Fabrication of indium gallium zinc oxide phototransistors via oxide-mesh insertion for visible light detection

Abstract

We developed amorphous indium gallium zinc oxide (a-IGZO) phototransistors that could sense up to the red light region of the visible light spectrum. The vital element of the utilized technology in this study is the creation of defects within the a-IGZO thin film that serves as the channel layer of a thin film transistor (TFT) through structural engineering. The defects are induced by inserting mesh-structured titanium dioxide (m-TiO₂) layers inside the channel. The oxygen vacancies thus generated create subgap states within the band gap of a-IGZO, enabling detection of the visible light with wide wavelengths. As a result, the optimized a-IGZO phototransistor with m-TiO₂ inserted inside the channel layer exhibited a photoresponsivity of 178.66 A W⁻¹, a photosensitivity of 1.04 × 10⁵, and a detectivity of 1.25 × 10¹⁰ Jones under monochromatic red light (635 nm wavelength) illumination.

---

Introduction

As technology advances, various machines facilitating daily tasks and recreational activities have exploded in number as people demand for a more comfortable life. Once people have tasted the comfort offered by these machines, their insatiable hunger for more has hastened further advances. As such, various electronics such as smart homes and autonomous vehicles are in development to offer direct interaction with various consumer electronics, home appliances, and even vehicles for more comfortable living. However, in order to realize such technology, all interlinked appliances must be able to sense the environment. For that purpose, various types of sensors are in extensive research for integration into these appliances. Among various types of sensors, photosensors are essential in this regard as they enable the detection of the environment through the change or presence of light. Therefore, photosensors are in extensive research and development for diverse applications such as vehicles, mobile devices, televisions, medical devices, and robots.^1–4

Presently, among various types of photosensors, thin-film transistor (TFT)-based photosensors, i.e., phototransistors, using amorphous oxide semiconductors (AOSs) as the channel layers are exceptionally promising as they have wavelength- and intensity-selectivity in terms of the incident light. This is because the AOSs that are used as the channel layers usually have a wide band gap of around 3.0 eV and subgap states within. When the incident light has much lower energy than the band gap energy, the light is absorbed only by electrons trapped at subgap states within the band gap. As the intensity of the light increases, more photons are emitted. Therefore, an increased number of photons are absorbed by more electrons trapped at subgap states and the drain current (I_D) increases accordingly. Similarly, as the wavelength of the incident light decreases, the light energy increases, and thus photons penetrate further into the band gap. This excites more electrons from the subgap states that lie further within the band gap, so this increases I_D accordingly.⁵

Additionally, AOS-based TFTs exhibit high uniformity over large area similar to amorphous silicon (a-Si), higher field-effect mobility (μ_FE) than a-Si-based TFTs, and lower off-current (I_off) than low-temperature poly-silicon (LTPS)-based TFTs. Therefore, many researchers have made efforts to further enhance the performance of AOS-based TFTs alone.^6–10 However, due to the band gap of around 3.0 eV, AOSs have high transparency. Thus, they cannot readily absorb visible light with wide wavelength as the energy of the visible light ranges from 1.7 to 3.1 eV (700 to 400 nm in wavelengths). As one of the widely used AOSs, amorphous indium gallium zinc oxide (a-IGZO) also has this characteristic with a band gap greater than 3.0 eV. Thus, a-IGZO-based phototransistors are usually used to detect ultraviolet (UV) or blue light with energy greater than 3.0 eV.^11–14 Therefore, there are many research studies to enable a-IGZO phototransistors to detect visible light of wavelength around 500 to 700 nm by depositing additional light-absorbing layers using various materials, or deliberately generating defects within the channel layer.^15–20 The resulting phototransistors show excellent sensitivity to visible light. However, some of them show detection only up to the green light (around 500 to 600 nm wavelength) or require materials that are either difficult to obtain or synthesize.

