1. Introduction

Solar-blind ultraviolet (UV) detector plays an important role in medical sterilization, missile tracking, hydrogen detection, fire warning, ozone hole detection, non-line-of-sight communication, power grid security monitoring and space exploration^[1−4]. Due to the ultra-wide bandgap of 4.4−4.9 eV^{[5, 6]}, Ga₂O₃ has the absorption edge of near 250−280 nm which just located in the solar-blind UV range^{[7, 8]}. Moreover, its absorption coefficient is as high as 10⁵ cm⁻¹, so it is one kind of ideal natural solar-blind UV detection materials^{[9, 10]}.

Ga₂O₃ solar-blind UV detector is mainly based on metal−semiconductor−metal (MSM) type^[11], Schottky type^[12−14], heterojunction type^{[15, 16]}, and p−n junction type^[17], among which MSM photodetectors have the simple structure and excellent performance. Nowadays, the Ga₂O₃ detector arrays based on the MSM photodetectors and its imaging are developing rapidly. In 2021, Xie et al. prepared an 8 × 8 photodetector array based on Ga₂O₃ film grown by pattern technique, showing good uniformity and repeatability^[18]. Their patterned growth of Ga₂O₃ provides a new idea for the development of photodetector arrays. In 2022, Liu et al. used metal−organic chemical vapor deposion (MOCVD) to grow Ga₂O₃ films and prepared a 4 × 4 photodetector array. With the photo-to-dark current ratio (PDCR) of 6 × 10⁷ and responsivity (R) of 634.15 A/W, the 16 units of the photodetector array exhibit good uniformity^[19]. In 2022, Ding et al. used inkjet printing technology to prepare a 4 × 4 photodetector array. Because of the high designability of inkjet printing, separated Ga₂O₃ film arrays can be well designed without the traditional imaging process, and all pixel units have good uniformity^[20]. In 2023, Shen et al. prepared an 8 × 8 photodetector array, which showed uniform responsivity and low chromatic aberration of graphic imaging at 5 V^[21]. In 2022, Hou et al. prepared a 10 × 10 photodetector array based on Ga₂O₃ film grown by defect and doping engineering, and reported the high temperature imaging under the conditions of 12 V and 280 °C for the first time^[22]. In 2023, Shen et al. prepared a 16 × 16 photodetector array based on Ga₂O₃ film grown by MOCVD, and further output photo-imaging with "0" and "1" signals^[23]. These photodetector arrays mentioned above almost adopt the single metal layer photolithography process. For the development of the arrays to a larger scale, the area of a single pixel unit and the wire arrangement should be a big challenge. Obviously, the single metal layer process cannot meet this challenge well. Therefore, the study of detector array based on multilayer metal process or crossbar structure should be very important and significant.

In this paper, a 10 × 10 MSM solar-blind UV detector array with double-layer wire structure is prepared based on Ga₂O₃ film grown by atomic layer deposition (ALD), which is designed by structural optimization with row/column-selective metal wires, and isolated at the row-column intersections by Al₂O₃ material. The detection characteristics of the unit photodetector at low bias and different light intensity (P) were investigated, and the response uniformity and imaging performance of the whole array at different P, bias and high temperature are analyzed.

2. Experiment

At first, the sapphire substrate was cleaned by ultrasonic, using acetone, anhydrous ethanol and deionized water for 10 min in turn. Next, Ga₂O₃ film with 120 nm thickness was grown on the sapphire substrate by ALD. Subsequently, Ga₂O₃ film was annealed for 10 min under nitrogen atmosphere and the surface morphology of the samples was observed by characterization technique. Next, the first photolithography and evaporation of metal (Ti/Au (100/50 nm)), was used to prepare the first metal layer (right side of the interdigital electrodes) as well as the column wires in the photodetector array. Afterwards, an Al₂O₃ film was grown as an insulating layer by ALD and etched it. Then the second photolithography and evaporation of metal (Ti/Au (100/50 nm)), was used to prepare the second metal layer (left side of the interdigital electrodes) and row wires. Fig. 1 illustrates this whole flow of process. Finally, the array was rapidly annealed to improve its Ohmic contact performance. The length, width and spacing of the interdigital electrodes are 300, 50 and 50 μm, respectively. The characteristic of the photodetector array was measured by Keysight B1505A semiconductor analyzer. A handheld UV mercury lamp with a wavelength of 254 nm was used in this process for light response measurements. During the measurement, P depends on the distance between the UV lamp and the samples. The distance was controlled by a lamp holder table, the height of which can be adjustable. To ensure stable and accurate P, a 254 nm-UV light intensity meter was used to measure the actual P.

