Enhancing the UV Response of All-Inorganic Perovskite Photodetectors by Introducing the Mist-CVD-Grown Gallium Oxide Layer

Abstract

All-inorganic perovskites, with their low-cost, simple processes and superior heat stability, have become potential candidate materials for photodetectors (PDs). However, they have no representative responsivity in the deep-ultraviolet (UV) wavelength region. As a new-generation semiconductor, gallium oxide (Ga₂O₃), which has an ultrawide bandgap, is appropriate for solar-blind (200 nm–280 nm) deep-UV detection. In this work, ultrawide-bandgap Ga₂O₃ was introduced into an inorganic perovskite device with a structure of sapphire/β-Ga₂O₃/Indium Zinc Oxide (IZO)/CsPbBr₃. The performance of this perovskite PD was obviously enhanced in the deep UV region. A low-cost, vacuum-free Mist-CVD was used to realize the epitaxial growth of β-Ga₂O₃ film on sapphire. By introducing the Ga₂O₃ layer, the light current of this heterojunction PD was obviously enhanced from 10⁻⁸ to 10⁻⁷, which leds its detectivity (D*) to reach 1.04 × 10¹² Jones under a 254 nm light illumination with an intensity of 500 μW/cm² at a 5 V bias.

1. Introduction

Photodetectors (PDs) are recognized as advanced optoelectronic devices, which play a significant role in information transmission [1,2,3]. Among them, ultraviolet (UV) PDs have wide applications in the fields of ultraviolet communication and imaging for both military and commercial uses [4,5,6]. Compared with the traditional silicon material, perovskite can be processed in a low-cost and simple manner. At the same time, it possesses excellent photoelectric properties that make it a superior potential candidate material for PDs, especially for all-inorganic perovskites [7,8]. Iinorganic perovskite PDs have been reported to exhibit excellent photodetection performance in both the visible- and near-UV spectra. However, they have no representative responsivity in the deep UV wavelength region. As a new-generation of semiconductor, gallium oxide (Ga₂O₃), which has an ultrawide bandgap of around 5.0 eV and a high breakdown field strength of about 8 MV/cm, has attracted more and more focus from researchers [9,10,11]. Because its cut-off absorption wavelength is below 280 nm, Ga₂O₃ is appropriate for the solar-blind deep UV region (200 nm–280 nm) [4,12,13,14]. If Ga₂O₃ is introduced into an inorganic perovskite device, the performance of perovskite UV PDs might be enhanced.

In typical perovskite devices, a Transparent Conducting Oxide (TCO) is usually used as a common electrode, and Ga₂O₃ is also reported to be used as a transport layer. For example, in our previous work [15], we used a structure of fluorine-doped tin oxide (FTO)/Ga₂O₃/SnO₂/perovskite/spiro-OMeTAD/Ag in perovskite solar cells. It was shown that Ga₂O₃ has a deep-valence-band minimum value and can thus function as an effective blocking layer. This led the devices it was used in to achieve power conversion efficiencies (PCEs) of 21.56%. Furthermore, we used Ga₂O₃ as an interlayer in PDs that greatly suppressed carrier recombination at interfaces. These PDs, which had structures of FTO/TiO₂/Ga₂O₃/CsPbIBr₂/Carbon, achieved a detectivity (D*) as high as 1.83 × 10¹² Jones [16]. Some reports also revealed the high performance of PDs by combining inorganic wide-bandgap semiconductors and perovskites. For example, Dong’s work used a Au/CH₃NH₃PbI₃/β-Ga₂O₃/Ti/Au hybrid structure. This structure’s photo-to-dark current ratio was 1460, and its responsivity (R) and D* were 85 mA/W and 1.28 × 10¹¹ Jones [17], respectively. Gao’s work showed lead-free perovskite PDs with heterojunction structures of CsCu₂I₃/β-Ga₂O₃ [18]. The β-Ga₂O₃ film was prepared with MOCVD, and the optimized devices reached a D* of 10⁷ Jones. Qi et al. used MOCVD to heteroepitaxially grow the β-Ga₂O₃ film and then combined it with CuBiI₄ to achieve a self-powered heterojunction PD [19]. Gong’s work prepared the β-Ga₂O₃ film using a molecular beam epitaxy (MBE) technique, resulting in metal–semiconductor–metal perovskite PDs with structures of β-Ga₂O₃/Au/MAPbBr₃ [20]. In addition to single-crystal Ga₂O₃ films, many researchers also used Ga₂O₃ nanowires and amorphous Ga₂O₃ to prepare Ga₂O₃ and perovskite hybrid devices. Palepu’s work used a vertical structure of β-Ga₂O₃ nanowires/CH₃NH₃PbI₃ heterostructures and exhibited a D* of 5.23 × 10¹¹ Jones under self-powered conditions [21]. Liu’s work reported a device with a structure of FTO/α-Ga₂O₃ nanowires/MAPbBr₃/Ag and a D* of 1.21 × 10¹² Jones [22]. Liu’s work also reported the radio-frequency magnetron sputtering of the prepared amorphous Ga₂O₃ film. Devices with structures of FTO/Ga₂O₃ (amorphous)/MAPbCl₃/spiro-OMeTAD/Ag had a D* of 5.4 × 10¹⁰ Jones [23].