In this study, a-IGZO phototransistors that could sense up to the red light with a wavelength of 635 nm are introduced via the engineering of the channel layer of the phototransistors by inserting mesh-structured titanium dioxide (m-TiO₂). One of the reasons for choosing TiO₂ is because titanium (Ti) is a widely known oxygen getter element,^21–23 and one of many materials that finds various applications in photosensing-related devices.^24–29 Also, this structure was chosen to increase the number of subgap states within the a-IGZO channel layer as this is theorized and proved to enhance light absorption of AOSs.³⁰

Experimental section

Fabrication of a-IGZO phototransistors

We have developed two types of a-IGZO phototransistors that slightly differ in structures as shown in Fig. 1(a) and (b). Firstly, heavily boron-doped silicon (p⁺–Si) wafers with a layer of thermally oxidized silicon dioxide (SiO₂) about 120 nm were prepared as the substrate where p⁺–Si and SiO₂ are the gate electrode and the gate insulator, respectively. For the first structure [Fig. 1(a)], an m-TiO₂ layer was first fabricated on top of the gate insulator. Then, the first layer of the a-IGZO film of about 20 nm was deposited by radiofrequency (RF) magnetron sputtering under 5 mTorr of argon (Ar) plasma without oxygen with a sputtering power of 150 W at room temperature. The IGZO target used in this experiment consisted of indium oxide (In₂O₃), gallium oxide (Ga₂O₃), and zinc oxide (ZnO) in a 1 : 1 : 1 ratio. A second layer of m-TiO₂ was then fabricated again, and a final a-IGZO film of about 20 nm was deposited as described above. This completes the fabrication of the channel layer of one of the a-IGZO phototransistors. We term this type of phototransistor the highly defective interface (HDI) a-IGZO phototransistor because the first m-TiO₂ layer creates a high number of defects at the interface. For the second structure [Fig. 1(b)], all the layers are deposited in the order described, where all the a-IGZO films are around 13 nm: an a-IGZO, m-TiO₂, a-IGZO film, m-TiO₂, and a-IGZO film. All of the a-IGZO films are deposited as described above. We term this type of phototransistor the oxide-mesh sandwiched (OMS) a-IGZO phototransistor because an oxide-mesh (m-TiO₂) is sandwiched between a-IGZO films. After completion of the channel layers for both types, both samples were thermally annealed at 300 °C in air for 1 hour. Finally, aluminium (Al) was deposited as the source and drain electrodes for both types of phototransistors by RF magnetron sputtering. The length and width of the channel were calculated as 150 and 1000 μm, respectively.

Fig. 1 Fabrication processes of the (a) HDI and (b) OMS a-IGZO phototransistors.

Fabrication of mesh-structured TiO₂

As shown in Fig. 2(a), the samples were exposed to UV light with wavelengths 185 and 254 nm first for 30 minutes to form hydrophilic surfaces on the wafers. Next, colloidal silica was spin-coated onto the wafers, which were dried for about 5 minutes to trigger self-assembly of island formations by self-aggregation of silica nanoparticles. This formed cracked templates on the wafer, where TiO₂ of about 8 nm was then deposited by RF magnetron sputtering at 150 W under 5 mTorr of Ar plasma without oxygen. The cracked templates were then lifted-off by ultrasonication in deionized water to form m-TiO₂.³¹Fig. 2(b) shows a scanning electron microscope (SEM) image of m-TiO₂ on the wafer.

Fig. 2 (a) Fabrication process of the mesh-structured TiO₂ on a wafer. (b) SEM image of the mesh-structured TiO₂.

Measurement and analysis

The electrical properties of all phototransistors were measured using a semiconductor parameter analyser (HP 4156C; Agilent Technologies) in the dark. Also various diode lasers emitting monochromatic light of wavelengths 405, 532, and 635 nm with intensities of 1, 3, 5, and 10 mW mm⁻² were employed to measure the photosensing capacity during the electrical property measurements. The endurance of the phototransistor was also measured by applying 0.1 Hz red laser pulse (635 nm) with an intensity of 10 mW mm⁻². Also the percentages of metal–oxygen (M–O), oxygen vacancy (V_o), and hydroxide radical (–OH) bonding within the channel layers of all the phototransistors were measured using an X-ray photoelectron spectroscope (XPS; K-Alpha; Thermo Scientific Inc.) to evaluate the formation of V_o where m-TiO₂ is inserted. Also the formation of m-TiO₂ was observed using an SEM (7610F-Plus; JEOL).