3. Results and discussion

Fig. 2(a) shows the X-ray diffraction (XRD) pattern of Ga₂O₃ film grown on sapphire substrate by ALD technique and rapidly annealed at 700 °C. Besides strong sapphire substrate diffraction peak, the diffraction peaks can be observed at 18.95°, 38.62°, and 59.36°, which correspond to the (−201), (−402), and (−603) crystallographic planes of β-Ga₂O₃, respectively (PDF card: No. 43−1012). The three planes are parallel to each other and belong to the same (−201) crystal plane family, indicating that the film is β-Ga₂O₃ film with a single growth orientation and good quality. Fig. 2(b) shows the UV−visible absorption spectrum of the grown β-Ga₂O₃, measured in the wavelength range (190−750 nm). Noted that the film has a very low absorption rate of only 10% in the visible range (400−750 nm) and a significant UV absorption in the solar-blind UV band (200−280 nm). The (αhν)²−hν curve is derived, as shown in the inset of Fig. 2(b). Then the film’s band gap E_g is calculated to be 4.88 eV based on the Tauc method which is as follows^[24].

$${\left( {\alpha h\nu } \right)^2} = \rm{C}\left( {h\nu - {E_{\text{g}}}} \right) ,$$

(1)

where α is the absorption coefficient, C is a constant, h is Planck constant, ν is the frequency of the photon. Fig. 2(c) shows the image of prepared array, the size of which is 1.86 cm × 2.08 cm and the enlarged view in details, respectively. It can be seen that there is no metal adhesion between the interdigital electrodes and the electrode edges are clear.

For detector, the photo current (I_photo) comes from the physical mechanism of photo-genereated electron−hole pair when the UV light irradiates it, as shown in Fig. 3(a). In order to test the photoelectric performance of the device, the individual pixel units were firstly tested one by one under the conditions of dark and solar-blind UV illumination with P of 1800 μW/cm². The experimental results are shown in Fig. 3(b). At the bias of 3 V, the dark current (I_dark) of the device is as low as about 18.5 pA, and I_photo can reach 10.4 μA. The PDCR is more than or close to 10⁶. This maybe result from the good quality of grown Ga₂O₃, which can suppress the generation of I_dark and produce a large number of excess photo-generated electron−holes under 254 nm UV light. The high PDCR means that the device has high sensitivity and can detect extremely weak UV light, which is also the basis for the array to work under extreme conditions. Figs. 3(c) and 3(d) present the distribution of I_dark and I_photo, carried out by Gaussian distribution statistics, respectively. It can be seen that each pixel unit has highly similar I_dark and I_photo, indicating that all pixel units in the array have good uniformity.

Figs. 4(a) and 4(b) show the linear and semi-logarithmic I_photo curves measured under 254 nm UV light with different P. It can be seen that I_photo increases with increasing P from 400 to 2945 μW/cm². Generally, I_photo has a power law relationship with P^[21]. Fig. 4(c) shows that I_photo also follows this power law and I_photo∝P^0.50, indicating that the device has a good regularity as P changes. To further explore the device’s performance under different P, R, external quantum efficiency (EQE) and detectivity (D^*) were extracted from Fig. 4(a) and Fig. 3(b). R reflects the ability of the device converting incident light into I_photo^[25] and is expressed as：

$$R = \frac{{{I_{{\text{photo}}}} - {I_{{\text{dark}}}}}}{{P \times S}} ,$$

(2)

where S is the effective area. The higher R is, the more photo-generated electron−hole pairs under a certain irradiation, and the better device performance. According to Eq. (2), R is inversely proportional to P. So R decreases with the increase of P, as shown in Fig. 4(d). When P = 1800 μW/cm², R can reach 4.28 A/W at 3 V. EQE is the ratio of incident photons to photo-generated electron−hole pairs. It is an important index to measure the photon utilization rate of the detector^[25]. EQE is as follows:

$${\text{EQE}} = \frac{{hcR}}{{q\lambda }} ,$$

(3)

where q is the electron’s charge, c is the velocity of incident light, λ is the wavelength of UV light, and R is the responsivity. According to Eq. (3), EQE of prepared device is calculated as 2.1 × 10³%. EQE is proportional to R according to Eq. (3) and so it also decreases as increasing P in Fig. 4(d). The high R and EQE maybe related to the photo-generated carriers that can participate in the transport in the active layer fast and effectively^[26]. D^* is proposed on the basis of noise equivalent power in order to represent device performance more accurately^{[27, 28]}.

$${D^*} = \frac{R}{{\sqrt {2q{I_{{\text{dark}}}}/S} }} .$$

(4)

Based on Eq. (4), the curve of D^* with the change of P was extracted, as shown in the inset of Fig. 4(d). D^* decreases with decreasing R and then decreases with increasing P, and the maximum D^* is 1.5 × 10¹⁴ Jones at 400 μW/cm²_. Although this extracted curve of D^* changes with P, according to the exact definition, D^* should be independent of P, so we only use D^* corresponding to the lowest P to represent the detectivity of the device.