However, most of the above reports prepared their Ga₂O₃ film using conventional methods such as PLD, MOCVD, or MBE, which have high costs and require a vacuum environment. Compared with these growth methods, Mist-CVD has the advantages of being vacuum-free, a simpler process, and low-cost [24,25,26,27,28]. It is expected to become a mainstream method for growing Ga₂O₃ film epitaxially on other semiconductor films and to push down the cost of growing epitaxial Ga₂O₃ films. In this work, Mist-CVD was used to grow Ga₂O₃ films on sapphire. We prepared a heterojunction PD with a structure of sapphire/β-Ga₂O₃/Indium Zinc Oxide (IZO)/CsPbBr₃. After the epitaxial growth of the β-Ga₂O₃, IZO electrodes and CsPbBr₃ films were deposited onto it. It was shown that the response of the heterojunction PD in the deep UV wavelength region was clearly enhanced compared to its perovskite-only counterpart. The device, which has a sapphire/β-Ga₂O₃/IZO/CsPbBr₃ structure, reached a D* of 1.04 × 10¹² Jones, which was illuminated by a 254 nm light at an intensity of 500 μW/cm² at a 5 V bias.

2. Materials and Methods

2.1. Materials Preparation

Ga₂O₃ was grown using the Mist-CVD method, which is a simple method that is expected to become mainstream for the epitaxial growth of Ga₂O₃ [24,25]. A 2-inch c-plane sapphire sample was cleaned with deionized water, acetone, and ethanol, successively, in an ultrasonic cleaner. A 0.05 mol/L Ga₂O₃ precursor source solution was made from 1.83 g Ga(Acac)₃ (Aladdin, 99.99%), 1 mL hydrochloric acid, and 100 mL ultra-pure water. We used 1.7 MHz ultrasonic atomizers to produce small droplets of the solution, which were then transported to the substrate using a mixed gas of nitrogen and compressed air. The β-Ga₂O₃ film was epitaxially grown using Mist-CVD at 850 °C. Its growth rate was controlled at 400 nm/h, and it took 1 h to obtain a high-quality Ga₂O₃ film.