Results and discussion

As mentioned in the introduction, an a-IGZO phototransistor is usually used for UV light detection as an a-IGZO film has a wide band gap of around 3.0 eV. Therefore, we have developed a method for creating subgap states within the band gap of a-IGZO by inserting m-TiO₂ inside the channel. This enabled a-IGZO phototransistors to detect visible light because the subgap states absorb energy less than the band gap energy. To obtain an optimal structure, we first fabricated eight different structures as shown in Fig. 3 by changing the locations and number of m-TiO₂ layers. Then, we compared the performances of each phototransistor in the dark, under red (635 nm), green (532 nm), and blue (405 nm) lights with an intensity of 10 mW mm⁻² with a drain voltage (V_D) of 10 V. Among these eight types, two types [Fig. 3(e) and (h)] showed exceptional detection of red light with considerable threshold voltage (V_th) shifts and large differences between I_D under light illumination (I_illuminated) and in the dark (I_dark) at a gate voltage (V_G) of −1 V. They are the HDI and OMS a-IGZO phototransistors that have been described earlier in the Experimental section and are analysed further in this study.

Fig. 3 Transfer characteristics in the dark and under red (635 nm), green (532 nm), and blue (405 nm) light (intensity 10 mW mm⁻²) for (a) a-IGZO, (b) a-IGZO/m-TiO₂, (c) double a-IGZO, (d) a-IGZO/m-TiO₂/a-IGZO, (e) a-IGZO/m-TiO₂/a-IGZO/m-TiO₂ (HDI), (f) triple a-IGZO, (g) double a-IGZO/m-TiO₂/a-IGZO, and (h) a-IGZO/m-TiO₂/a-IGZO/m-TiO₂/a-IGZO (OMS) phototransistors.

Fig. 4 shows the transfer characteristics of the a-IGZO phototransistors in the dark, under red (635 nm), green (532 nm), and blue (405 nm) light with intensities of 1, 3, 5, and 10 mW mm⁻² with a V_D of 10 V. Fig. 4 (a)–(c) show the transfer characteristics of the a-IGZO phototransistor without m-TiO₂ inserted inside the channel, which we term the pristine a-IGZO phototransistor, under red, green, and blue light, respectively. The pristine phototransistor exhibited high sensitivity to blue light that has enough energy to excite electrons trapped at subgap states that exist deep inside the band gap. However, as the wavelength of the incident light increases, the transfer characteristics show minimal changes because less number of subgap states exists within the band gap where the energy level corresponds to the energy of the incident light. In contrast, the transfer characteristics of both the HDI and OMS a-IGZO phototransistors show considerable changes in V_th and greater differences between I_illuminated and I_dark under red, green, and blue light with different intensities [Fig. 4(d)–(i)]. Hence, both types showed improved sensitivity to visible light. Moreover, both types showed improved optoelectronic characteristics compared to the pristine phototransistor, and the HDI was optimal among the three types of phototransistors.

Fig. 4 Transfer characteristics of the pristine a-IGZO phototransistor under (a) red, (b) green, and (c) blue light with different intensities. Transfer characteristics of the HDI phototransistor under (d) red, (e) green, and (f) blue light with different intensities. Transfer characteristics of the OMS phototransistor under (g) red, (h) green, and (i) blue light with different intensities.

There are three parameters that are commonly used to evaluate photosensors: photoresponsivity (PR), photosensitivity (PS), and detectivity (D*). Eqn (1)–(3) define the parameters:

PR = J_photo/P (1)

PS = I_photo/I_dark (2)

D* = PR/(2qJ_dark)^1/2 (3)

where J_photo is the current density during light illumination, P is the power density of the incident light, I_photo = I_illuminated − I_dark, q is the absolute value of the charge of an electron (1.6 × 10⁻¹⁹ C), and J_dark is the current density in the dark. The electrical properties of all the phototransistors were first measured in the dark. As shown in Table 1, we have calculated the electrical properties, such as μ_FE, V_th, and subthreshold swing (SS), and optoelectronic properties under red and green lights. For the electrical properties, the extraction methods are explained in the ESI† (see Fig. S1 for threshold voltage).^32–35

