Realization of Visible‐Light Detection in InGaZnO Thin‐Film Transistor via Oxygen Vacancy Modulation through N2 Treatments

InGaZnO (IGZO) thin‐film transistors (TFTs) have gained widespread use in active matrix (AM) displays due to their decent field‐effect mobility and low off current. However, the wide bandgap of IGZO limits their application in visible light detection. This study presents an industry process‐compatible method to achieve visible light detection in IGZO TFTs through N2 treatment during the sputtering and annealing process for the first time. A comparison with control IGZO TFTs shows that the N2‐treated IGZO TFTs exhibit a high responsivity of 0.66 A W−1 and detectivity of 5.40 × 1014 Jones for visible light detection. Based on X‐ray photoelectron spectroscopy analysis and technology computer‐aided design simulations, a model, focusing on oxygen vacancy modulation, is proposed to explain the visible‐light sensitivity in IGZO TFTs with N2 treatment. This work opens up new possibilities for the integration of IGZO photodetector into AM display panels to realize in‐cell environmental detection and biometric recognition.


Introduction
Low-cost and low-power photodetectors are essential for environmental detection [1] and biometric recognition. [2]ue to its transparency and low power, metal oxide (MO) thin-film transistor (TFT) has become a particularly appealing option for environmental monitoring [3] and biometric sensing. [4]mong MO TFTs, [5] InGaZnO (IGZO) TFT is the first marketable technology for active matrix (AM) displays because of its inexpensive process cost, decent field-effect mobility (μ FE ), and extremely low leakage current. [6]9][10] However, the applicability of IGZO TFT in visual detection is constrained by its wide bandgap (≈3.5 eV). [11][14][15][16][17] This light absorption layer is usually organic materials [14,15] or inorganicorganic hybrids. [16,17][20][21] QDs are a type of nanocrystal made from special semiconductor materials, with a size ranging from just a single molecule up to a large crystalline solid (≈1-100 Å). [18,19] The unique size and structure of QDs give rise to their exceptional optical properties, which are significantly different from those of bulk materials or individual molecules. [20,21][24][25] V O is one kind of defects inside the IGZO film, which can trap carriers and affect the photoconductivity of a device.It is worth noting that either the light absorption layer or the QDs are incompatible with the current AM display industry processes.Developing a display industry process-compatible method to extend the optical detection range for IGZO TFTs poses a significant challenge.
In this work, a visible-light photodetector based on IGZO TFTs is achieved by modulating V O via a series of N 2 treatments.By controlling the N 2 during the sputtering process and annealing process, the content of V O in the IGZO thin film can be tuned to introduce subgap states in the energy band of the IGZO, making IGZO TFTs sensitive to visible light.Compared with normal IGZO TFTs, IGZO TFTs with N 2 treatments exhibit higher sensitivity to visible light (450-700 nm).The best photoresponsibility (R) of 0.66 A W −1 and normalized detectivity (D*) of 5.40 × 10 14 Jones can be realized in IGZO TFTs at 450 nm visible light wavelength.By incorporating X-ray photoelectron spectroscopy (XPS) analysis and technology computer-aided design (TCAD) simulations, a model is tentatively proposed to understand the visible-light sensitivity in IGZO TFTs with N 2 treatments, which considers the increased V O as the subgap states in the IGZO energy band.This work provides an effective and AM display industry process-compatible approach to fabricating IGZO TFTs with high visible light sensitivity.Shown in Figure 1b are transfer curves of different kinds of IGZO TFTs, measured under the dark condition.The TFT-A shows a standard transfer characteristic with a typical μ FE of 16 cm 2 V −1 s −1 and a threshold voltage (V th ) of 1.0 V.For IGZO TFTs treated with N 2 , the transfer curve of all TFT-B, TFT-C, and TFT-D is negatively shifted.A shorter postannealing time brings a larger negative shift of the transfer curve.The V th of TFT-B, TFT-C, and TFT-D are, respectively, 0.38, −1.64, and −6.9 V. Compared with the TFT-A, the negative shift of transfer curves in IGZO TFT after N 2 treatment may result from the change of carrier concentration in the IGZO channel. [26]As the carrier concentration increases, the devices will turn on more easily, causing the negative shift of transfer curves.It is worth pointing out that if only introducing N 2 during the sputtering process or annealing process, the device will become conductive (Figure S1 in the Supporting Information).