2.2. Fabrication of PDs

Firstly, the Ga₂O₃ sample was cut into 2 cm × 2 cm pieces and cleaned again. A 100 nm IZO layer was deposited onto the sample by a sputter system with a shadow mask. The PbBr₂ precursor was obtained by dissolving 367 mg of PbBr₂ into DMF at 80 °C with sufficient stirring. Then, 80 μL of PbBr₂ solution was spin-coated onto the sample with the IZO electrode at 2000 rpm for 30 s. This was followed by a thermal treatment at 90 °C for 30 min. Next, 110 μL of a CsCl/H₂O solution with a concentration of 1 M was spin-coated onto the PbBr₂ film at 2000 rpm for 30 s. Finally, the sample was annealed at 250 °C to fulfill the crystallization of the CsPbBr₃ film, and thus, the fabrication of the CsPbBr₃/Ga₂O₃ UV PDs was finished.

2.3. Measurement Details

A Bruker D8 Advance powder X-ray diffractometer and spectrophotometer (U-4100, Hitachi, Tokyo, Japan) were used to record the X-ray diffraction (XRD) results and the UV-vis transmittance spectra, respectively. A Keysight 1505A semiconductor analyzer, a Keithley 2636B, and a Keithley 2450 recorded the dark current, light current, and time-dependent photo-response characteristics, respectively.

3. Results

The preparation procedures and the structure configurations of Ga₂O₃/CsPbBr₃ PDs are shown in Figure 1a. At high temperatures, small droplets of the gallium–acetylacetonate aqueous solution produced a chemical reaction on the surface of the sapphire that caused the Ga₂O₃ film to grow. Then, the IZO layer was used as an electrode between the β-Ga₂O₃ and the CsPbBr₃ films. After spin coating and annealing, the CsPbBr₃ perovskite was deposited on the IZO/β-Ga₂O₃ film. Figure 1b shows the cross-section of the CsPbBr₃/Ga₂O₃/Sapphire film. The thickness of the Ga₂O₃ film was about 420 nm and that of the CsPbBr₃ was about 410 nm. The results of an EDX (Energy-Dispersive X-ray Spectroscopy) also showed a clear demarcation. Although the CsPbBr₃ was grown on the β-Ga₂O₃, the CsPbBr₃ film was composed of the grain’s boundaries, as shown in the inset image. This indicates that every grain shows good contact with the others. As shown in Figure 1c, the XRD spectra exhibited a single β-crystal phase for the grown Ga₂O₃. The peaks of 18.8°, 38.5°, and 59.6° could be assigned to the (−201), (−402), and (−603) planes of the β-Ga₂O₃, respectively. This is consistent with our previous report that Mist-CVD is capable of growing single β-Ga₂O₃ crystals [26,27]. Additionally, a part of the data from the XRD have been enlarged in Figure 1d. There are some peaks (around 15.6°, 22.1°, and 31.1°) which prove that the CsPbBr₃ film was successfully deposited on the surface.

Moreover, the UV-vis spectra of the β-Ga₂O₃ film and the CsPbBr₃/IZO/β-Ga₂O₃ sample are demonstrated in Figure 1e. The light could transmit through the β-Ga₂O₃ film after about 250 nm. After 300 nm, the light transmission of the β-Ga₂O₃ film was around 80%. The bandgap of β-Ga₂O₃ can be calculated using the following formula [28,29]:

E_g ≈ 1240/λ

where λ is the absorption cutoff edge of the light, and E_g is the bandgap of the semiconductor material. The E_g of the β-Ga₂O₃ film is about 4.95 eV. The inset plot also exhibits this E_g. Because the bandgap of CsPbBr₃ is the smallest bandgap in structure of the CsPbBr₃/IZO/β-Ga₂O₃, the the CsPbBr₃/IZO/β-Ga₂O₃ sample shows transmittance after 500 nm. This is in line with the bandgap of CsPbBr₃, which has a value of about 2.40 eV. In these results, the CsPbBr₃ shows light absorption at the UV wavelength, but the β-Ga₂O₃ cannot absorb light after 258 nm.