Table 1 Optoelectronic properties of a-IGZO phototransistors

Sample	In the dark				Under red light (635 nm)			Under green light (532 nm)
	μ FE (cm^2 V^−1 s^−1)	V th (V)	SS (V dec^−1)	N t (cm^−2 eV^−1)	Max. PR (A W^−1)	Max. PS	Max. D* (Jones)	Max. PR (A W^−1)	Max. PS	Max. D* (Jones)
Pristine	9.23	1.81	0.38	9.10 × 10^11	28.29	9.52 × 10	2.19 × 10^7	70.38	6.07 × 10^4	3.21 × 10^9
HDI	6.22	1.91	0.54	1.37 × 10^12	178.66	1.04 × 10^5	1.25 × 10^10	215.77	4.38 × 10^5	5.07 × 10^10
OMS	12.14	4.39	1.02	2.74 × 10^12	138.74	9.21 × 10^3	1.44 × 10^9	265.57	3.59 × 10^5	4.06 × 10^10

---

As we can observe, the pristine phototransistor showed the best performance as it lacks defects created by m-TiO₂ inside the channel. The HDI showed decreased μ_FE along with slightly increased V_th and SS. The OMS showed a slightly increased μ_FE with greatly increased V_th and SS. For the HDI, defects are created directly at the interface of the gate insulator and the channel that serves as the front channel of the TFT where electrons are accumulated and flow during the on-state (V_G > V_th). These defects create traps that impede electron flow and decrease μ_FE. For the OMS, defects are not created directly at the front channel since m-TiO₂ is not inserted there, but rather on the top layer. Because of that, m-TiO₂ attracts oxygen from the front channel, creating V_o and inducing additional carrier (electron) generation that increases μ_FE.^36,37 On the other hand, the increases in V_th and SS of both the HDI and OMS a-IGZO phototransistors indicate formation of parasitic transistors by m-TiO₂. As a well-known semiconductor with a band gap of around 3.2 eV, TiO₂ is an n-type semiconductor confirmed by previous studies.^{24,26–29,38} Since the band gap of m-TiO₂ is narrower than the band gap of a-IGZO used in this work which is shown in Fig. 5, current flows through TiO₂ before a-IGZO when enough V_G is applied to the phototransistors. This also explains the hump observed in the transfer curves as V_G required to turn on TiO₂ is less than that required to turn on a-IGZO.^39,40 Also, greater increases in V_th and SS of the OMS than those of the HDI are explained by the greater number of defects (N_t) inside the channel (2.74 × 10¹² cm⁻² eV⁻¹ for the OMS, i.e., twice the number of defects of the HDI). The calculation method of N_t is explained in the ESI.†³⁴

Fig. 5 Tauc plots of (a) pristine and (b) HDI a-IGZO films.

Although defects induced by m-TiO₂ insertion caused degradation in transistor performances, they enhanced photosensor performances with the HDI exhibiting the best sensitivity to red light. The PR improved from 28.29 to 178.66 A W⁻¹, the PS from 9.52 × 10 to 1.04 × 10⁵, and the D* from 2.19 × 10⁷ to 1.25 × 10¹⁰ Jones under red light at an intensity of 10 mW mm⁻² with a V_D of 10 V. Under green light, the PR of the OMS showed higher improvement from 70.38 to 265.57 A W⁻¹ while PR of the HDI improved to 215.77 A W⁻¹. However, the PS and D* were higher for the HDI than the OMS. The PS under green light improved from 6.07 × 10⁴ to 4.38 × 10⁵ and 3.59 × 10⁵ for the HDI and OMS, respectively. D* under green light improved from 3.21 × 10⁹ to 5.07 × 10¹⁰ and 4.06 × 10¹⁰ for the HDI and OMS, respectively. Thus, with the exception of the PR under green light, the HDI a-IGZO phototransistor exhibited the best optoelectronic properties among the three phototransistors.