IGZO
Shown in Figure 1c-f are, respectively, transfer curves of TFT-A, TFT-B, TFT-C TFT-D under illuminations with different visible light wavelengths ( s ).For the control TFT-A, all the transfer curves overlap each other under visible light illuminations as shown in Figure 1c, indicating the TFT-A is insensitive to visible light.This is a normal phenomenon, which is also presented by other research work. [11,12,27]The control IGZO TFT (TFT-A) cannot absorb visible light photons due to its larger bandgap than visible light photon energy. [28]Therefore, TFT-A is incapable of generating photoinduced carriers when exposed to visible light and cannot induce a negative shift in transfer curves.After a series of N 2 treatments, the subthreshold current of TFT-B, TFT-C, and TFT-D becomes sensitive to visible light while the on-current is still almost insensitive to visible light since the photogenerated carriers only have a significant effect on subthreshold current. [11]The transfer curves exhibit a more negative shift as the incident light wavelength decreases from 700 to 450 nm.Shorter  has higher energy and can excite the more photogenerated carriers in the IGZO channel, [29][30][31] resulting in a greater response in transfer curves.For the different N 2 treatments, shorter postannealing time (TFT-D) brings a larger visible light response.An assumption can be made that N 2 treatment may introduce new states in the energy band of IGZO that aid in absorbing visible light photons.Shorter postannealing times may increase the number of such states, resulting in a larger visible light response.It is also observed that there is a significant decrease in current at ≈−10 V when the TFT-B is illuminated at  = 450 nm.This unusual I off decrease may be due to the discharge of the gate-to-drain capacitor, influenced by contact resistance and sweeping rates. [32]Besides, the transfer curves of TFT-C under visible light illuminations with different light intensities were also measured (Figure S2 in the Supporting Information).Not surprisingly, the response to visible light becomes stronger as the incident light intensity increases.
The R of a phototransistor is a crucial indicator of its sensitivity to light. [33]Figure 2a,b shows the R of different kinds of IGZO TFTs for various light  s , with the highest R values shown in Figure 2a and the R values at V on (R-V on ) shown in Figure 2b.For the highest R extraction, the V ds is fixed at 0.1 V.The V gs is defined as the gate voltage at which R takes its maximum value, which is different for all TFTs for different incident light  conditions.The R-V on of TFT-A is, respectively, 8.3 × 10 −6 , 2.3 × 10 −6 , 1.88 × 10 −5 , and 2.2 × 10 −4 A W −1 for  = 700, 600, 500, and 450 nm, indicating negligible response to visible light.After a series of N 2 treatments, the R gradually increases with a shorter postannealing time.For example, the maximum of R of TFT-B to TFT-D gradually increases from 0.0456 to 0.521 A W −1 at  = 500 nm.Furthermore, the R values of all devices exhibit an upward trend as the  decreases, consistent with the negative shift of the transfer curve.TFT-B, TFT-C, and TFT-D reach their highest R values at 450 nm, which are 0.092, 0.059, and 0.66 A W −1 , respectively.