Figure 2a demonstrates the dark currents of the Ga₂O₃ PD and Ga₂O₃/CsPbBr₃ heterojunction PD. The Ga₂O₃ PD reached a dark current of 7.00 × 10⁻¹¹ A at a −5 V bias. However, after the successful deposition of the CsPbBr₃, the dark current of the Ga₂O₃/CsPbBr₃ heterojunction PD exhibited a small increase and achieved 2.56 × 10⁻¹⁰ A at a −5 V bias. This phenomenon is attributable to the introduction of defects with a polycrystalline-phase CsPbBr₃. Moreover, the CsPbBr₃ device was also prepared and achieved 2.09 × 10⁻⁸ A at a 5 V bias under 254 nm of illumination with a light intensity of 500 μW/cm². The Ga₂O₃ PD reached 2.01 × 10⁻⁶ A under the same measurement conditions, which demonstrates an increase of 2 orders of magnitude. In Figure 2c,d, the Ga₂O₃/CsPbBr₃ heterojunction PD exhibits a photo response to 254 nm, 365 nm, and 405 nm with a light density of 500 μW/cm², 800 μW/cm², and 800 μW/cm², respectively. When the light was illuminated from the CsPbBr₃ side, as in Figure 2c, the Ga₂O₃/CsPbBr₃ heterojunction PD showed currents of 2.39 × 10⁻⁷ A, 7.45 × 10⁻⁸ A, and 9.66 × 10⁻⁸ A under a bias voltage of 5 V for the light wavelengths of 254 nm, 365 nm, and 405 nm, respectively. When the light was illuminated from the sapphire side, as in Figure 2d, the Ga₂O₃/CsPbBr₃ heterojunction PD showed a current of 2.03 × 10⁻⁶ A, 7.11 × 10⁻⁸ A, and 6.18 × 10⁻⁸ A under a bias voltage of 5 V for the light wavelengths of 254 nm, 365 nm, and 405 nm, respectively. These data have been collated in

Table 1. The light current statistics of the Ga₂O₃/CsPbBr₃ heterojunction PD.

	From the CsPbBr3 Side (Based on 5 V)	From the Sapphire Side (Based on 5 V)
254 nm (500 μW/cm2)	2.39 × 10−7 A	2.03 × 10−6 A
365 nm (800 μW/cm2)	7.45 × 10−8 A	7.11 × 10−8 A
405 nm (800 μW/cm2)	9.66 × 10−8 A	6.18 × 10−8 A

. For the sake of description, light illuminated from the sapphire side and from the CsPbBr₃ side are abbreviated as “from the sapphire side” and “from the CsPbBr₃ side”.

As shown in Table 1, the Ga₂O₃/CsPbBr₃ heterojunction PD demonstrated similar response properties under 365 nm and 405 nm lights from both the CsPbBr₃ side and the sapphire side. This is because Ga₂O₃ cannot absorb long-wavelength lights, which can only be absorbed by CsPbBr₃ regardless of whether they come from the CsPbBr₃ side or the sapphire side. Under the 254 nm light’s illumination, the current’s response exhibited an order-of-magnitude change. That is to say, the light current from the sapphire side increased tenfold compared to the CsPbBr₃ side. This phenomenon was maintained after multiple measurements. This is because the 254 nm light can be absorbed by both the Ga₂O₃ and CsPbBr₃ films. When the 254 nm light was illuminated from the CsPbBr₃ side, it was absorbed by the CsPbBr₃ film. When the 254 nm light is was illuminated from the sapphire side, it was absorbed mainly by the Ga₂O₃ film. This indicates that the responsivity in the deep-ultraviolet (UV) wavelength region was mainly determined by the absorption of Ga₂O₃ in the Ga₂O₃/CsPbBr₃ heterojunction PD. With the introduction of the Ga₂O₃ layer, the photocurrent increased from 10⁻⁸ to 10⁻⁷ A at a bias voltage of 5 V when compared with the Ga₂O₃/CsPbBr₃ heterojunction PD and the CsPbBr₃-only device. This obviously indicates that the ability of the CsPbBr₃ film to absorb UV light increased with the help of β-Ga₂O₃.