We next investigated why the optoelectronic performance improved by analyzing the XPS depth profiles of the O 1s peaks for all three phototransistors. Fig. 6(a) shows that the channel layer of the pristine a-IGZO phototransistor mostly consisted of M–O bonding and as the depth changes, there are only slight changes in the percentages of M–O bonding, V_o, and –OH bonding. Because the percentage of M–O bonding is dominant among all O 1s peaks, the pristine phototransistor showed the best electrical properties in the dark. Also, the presence of V_o explains its sensitivity to the blue light even though the band gap of the pristine a-IGZO film is around 3.70 eV. Now for both the HDI and OMS a-IGZO phototransistors as shown in Fig. 6(b) and (c), the percentages of M–O bonding decreased significantly from the near-surface (2.1 nm depth for the HDI and 2.8 nm depth for the OMS) to where m-TiO₂ was inserted (19.0 and 35.9 nm depths for the HDI and 14.0 and 28.1 nm depths for the OMS). In turn, the percentages of V_o for the HDI increased from the near-surface to the depths mentioned above. For the OMS, V_o stayed almost the same for all the depths but started with higher percentage at the near-surface. For the HDI a-IGZO phototransistor, the increase in the percentage of V_o indicates the formation of subgap states within the band gap of a-IGZO that allow absorption of red light.¹⁹ For the OMS a-IGZO phototransistor, the percentage of V_o remained almost the same but started with higher percentage than that of the pristine phototransistor. This is because m-TiO₂ in the OMS is inserted nearer to the surface than in the HDI. Also similar to the HDI, this higher percentage of V_o also indicates formation of subgap states. Moreover, the percentages of –OH bonding increased as the depth became deeper. This generation of more –OH bonding is because the defects bonded with hydrogens and generated –OH radicals as both the HDI and OMS a-IGZO phototransistors showed higher percentages than the pristine phototransistor. However, as the percentage of V_o is higher than the percentage of M–O bonding, the HDI a-IGZO phototransistor showed degraded electrical performance in the dark including decreased μ_FE. As for the OMS, the percentage of M–O bonding was the highest and thus the extra carriers generated by increased amounts of V_o and –OH bonding caused an increase in μ_FE compared to those of the pristine and the HDI a-IGZO phototransistors.⁴¹

Fig. 6 De-convoluted O 1s XPS depth profiles at various depths for the channel layers of (a) the pristine, (b) the HDI, and (c) the OMS a-IGZO phototransistors.

The increased amount of V_o is caused by Ti as it strongly bonds with oxygen (O) than any other element of the a-IGZO film. As a well-known O getter, Ti has a strong Ti–O bond energy of about 7.0 eV whereas In–O, Ga–O, and Zn–O bond energies are 1.7, 2.0, and 1.5 eV, respectively.^36,37 Therefore, during thermal annealing of the phototransistors, the O atoms are attracted towards Ti atoms more so than towards In, Ga, or Zn within the a-IGZO film. Thus, more V_os are generated where m-TiO₂ is inserted as shown in Fig. 7. From here on, we do not further discuss the OMS a-IGZO phototransistor because the explanation would be redundant with the HDI a-IGZO phototransistor although the number of V_o generated is different.

Fig. 7 Comparison of schematic illustrations of the formation of oxygen vacancies at the interface of the a-IGZO and SiO₂ layers for the pristine and the HDI a-IGZO phototransistors.

As V_os are formed within the channel of the HDI a-IGZO phototransistor, more subgap states are created within the band gap of a-IGZO as shown in Fig. 8.^19,30,36,41 As visible light is illuminated on the phototransistor, the pristine phototransistor cannot absorb it given the large band gap and lack of enough number of subgap states. However, the HDI a-IGZO phototransistor can absorb the light through the abundant number of subgap states which repeatedly trap and excite electrons and contribute them to the photocurrent.

Fig. 8 Comparison of operation mechanisms for (a) the pristine and (b) the HDI a-IGZO phototransistors under visible light illumination and V_G < 0 V.

