities of a-IGZO TG-TFTs were investigated by controlling the 1^st-annealing temperature, and the mechanism of 1^st-annealing treatment on a-IGZO films was explored through PL spectra and XPS analysis. The TG-TFT achieved good switching function when 1^st-annealing temperature at 200°C – 300°C. The TG-TFT with 1^st-annealing treatment at 300°C resulted in the best electrical properties, PBS stability, and NBS stability.

The PL spectra and XPS spectra revealed the change of O defect states and Ga-O bond state in the 1^st-annealed a-IGZO films. O defects were reduced with 1^st-annealing treatments, but some shallow states were produced when the 1^st-annealing temperature increased to 300°C. The best coordination situation between Ga and O atoms occurred for a-IGZO film with 1^st-annealing treatment at 250°C, leading to the biggest positive V_th of TG-TFTs. Furthermore, the highest content of Ga-O bond and minimal defect intensity were both obtained in a-IGZO film with 1^st-annealing at 300°C, which contributed to the excellent PBS and NBS stabilities of TG-TFTs.