To account for differences in device area and response speed among various photodetectors, the D* is commonly utilized as a metric to evaluate their ability to detect low-intensity light. [34]he maximum of the D* and the D* at V on (D*-V on ) are, respectively, shown in Figure 2c,d.The test results show that the D* of the device without N 2 annealing can be ignored, which are respectively 1.42 × 10 9 Jones at 700 nm, 1.34 × 10 9 Jones at The other two crucial optical parameters are external quantum efficiency (EQE) [35] and linear dynamic range (LDR). [31]The EQE characterizes the efficiency of converting photons into a stream of electrons, whereas LDR characterizes the linear response range of a photodetector.The calculated results of EQE and LDR are, respectively, shown in Figure 2e TFT-D exhibits the best performance in all optical parameters since it may produce more photogenerated carriers after illumination.This suggests that N 2 treatments should introduce new states in the energy band of IGZO thin film, which can absorb visible light.Shorter postannealing times increase a larger number of such states.Furthermore, the photocurrent of TFT-C as a function of time under visible light pulse illumination was also performed (Figure S3 in the Supporting Information).Consistent with other light sensors based on MO TFTs, [36] an obvious persistent photoconductivity effect [37] was observed.The optical response does not reach saturation even when applying a light pulse with a period of 100 s and a duty ratio of 50% (Figure S3 in Supporting Information).Based on the data shown in Figure S3 in the Supporting Information, the rising time and recovery time are calculated to be about 37 s and more than 50 s, which are defined as the time to reach 90% and 10% of the maximum of drain current, respectively.It should be noted that the real rising time is larger than 37 s.This phenomenon is mainly due to the composite relaxation time of the photogenerated electron-hole pair being greater than the switching speed, [37] or possibly due to a slow recombination process at the deep trap site. [13]o investigate the effects of N 2 treatments on IGZO TFTs, Xray diffraction (XRD) analysis, UV-visible spectra examination, TCAD simulations, and XPS evaluation were subsequently conducted.Shown in Figure 3a are XRD measurements of different kinds of IGZO thin films.In comparison with Film-A (control IGZO film), the IGZO films did not crystallize after N 2 treatments.This indicates that the use of N 2 treatment for achieving visible light detection in IGZO in this study is not based on the crystallization of the IGZO film.
The transmittance spectra of IGZO films on the glass substrate were also measured to analyze the influence of N 2 treatments on IGZO films, as shown in Figure 3b.The optical transmittance of N 2 -treated IGZO films decreases obviously in the range of visible light compared with O 2 -treat IGZO film.It can be inferred that N 2 treatment of IGZO films introduces subgap states in the IGZO film, which are favorable for trapping carriers and result in a decrease in light transmittance. [38]ext, TCAD simulations of TFT-A, TFT-B, TFT-C, and TFT-D in the dark conditions were performed.Shown in Figure 4a are the fitting results of the transfer characteristics of four kinds of IGZO TFT in a dark state and shown in Figure 4b are the corresponding density of states (DOS).In this work, the DOS of IGZO is assumed to consist of band-tailed states characterized by shallow energy levels and deep energy levels away from the forbidden band edge, respectively.The distribution functions of the DOS for the acceptor-like state defects (g A ) and donor-like state defects (g D ) can be expressed respectively as the following formulas [39] g and which can be summarized as exponentially decaying band-tailed states and deep energy levels characterized by Gaussian distributions.N TA and N TD are, respectively, the DOS at the conduction band edge and valence band edge, while W TA and W TD determine the change of band-tailed state in the forbidden band.N GA and N GD are, respectively, the peak value of DOS for the acceptor-like state and donor-like state, while W GA and W GD are their characteristic recessions.E GA and E GD are, respectively, the energy levels at energy peaks.It can be observed that the fitting data fit the experimental data well, as shown in Figure 4a.Typically, in IGZO TFTs, the donor bump state density is directly related to the V O concentration, while the acceptor tail state density is related to the quality of the channel. [40]For DOS, there is no change in acceptor bump state density and donor tail state density, as shown in Figure 4b.The acceptor tail state density changes negligibly and the donor bump state density in IGZO TFTs with N 2 treatments (TFT-B, TFT-C, and TFT-D) increases as compared to that without N 2 treatments (TFT-A).Shorter N 2 postannealing time results in a higher increase in the acceptor tail state density and donor bump state density.