The time-dependent photoresponse characteristic of the HDI a-IGZO phototransistor was investigated by switching the red laser on and off with an intensity of 10 mW mm⁻² at a rate of 0.1 Hz. The I_D was measured at V_G = −1 V and V_D = 10 V. As red light is significantly more difficult to detect than any other visible light, measurement under red light is of particular importance compared to any other visible light. However, Fig. 9(a) shows slight increases in both I_illuminated and I_dark due to the existence of persistent photocurrent (PPC) originating from the slow recombination due to the traps that exist uniquely in oxide semiconductors.⁴² As the defects within the HDI a-IGZO phototransistor are deliberately created at the interface of the channel and the gate insulator, electrons become trapped more often, reducing both the response time and recovery time. However, this PPC issue could be resolved by applying a short but highly positive gate pulse bias (+35 V for 5 ms) immediately after the light is turned off [Fig. 9(b)]. This highly positive gate pulse bias induces recombination and detrapping of photoexcited electrons.^7,8,15,19,42 After the endurance test, the negative bias stress (NBS) test was also performed to observe the stability of the device under ambient conditions without light at a gate voltage of −1 V, a voltage also used for the endurance tests and for the detection of the light. The NBS tests were performed for the HDI phototransistors as shown in Fig. S2 (ESI†). As the graphs show, there is only a minimal change in the threshold voltage which was about −2.11 V, showing that the ambience has almost no influence under negative bias.^43–46 In order to check this, we have also deposited a passivation layer (Y₂O₃) and tested the stability under NBS. The threshold voltage shift was around −1.48 V, which has reduced but similar enough to show that the ambience has almost no effect on the device under negative bias. On the other hand, for the positive bias stress (PBS) test, the HDI phototransistor showed a significant positive threshold voltage shift of around +8.11 V without passivation. With a passivation layer, the positive threshold voltage shift has reduced to +5.64 V, which indicates that the ambience, mainly oxygen and water molecules,^46–49 has a significant effect in addition to the defects that have been made inside the channel layer. However, with a suitable passivation layer, the degradation could be reduced, and thus the device could be used without many problems.^19,48–50

Fig. 9 (a) Time-dependent photoresponse characteristic of the HDI a-IGZO phototransistor under periodic dark and red light with an intensity of 10 mW mm⁻² with a rate of 0.1 Hz for 600 seconds. (b) Comparison of one cycle of photoresponse characteristics with and without a gate pulse bias of +35 V for 5 ms.

Conclusions

In conclusion, we developed a-IGZO-based phototransistors that successfully detected wide-wavelength visible light by inserting m-TiO₂ inside the channel. We exploited the fact that Ti bonds strongly with O, creating V_o near where TiO₂ is formed for the absorption of visible light. Unlike previously developed oxide-based phototransistors, our device does not require a whole extra light absorption layer, but rather we engineered the channel directly. For this reason, the device is highly transparent, consists of simply synthesized materials, and functions relatively well as a transistor. This research has demonstrated that the insertion of an oxide-mesh at the interface of the channel and the gate insulator of an a-IGZO phototransistor increased the number of V_o and thus the number of subgap states absorbing light of lower energy than the actual band gap of the channel material of the device. In future work, we will improve both the photosensor and transistor performance by eliminating transistor degradation that has been shown in this work.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2017R1A2B3008719).