Fig. 5(a) shows the current−time (I−T) curves of the device at different bias under P of 1800 μW/cm². It can be seen that when the applied bias is changed, the device has good time response characteristics to different applied bias. In order to further explore the response speed of the device, there are two parameters: rising time (τ_r) and decay time (τ_d)^[29]. τ_r is the time of the I_photo increasing from 10% to 90% of the maximum value. On the contrary, τ_d is the time of I_photo decreasing from 90% to 10% of its maximum value. As shown in Fig. 5(b), a cycle was extracted from the I−T curve of Fig. 5(a) at 5 V. The device has a fast response speed, which can be seen from that, τ_r and τ_d of the device calculated based on the experimental data are 74 and 130 ms, respectively. Noted that τ_r and τ_d are relatively long. This long response time should be related with the quality of the material^[30]. Fig. 5(c) shows the I−T curve of the device under different P at 5 V. As increasing P, I_photo increases. This is attributed to that the greater P is, the more photo-generated carriers are generated, and the greater I_photo is. Subsequently, the I−T curve under multiple switching cycles was tested. It can be seen from Fig. 5(d) that the device has high light response characteristics and maintains nearly consistent I_photo after multiple cycles, indicating that the device has good reproducibility and robustness.

Fig. 6(a) presents the imaging system for array, including UV light source, photomask, data collection system and temperature-controlled heater. Figs. 6(b)−6(d) show the experimental imaging results of the array under different P at room temperature (RT) at 5 V. It can be seen that the I_photo in the array decreases slightly with the decrease of P, but it can still be clearly imaged, which indicates that the array also has a sensitive response to weak light. Figs. 6(e) and 6(f) show the imaging results when P is 1800 μW/cm² and the voltage is 3 and 1 V at RT, respectively. The array imaging effect is good under the ultra-low voltage. This array can be applied in low-voltage circuits, which greatly reduces the circuit power consumption. Noted that I_dark, I_photo, PDCR and R increase as bias increases under P of 1800 μW/cm², as shown in Figs. 6(g) and 6(h).

Table 1 summarizes the key performance parameters of different photodetectors and photodetector arrays at RT reported in the literature. The performance of the photodetector array prepared in this work is at a high level.

Further, Figs. 7(a) and 7(b) present the imaging results of 5 and 1 V under different temperatures, respectively. All of these images are very clear even at ultra-low operating voltage of 1 V. Their PDCRs of 5 and 1 V under the given temperatures are similar to each other, which changes from 10⁵ to 10¹. With the increase of temperature, both of I_photo and I_dark show an increasing trend in Fig. 7(c). I_dark of the array at 50, 80, 110, and 150 °C are significantly greater than that at RT, but I_photo are almost as same as that at RT. Noted that these PDCRs are about 10² in Fig. 7(d), indicating that the array’s performance remains stable when the temperature located in the range of 50−150 °C, suggesting that the array has excellent temperature stability. For the cases of over 200 °C, I_photo and I_dark of the array increase significantly, and the performance begins to deteriorate as temperature increases. However, even if PDCR worsened under high temperature conditions, clear images can still be obtained at 300 °C. In Fig. 7(d), under 5 V, PDCR and R can reach 1.65 × 10¹ and 34.4 A/W, respectively. And in this case EQE is as high as 1.7 × 10⁴%. Noted that even at 300 °C, PDCR still reaches 1.70 × 10¹ at 1 V. In order to study the performance of the array at 300 °C, the uniformity diagram is shown in Fig. 7(e) using the gray image processing technology. The results show that the gray image is very clear. Table 2 compares the performance parameters of Ga₂O₃ photodetectors at maximum temperature reported by far with this work. The devices prepared in this work still have excellent performance at a high temperature of 300 °C.

4. Conclusion

In this paper, β-Ga₂O₃ film was grown on sapphire substrate by ALD. A 10 × 10 solar-blind UV detector array with MSM structure was prepared by two photolithography and evaporation, and etch. The uniformity of pixel units in the array and the photoelectric performance of single detection units were investigated. Under 3 V, the detector has ultra-low I_dark (1.85 × 10⁻¹¹ A), high R (4.28 A/W), high EQE (2.1 × 10³%) and high D^* (1.5 × 10¹⁴ Jones), and the PDCR reaches 5.5 × 10⁵. In addition, the array has good time response characteristics, the response time is only 100/150 ms. The imaging characteristics of the array at different temperatures are investigated by building an imaging system, and the array presents clear imaging at a high temperature of 300 °C. When the applied voltage is 5 V, PDCR is 1.65 × 10¹, R is as high as 34.4 A/W, and EQE reaches 1.7 × 10⁴%, showing excellent temperature stability and high temperature resistance. There are some beneficial explorations for the application of Ga₂O₃ solar-blind UV detector array in this work. Different from the single-layer wire structure used widely, the double-layer wire structure is adopted through structural optimization in the detector. And at ultra-low voltage, the array presents high imaging performance, which could reduce the circuit power consumption effectively. Meanwhile, this array still realized a good imaging under very high temperature of 300 °C. The experimental results show that the double-layer array structure has good characteristics. Furthermore, based on this structure, photodetector array should be easy to expand to the larger scale for high pixel density. Of course, there should be the detector uniformity and crosstalk issues as the array becomes too large. However, in turn, seeking solutions to these issues should also drive the development of Ga₂O₃ image sensor.

Acknowledgments

This work was supported by Natural Science Basic Research Program of Shaanxi Province of China (No. 2023-JC-YB-574) and National Natural Science Foundation of China (Grant No. 62304178).