The responsivity (R) and the D* of the Ga₂O₃/CsPbBr₃ heterojunction PD can be calculated using the formulas [29]:

R = (I_photo − I_dark)/P_λS

D* = RS^1/2/(2qI_dark)^1/2

where P_λ is the radiation intensity of the light; S is the effective illumination area of the device; I_photo and I_dark are the light and dark currents of the device, respectively; S is the effective illumination area of the device; and q is the quantity of the electric charge.

The responsivity (R) of the Ga₂O₃/CsPbBr₃ heterojunction PD that was illuminated from the sapphire side is shown in Figure 3a. Because the Ga₂O₃ film exhibited great absorption of the 254 nm light, its R was obviously increased and reached 0.016 mA/W with a light density of 500 μW/cm² at a 5 V bias. The PD achieved 0.493 mA/W and 0.567 mA/W with 365 nm and 405 nm of illumination, respectively, under a light density of 800 μW/cm² at a 5 V bias. When illuminated from the CsPbBr₃ side, the PD reached 1.91 mA/W, 0.594 mA/W, and 0.771 mA/W, respectively. The D* values of the Ga₂O₃/CsPbBr₃ heterojunction PD reached 1.22 × 10¹¹ Jones, 3.80 × 10¹⁰ Jones, and 4.93 × 10¹⁰ Jones under 254 nm, 365 nm, and 405 nm of illumination, respectively, all with a 5 V bias. In contrast, when illuminated from the sapphire side, the Ga₂O₃/CsPbBr₃ heterojunction PD achieved D* values of 1.04 × 10¹² Jones, 3.15 × 10¹⁰ Jones, and 3.63 × 10¹⁰ Jones, respectively. These data have been collected in

Table 2. The R and D* of the Ga₂O₃/CsPbBr₃ heterojunction PD.

	R (From the CsPbBr3 Side)	D* (From the CsPbBr3 Side)	R (From the Sapphire Side)	D* (From the CsPbBr3 Side)
254 nm	1.91 mA/W	1.22 × 1011 Jones	0.016 mA/W	1.04 × 1012 Jones
365 nm	0.594 mA/W	3.80 × 1010 Jones	0.493 mA/W	3.15 × 1010 Jones
405 nm	0.771 mA/W	4.93 × 1010 Jones	0.567 mA/W	3.63 × 1010 Jones

. The R and D* of the heterojunction PD showed similar response properties above 254 nm from either side. Therefore, in the forthcoming section, light wavelengths above 254 nm will not be distinguished between the CsPbBr₃ side and the sapphire side.

Table 3. Comparison of perovskite-/Ga₂O₃-based UV PDs reported.

Structure	Ga2O3 Deposit Method	D* (Jones)	Reference
Ti/Au/CsCu2I3/β-Ga2O3(Ti/Au)	MOCVD	107 (10 V, 254 nm, 200 μW/cm2)	[16]
Au/CH3NH3PbI3/Au	bulk material	1.28 × 1011 (254 nm, 0.86 μW/cm2)	[17]
Ti/Au/CuBiI4/β-Ga2O3(Ti/Au)	MOCVD	3.72 × 1015 (5 V, 254 nm, 0.1 μW/cm2)	[19]
β-Ga2O3/Au/MAPbBr3	MBE	7.3 × 1010 (0 V, 240 nm)	[20]
Ag/NiO/CH3NH3PbI3/β-Ga2O3/n-Si	Sputtering (β-Ga2O3 nanowire)	5.23 × 1011 (0 V, 275 nm, 2.54 μW/cm2)	[21]
Ag/Spiro-OMeTAD/MAPbCl3/ZGO/FTO	spin-coated (ZGO)	1.21 × 1012 (0 V, 398 nm)	[22]
β-Ga2O3/IZO/CsPbBr3	Mist-CVD	1.04 × 1012 (5 V, 254 nm, 0 V, 500 μW/cm2)	This work

shows a comparison of the perovskite-/Ga₂O₃-based UV PDs reported. Most reports have prepared Ga₂O₃ film using conventional methods such as PLD, MOCVD, or MBE, which require a vacuum environment and have high costs. This work suggests that using Mist-CVD could decrease costs and achieve higher performance.