Notes and references

1. M. W. Seo, S. W. Choi, S. L. Lee, E. Y. Oh, J. S. Baek and S. J. Kang, Sensors, 2017, 17, 1112.
2. K. Kasai, Y. Haruyama, T. Yamada, M. Akiba, Y. Tomonari, T. Kaji, T. Terui, F. Peper, S. Tanaka, Y. Katagiri, H. Kikuchi, Y. Okada-Shudo and A. Otomo, Proc. SPIE, 2013, 8817, 88170N.
3. N. M. Z. Hashim, N. A. Ali, A. Salleh, A. S. Ja’afar and N. A. Z. Abidin, Int. J. Eng. Technol., 2013, 5, 3601.
4. S. Lee, S. Jeon, R. Chaji and A. Nathan, Proc. IEEE, 2015, 103, 644.
5. S. E. Ahn, I. Song, S. Jeon, Y. W. Jeon, Y. Kim, C. Kim, B. Ryu, J. H. Lee, A. Nathan, S. Lee, G. T. Kim and U. I. Chung, Adv. Mater., 2012, 24, 2631.
6. K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano and H. Hosono, Nature, 2004, 432, 488.
7. W.-G. Kim, Y. J. Tak and H. J. Kim, J. Inf. Disp., 2018, 19, 39.
8. H. Lee, K. S. Chang, Y. J. Tak, T. S. Jung, J. W. Park, W.-G. Kim, J. Chung, C. B. Jeong and H. J. Kim, J. Inf. Disp., 2017, 18, 131.
9. Y. Lim, N. Hwang, J. Lee, S. Lee and M. Yi, J. Nanosci. Nanotechnol., 2019, 19, 1470.
10. K. M. Kim, W. H. Jeong, D. L. Kim, Y. S. Rim, Y. Choi, M.-K. Ryu, K.-B. Park and H. J. Kim, IEEE Electron Device Lett., 2011, 32, 1242.
11. S. Knobelspies, A. Daus, G. Cantarella, L. Petti, N. Münzenrieder, G. Tröster and G. A. Salvatore, Adv. Electron. Mater., 2016, 2, 1600273.
12. Z. Tao, X. Liu, W. Lei and J. Chen, Mater. Lett., 2018, 228, 451.
13. H. Zhu, A. Liu, F. Shan, W. Yang, W. Zhang, D. Li and J. Liu, Carbon, 2016, 100, 201.
14. H. Lu, X. Zhou, T. Liang, L. Zhang and S. Zhang, IEEE J. Electron Devices Soc., 2017, 5, 504.
15. M. G. Yun, Y. K. Kim, C. H. Ahn, S. W. Cho, W. J. Kang, H. K. Cho and Y. H. Kim, Sci. Rep., 2016, 6, 3199.
16. J. Yang, H. Kwak, Y. Lee, Y. S. Kang, M. H. Cho, J. H. Cho, Y. H. Kim, S. J. Jeong, S. Park, H. J. Lee and H. Kim, ACS Appl. Mater. Interfaces, 2016, 8, 8576.
17. Y. J. Tak, D. J. Kim, W.-G. Kim, J. H. Lee, S. J. Kim, J. H. Kim and H. J. Kim, ACS Appl. Mater. Interfaces, 2018, 10, 12854.
18. A. D. Mottram, Y. H. Lin, P. Pattanasattayavong, K. Zhao, A. Amassian and T. D. Anthopoulos, ACS Appl. Mater. Interfaces, 2016, 8, 4894.
19. B. H. Kang, W.-G. Kim, J. Chung, J. H. Lee and H. J. Kim, ACS Appl. Mater. Interfaces, 2018, 10, 7223.
20. C. H. Ahn., W. J. Kang, Y. K. Kim, M. G. Yun and H. K. Cho, ACS Appl. Mater. Interfaces, 2016, 8, 15518.
21. W. D. Sylwestrowicz, J. Electrochem. Soc., 1975, 122, 1504.
22. P. A. Crozier, M. R. McCartney and D. J. Smith, Surf. Sci., 1990, 237, 232.
23. W. Zhang, J. Mao, W. K. Pang, X. Wang and Z. Guo, Nano Energy, 2018, 49, 549.
24. A. M. Selman, Z. Hassan, M. Husham and N. M. Ahmed, Appl. Surf. Sci., 2014, 305, 445.
25. Y. Mansourpanah, S. S. Madaeni, A. Rahimpour, A. Farhadian and A. H. Taheri, J. Membr. Sci., 2009, 330, 297.