IGZO film contains numerous defects, including V O . [41][43][44] When N 2 is introduced to replace O 2 during the sputtering process, V O repairment due to O 2 begins to slow down.The as-fabricated IGZO TFT without postannealing becomes extremely negatively shifted due to a huge number of electrons donated by V O in the channel (Figure S4 in the Supporting Information).If only O 2 is introduced during the postannealing process, the numerous V O in IGZO TFTs with N 2 treatment during the sputtering process could be quickly repaired, resulting in insensitivity to visible light (Figure S5 in the Supporting Information).Therefore, precise control of the postannealing atmosphere and postannealing time becomes critical.In other words, the second N 2 treatment during the postannealing process must partially repair the V O .
Compared with IGZO TFT without N 2 treatments (TFT-A), the V O content in IGZO TFT with N 2 treatments obviously increase (Figure 4b).The increased V O acts as subgap states in the IGZO bandgap and donates electrons, [45,46] resulting in a negative deviation of the device transfer curves (Figure 1b) [42] and causing a decrease in the optical transmittance of IGZO films (Figure 3b). [45]lso due to the introduction of subgap states, IGZO TFTs can absorb visible light with energy less than the bandgap width, then generate photogenerated carriers, make the transfer curve negatively biased, and lead to luminescence response.
To confirm this hypothesis, XPS analysis was conducted on four sets of IGZO films, as shown in Figure 4c-f.Film-A is the control IGZO film with a standard sputtering and annealing process without N 2 , while Film-B, Film-C, and Film-D are treated with N 2 sputtering process and annealing process.The postannealing time of Film-B, Film-C, and Film-D are respectively fixed at 5, 4, and 3 min, which is the same as that of TFT-B, TFT-C, and TFT-D.In general, the O 1s spectra can be divided into three sub-peaks with binding energy centered at 530 ± 0.5 (peak 1), 531 ± 0.5 (peak 2), and 532 ± 0.5 (peak 3), respectively, referring to metal─oxygen (M─O) bonds, V O and metal─hydroxyl (M─OH). [47]For the IGZO thin film after O 2 annealing (Film-A), the composition of peak 1: peak 2: peak 3 is 55.95%: 17.12%: 26.93% (Figure 4c).After N 2 annealing (Film-B to Film-D), the composition of peak 1: peak 2: peak 3 becomes 56.00%: 18.98%: 25.02%, 52.72%: 23.11%: 24.17%, and 53.07%: 25.58%: 21.35%.Compared with the untreated IGZO thin film (Film-A), the concentration of V O (peak 2) increases from 17.12% to 25.58%.Shortening the N 2 annealing time would further reduce the restoration time of V O , allowing more V O to be retained in the IGZO film.
To enhance comprehension of the involved in O 2 annealing for V O repair and N 2 annealing for inducing a visible light response in IGZO TFTs, the energy band diagrams for IGZO TFTs under different annealing conditions are presented in Figure 5.In general, the IGZO film will experience a conversion between a V O and a divalent oxygen vacancy (V O 2+ ), which can be expressed by the following formula [48] V Process 1 refers to V O 2+ converting into V O after obtaining two electrons, while process 2 is the inverse process.Due to the decrease in oxygen partial pressure during the device fabrication process and the thermal excitation within the device, a portion of V O s is more prone to converting into V O 2+ (Process 2).Conversely, Process 1 is less likely to occur due to the low DOS of the acceptor-like state in the IGZO near the bottom of the conduction band. [48]s shown in Figure 5a, at high temperatures, O 2 can enter into IGZO film and can react with V O present in the film, leading to the formation of oxides. [49]At the same time, the V O is repaired, indicated by a smaller and more transparent V O as shown in Figure 5c.The concentration of V O is thus reduced and Process 2 is attenuated, leading to the reduction of the generation of free electrons and a positive shift in the transfer curve.For N 2 treatments, as shown in Figure 5b, the conversion of V O in either Process 1 or Process 2 will not be affected.However, compared with O 2 treatments, the N 2 treatment can impede the repair of V O due to O 2 , leading to a higher concentration of V O remaining within the IGZO film and a negative shift in the transfer curve.