Figure 4a presents the time-dependent photo responses of the Ga₂O₃/CsPbBr₃ heterojunction PD at a 5 V bias under a 254 nm light from both the sapphire side and the CsPbBr₃ side. Using a timer switch every ten seconds, and under a 254 nm UV light with a light density of 500 μW/cm² and a 405 nm laser with a light density of 1 μW/cm², the on/off switching behavior also presents a smooth curve after manifold cycles.

The time-dependent photo response of the Ga₂O₃/CsPbBr₃ heterojunction PD from the sapphire side is higher than it is from the CsPbBr₃ side. As a PD, it is important for the device to obtain data quickly. The response times of devices have been calculated using the following formula [29]:

I = I₀ + Ae^(−t/τ1) + Be^(−t/τ2)

where I₀ is the steady-state photocurrent, A and B are the fitting constants, and τ₁ and τ₂ are the relaxation time constants for the fast and slow components, respectively. Under a 254 nm light, this PD achieved increased relaxation times of 2.22 s (slow) and 0.31 s (fast) when illuminated from the sapphire side. When illuminated from the CsPbBr₃ side, the PD reached 2.36 s (slow) and 0.55 s (fast). Under 405 nm illumination, the PD reached 0.05 s (slow) and 0.05 s (fast). These results show that the Ga₂O₃/CsPbBr₃ heterojunction PD’s performance is superior.

Moreover, the relationship between current density (J) and light intensity is shown in Figure 4c on the test results of the device under 254 nm of illumination with a light intensity ranging from 100 μW/cm² to 700 μW/cm² at a 5 V bias. The light–current density values versus the light–intensity plots, along with the fitting results obtained using the power law of J−P^α, are shown in Figure 4c. As fitted with the power law of J−P^α, the device received an α value of about 0.99. Figure 4d shows the power law results of the Ga₂O₃/CsPbBr₃ heterojunction PD under a 532 nm light with a density ranging from 200 μW/cm² to 2.5 μW/cm² at a 5 V bias, where this device received an α value of about 0.86. When the value of α approaches 1, it suggests that the photocurrent has linear dependence on incident light–power density. It also indicates that photogenerated carriers can separate efficiently [16]. Therefore, these data prove that the introduction of Ga₂O₃ into an inorganic perovskite device can enhance the performance of perovskite UV PDs.

4. Conclusions

Perovskites can be prepared using a low-cost and simple process, thus becoming a potential candidate material for PDs. As a deep-ultraviolet photodetector material, Ga₂O₃ has also attracted great attention from many researchers due to its solar-blind nature. To enhance the UV detectivity of perovskite PD, we combined the Ga₂O₃ with a CsPbBr₃ perovskite to prepare a heterojunction PD. We used a low-cost method to prepare the β-Ga₂O₃ using Mist-CVD, and the CsPbBr₃ exhibited good coverage on the β-Ga₂O₃ film. This Ga₂O₃/CsPbBr₃ heterojunction PD exhibited a high response for 254 nm lights, and the R of the PD increased and reached 0.016 mA/W under a 5 V bias and a light density of 500 μW/cm². The detectivity of the Ga₂O₃/CsPbBr₃ heterojunction PD achieved 1.04 × 10¹² Jones. The PD exhibited a good linear dependence between light power densities and photocurrents in the UV region. It also achieved increased relaxation times of 2.22 s (slow) and 0.31 s (fast). Our work hopes to offer an effective method for enhancing the performance of perovskite detectors at the UV wavelength.