26. A. A. Hendi and F. Yakuphanoglu, J. Alloys Compd., 2016, 665, 418.
27. A. M. Selman and Z. Hassan, Sens. Actuators, A, 2015, 221, 15.
28. M. Okuya, K. Shiozaki, N. Horikawa, T. Kosugi, G. R. A. Kumora, J. Madarasz, S. Kaneko and G. Pokol, Solid State Ionics, 2004, 172, 527.
29. H.-Y. Liu, R.-C. Huang, Y.-Y. Li, C.-S. Lee and W.-C. Hsu, IEEE Electron Device Lett., 2017, 38, 756.
30. J. Yao, N. Xu, S. Deng, J. Chen, J. She, H.-P. D. Shieh, P.-T. Liu and Y.-P. Huang, IEEE Trans. Electron Devices, 2011, 58, 1121.
31. Y.-g. Kim, Y. J. Tak, S. P. Park, H. J. Kim and H. J. Kim, Nanomaterials, 2017, 7, 214.
32. A. Ortiz-Conde, F. J. Garcia Sanchez, J. J. Liou, A. Cerdeira, M. Estrada and Y. Yue, Microelectron. Reliab., 2002, 42, 583.
33. R. Martins, P. Barquinha, I. Ferreira, L. Pereira, G. Concalves and E. Fortunato, J. Appl. Phys., 2007, 101, 044505.
34. G. H. Kim, B. D. Ahn, H. S. Shin, W. H. Jeong, H. J. Kim and H. J. Kim, Appl. Phys. Lett., 2009, 94, 233501.
35. J. H. Park, Y.-g. Kim, S. Yoon, S. Hong and H. J. Kim, ACS Appl. Mater. Interfaces, 2014, 6, 21363.
36. T. Kamiya, K. Nomura and H. Hosono, Phys. Status Solidi A, 2010, 207, 1698.
37. H.-H. Hsu, C.-Y. Chang, C.-H. Cheng, S.-H. Chiou and C.-H. Huang, IEEE Electron Device Lett., 2014, 35, 87.
38. W. S. Shih, S. J. Young, L. W. Ji, W. Water and H. W. Shiu, J. Electrochem. Soc., 2011, 158, 609.
39. A. Valletta, P. Gaucci, L. Mariucci, G. Fortunato and F. Templier, J. Appl. Phys., 2008, 104, 124511.
40. M. J. Seok, M. Mativenga, D. Geng and J. Jang, IEEE Trans. Electron Devices, 2013, 60, 3787.
41. J. Bang, S. Matsuishi and H. Hosono, Appl. Phys. Lett., 2017, 110, 232105.
42. S. Jeon, S. E. Ahn, I. Song, C. J. Kim, U. I. Chung, E. Lee, I. Yoo, A. Nathan, S. Lee, J. Robertson and K. Kim, Nat. Mater., 2012, 11, 301.
43. S. Kim, S. Kim, C. Kim, J. Park, I. Song, S. Jeon, S.-E. Ahn, J.-S. Park and J. K. Jeong, Solid-State Electron., 2011, 62, 77.
44. J. H. Shin, J.-S. Lee, C.-S. Hwang, S.-H. Ko Park, W.-S. Cheong, M. Ryu, C.-W. Byun, J.-I. Lee and H. Y. Chu, ETRI J., 2009, 31, 62.
45. J. Kim, J. Bang, N. Nakamura and H. Hosono, APL Mater., 2019, 7, 022501.
46. C.-L. Fan, F.-P. Tseng, B.-J. Li, Y.-Z. Lin, S.-J. Wang, W.-D. Lee and B.-R. Huang, Jpn. J. Appl. Phys., 2016, 55, 02BC17.
47. J. K. Jeong, H. W. Yang, J. H. Jeong, Y.-G. Mo and H. D. Kim, Appl. Phys. Lett., 2008, 93, 123508.
48. K. Y. Shin, Y. J. Tak, W.-G. Kim, S. Hong and H. J. Kim, ACS Appl. Mater. Interfaces, 2017, 9, 13278.
49. S. Hong, S. P. Park, Y.-g. Kim, B. H. Kang, J. W. Na and H. J. Kim, Sci. Rep., 2017, 7, 16265.
50. H. Yoo, Y. J. Tak, W.-G. Kim, Y.-g. Kim and H. J. Kim, J. Mater. Chem. C, 2018, 6, 6187.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9tc04982g

---