In the absence of N 2 treatment, the V O in TFT-A undergoes significant repair by O 2 , making it challenging to create subgap states within the forbidden band of IGZO (Figure 5c).Consequently, when the device is exposed to visible light of various wavelengths, TFT-A exhibits no light response due to the energy of the bandgap being higher than the energy of visible light [28] (Figure 1c).After the N 2 treatment, a higher concentration of unrepaired V O remains within the IGZO film.Compared to the O 2 treatment, the number of unrepaired V O s increases after the N 2 treatment, leading to an increase in V O 2+ and carrier concentration.Higher carrier concentration leads to an upward shift in the Fermi energy level, [50] as shown by the red dashed line in Figure 5d.The increase of V O 2+ is positively charged in the IGZO, thus reducing the potential barrier between the channel and source/drain, causing a negative V th shift of IGZO TFTs. [51]s a result, devices after N 2 treatment are easier to turn on.These unrepaired V O act as subgap states within the IGZO bandgap, represented by the blue lines in Figure 5b.When the N 2 -treated device is exposed to visible light, these subgap states assist in absorbing photons and exciting the electrons in the V O to the conduction band, while the V O converts to V O 2+ .This leads to an augmented flow of carriers into the conduction band, thereby increasing the carrier concentration in the IGZO channel.Consequently, the device turns on earlier, resulting in a negative shift in the transfer curves and enhanced visible light response (Figure 1).

Conclusion
In this work, the visible detection capability of IGZO TFTs was achieved through N 2 treatments during the sputtering and annealing process for the first time.The N 2 -treated TFTs exhibit a high sensitivity to visible light with the highest R of 0.66 A W −1 and the largest D* of 5.40 × 10 14 Jones.A V O modulation model is proposed by integrating XPS results and TCAD simulations.This work presents an efficient and process-compatible approach for fabricating IGZO TFTs with elevated visible light sensitivity, holding great potential for applications in the AM display industry.

Experimental Section
IGZO TFTs used in this work are of bottom-gate and top-contact structure, as shown in Figure 1a.The substrate is a heavily doped P-type silicon (P-Si) wafer covered with 100 nm SiO 2 .The P-Si substrate serves as the gate while the SiO 2 serves as the gate insulator.
For the control IGZO TFT (TFT-A), a 40 nm IGZO thin film was first deposited on the substrate by RF sputtering an IGZO target (In 2 O 3 :Ga 2 O 3 :ZnO = 1:1:1 at%) provided by ZhongNuo Advanced Material (Beijing) Technology Co., Ltd, using a shadow mask.The sputtering power, sputtering pressure and Ar/O 2 flow ratio during the sputtering were, respectively, fixed at 120 W, 3 mTorr and 28.5/4.5 sccm.The IGZO thin film was subsequently annealed through a rapid thermal process (RTP) with an annealing temperature of 400 °C, an annealing time of 30 min and a O 2 flow rate of 10 sccm.After the in-process annealing, a layer of Al was deposited by DC sputtering as the source and the drain using the shadow mask, followed by a post-RTP annealing with an O 2 flow rate of 10 sccm at 300 °C for 5 min.
For the IGZO TFTs with N 2 treatments (TFT-B, TFT-C, and TFT-D), during the sputtering process, the previous O 2 was replaced by N 2 .The Ar/N 2 flow ratio is fixed at 28.5/4.5 sccm.The other sputtering parameters were kept the same as the TFT-A.For the post-RTP annealing process, an additional N 2 flow was introduced and N 2 /O 2 flow ratio was fixed at 4/1 sccm.For the in-process annealing, except for the N 2 introduction, the other annealing parameters were still kept the same as the TFT-A.For the post-RTP annealing process, except for the N 2 introduction and annealing time variation, the other annealing parameters were still kept the same as the TFT-A.The postannealing time for TFT-B, TFT-C, and TFT-D are, respectively, 5, 4, and 3 min.
The transfer curves of IGZO TFTs were characterized by a Keysight B1500A semiconductor analyzer connected with a probe station.The visible light was provided by a tunable laser (Photonics, SC-PRO-7).If not emphasized, the light power densities at a specific wavelength were fixed at 250 μW cm −2 .The XPS and XRD characteristics were obtained by Thermo Scientific K-Alpha and Rigaku Smartlab, respectively.The transmittance of the IGZO films were characterized by a UV-VIS-NIR spectrometer (SHIMADZU, SolidSpec-3700).The TCAD simulation of IGZO TFTs was performed using Silvaco Atlas simulator.
The V th is defined as the intercept of the reverse extension of the transfer curve (V gs > 0) with the x-axis in linear coordinates.The V on is defined as the gate voltage when the drain current is 1.0 × 10 −11 A. The μ FE is extracted from the following equation [52]  FE = Ldg m W ox V ds (4)   where W, L, d, g m ,  ox , are, respectively, the device channel width, the channel length, the gate dielectric thickness, the maximum of device transconductance at V ds = 0.1 V and gate dielectric permittivity.The R, D*, EQE, and LDR at different wavelengths ( = 450, 500, 600, 700 nm) can be calculated using the following formulas respectively [31]  I light , I dark , P in , q, A, I light-max , and I photocurrent , respectively, refer to the drain current under illumination, the drain current in the dark state, the incident light power (equal to the incident light power density multiplied by the active layer area of the device), the basic charge, the effective area of the detector receiving light, the maximum drain current after illumination, and the drain current under illumination when the V gs is equal to the V on .It should be noted that the value of D* calculated from the above equation may be overestimated.The value of D* derived from the noise power spectral density measurements [30] would be more accurate.

Figure 1 .
Figure 1.a) Schematic diagram of a photodetector based on IGZO TFTs.b) Transfer curves of different kinds of IGZO TFTs.Transfer curves of c) TFT-A, d) TFT-B, e) TFT-C, and f) TFT-D under illuminations with different visible light  s .
TFTs used in this work are of the bottom gate and top contact structure.The schematic diagram is shown in Figure 1a.There are four kinds of IGZO TFTs fabricated under various conditions, namely TFT-A, TFT-B, TFT-C, and TFT-D.TFT-A is the control IGZO TFT with a standard sputtering and annealing process without N 2 involvement.TFT-B, TFT-C, and TFT-D are treated with N 2 during the sputtering process and an-nealing process.The post annealing time of TFT-B, TFT-C, and TFT-D are respectively fixed at 5, 4, and 3 min.The detailed fabrication conditions can be found in the Experimental Section.

Figure 2 .
Figure 2. Comparison of a) R, b) R-V on , c) D*, d) D*-V on , e) EQE, and d) LDR among different kinds of IGZO TFTs.

Figure 3 .
Figure 3. a) XRD measurements of different kinds of IGZO thin films.b) The transmittance spectra of the pure glass substrate and IGZO thin films.
,f.The trends for these two parameters are consistent with the trend for R and D*.The best EQE of TFT-A, TFT-B, TFT-C, and TFT-D are respectively 1886.64, 2295.51,1968.74, and 2750.16,while the best LDR are, respectively, 17, 63, 78, and121 dB, which are both achieved at  = 450 nm.

Figure 4 .
Figure 4. a) TCAD simulation of IGZO TFTs under different annealing conditions (symbols: experimental data, lines: fitting data).b) Distribution of state density in IGZO TFTs under different annealing conditions.XPS text results of c) Film-A, d) Film-B, e) Film-C, and f) Film-D under different annealing conditions.

Figure 5 .
Figure 5.The schematic diagram of V O in IGZO film a) after O 2 annealing, and b) after N 2 annealing.Energy band diagrams in IGZO TFT c) after O 2 annealing, and d) after N 2 annealing.