Enhancing the Solar-Blind UV Detection Performance of β-Ga₂O₃ Films Through Oxygen Plasma Treatment

Abstract

This study systematically investigated the effects of oxygen plasma treatment on oxygen vacancy defects in sputtered β-gallium oxide (β-Ga₂O₃) films and their corresponding ultraviolet (UV) detection performance. The sputtered β-Ga₂O₃ film subjected to 1 min of oxygen plasma treatment exhibited optimal photodetection properties. Compared to the untreated sample, the dark current was reduced by approximately one order of magnitude to 0.378 pA at 10 V bias. It exhibited an 86% (from 2.92 s to 0.41 s) decrease in response time, a 41.6% increase in photocurrent, a very high photo-to-dark current ratio of 9.18 × 10⁵, and a specific detectivity of 2.62 × 10¹⁰ cm·Hz^1/2W⁻¹ under 254 nm UV illumination intensity of 799 μW/cm² at 10 V bias. Notably, appropriate oxygen plasma treatment minimizes electron capture, enhances the separation and collection of photogenerated carriers, and suppresses the persistent photoconductivity (PPC) effect, thus ultimately shortening the response time. Oxygen plasma processing thus provides an effective approach to fabricating high-performance β-Ga₂O₃ solar-blind photodetectors (SBPDs).

1. Introduction

With the growing demand for high-precision and anti-interference detection technologies in fields such as atmospheric environmental monitoring, public safety early warning, and national defense security, solar-blind ultraviolet (UV) band (200–280 nm) detection has emerged as a research hotspot in the optoelectronic field in recent years, owing to its unique advantage of zero interference from solar radiation [1,2]. Photodetectors operating in the solar-blind UV band (200–280 nm) exhibit low false alarm rates, high sensitivity, and low background noise [3,4], enabling their wide applications in ozone detection, fire alarms, optical communications, and missile monitoring [5,6,7]. Candidate materials for such detectors include MgZnO [8], AlGaN [9], diamond [10], Ga₂O₃ [11], etc. Among these, Ga₂O₃ is particularly promising for high-performance solar-blind photodetectors, owing to its excellent chemical stability, thermal stability, ultrahigh breakdown voltage, and a bandgap that matches the solar-blind region [12]. β-Ga₂O₃ is the most extensively studied crystalline phase for photodetection applications among the five polymorphs of Ga₂O₃, primarily due to its thermodynamic stability [13,14]. With a bandgap aligned to the solar-blind region, β-Ga₂O₃ possesses distinct advantages for solar-blind UV photodetector development. However, heteroepitaxially grown β-Ga₂O₃ films inevitably contain oxygen vacancy defects, which pose a critical bottleneck to the enhancement of photodetector performance.

Radio frequency magnetron sputtering (RFMS) is a well-established technique for β-Ga₂O₃ film deposition, featuring a simple operation process and easy control over film growth attributes that make it suitable for large-scale, cost-effective production [15]. Nevertheless, due to the high cost and preparation challenges of Ga₂O₃ single-crystal substrates, the epitaxial growth of β-Ga₂O₃ films on heterogeneous substrates (e.g., sapphire and silicon wafers) is preferred. Lattice mismatch between β-Ga₂O₃ and heterogeneous substrates can induce local structural distortions in the β-Ga₂O₃ crystal lattice, leading to the incomplete filling of oxygen atoms and the formation of oxygen vacancy defects. These defects act as charge carrier traps, increasing the dark current and reducing carrier mobility; consequently, they cause persistent photoconductivity (PPC) and significantly degrade the response speed of detectors [16].

The essence of the PPC effect lies in the trapping and slow release of photogenerated carriers via deep-level defects in the material (such as oxygen vacancies and impurity levels). The impact of PPC on device performance is highly scenario-dependent; in the field of photodetection where rapid response is paramount, PPC causes the current to fail to drop synchronously after light termination, manifesting as prolonged response times and degraded dynamic switching performance, thereby posing a key bottleneck to device practicality. Conversely, the presence of PPC can enable various alternative applications, such as enhancing detection responsivity and enabling its use in memory devices. Wang et al. demonstrated a feasible strategy to preserve hot electrons for efficient collection by intentionally engineering a mix of cubic zinc-blende and hexagonal wurtzite phases in III–V semiconductor nanowires. This heterostructure design creates additional energy levels that are above the conduction band minimum, which capture and store hot electrons before they thermalize to the band edges. Consequently, the lifetime of photogenerated carriers is prolonged, leading to enhanced responsivity [17]. Yang et al. report a novel negative photoconductivity (NPC) mechanism in n-type indium arsenide nanowires (NWs), revealing competing positive photoconductivity to coexist with NPC. Low temperatures slow post-illumination conductivity recovery, indicating a thermally activated detrapping mechanism. At 78 K, conductance’s spontaneous recovery is quenched, enabling a reversible memory device switchable via light and gate voltage pulses. This NPC-based platform shows promise for high-performance photodetection and low-power nonvolatile memory [18]. This study demonstrates that regulating defect concentration to reduce oxygen vacancies and surface defects can significantly suppress the PPC effect, providing a critical solution for optimizing the performance of β-Ga₂O₃ photodetectors.

Modulating defect concentration is critical for optimizing the performance of semiconductor devices. Current mainstream strategies include the following: (1) suppressing defect generation by optimizing process parameters (e.g., growth temperature and oxygen partial pressure) [19,20]; (2) reducing defect density via high-temperature thermal annealing, which repairs crystal structures and relieves residual stress [21]; and (3) selectively doping impurity atoms to reconstruct electronic structures and regulate defect state distributions [22].

To address the challenge of defect modulation in semiconductor materials, plasma surface treatment technology has been successfully applied in wide-bandgap semiconductor materials, such as ZnO and GaN, owing to its inherent advantages, including low-temperature operation, high efficiency, and precise controllability. Plasma surface treatment has recently emerged as a promising technology for use in semiconductor modification. Liu et al. demonstrated that oxygen plasma treatment reduces oxygen vacancy concentration in ZnO films, yielding ZnO-based photodetectors with response/recovery times below 50 µs and a responsivity of 1–10 A/W [23]. Surface passivation (via plasma or other methods) has also been shown to significantly enhance the performance of Ge-, GaAs-, and GaN-based UV photodetectors; it reduces defect density, suppresses leakage currents, and, thereby, improves photodetection sensitivity [16,24,25]. For Ga₂O₃-based materials, Zhang et al. fabricated β-Ga₂O₃ films via metal–organic chemical vapor deposition (MOCVD); through the combination of oxygen annealing and plasma treatment, these films achieved a dark current of 29 pA, a normalized detectivity of 1.3 × 10¹² Jones, and a response rejection ratio of 8.6 × 10⁶, under a 10 V bias [26]. Existing studies confirm that various plasma treatments can effectively regulate the structural and electrical properties of Ga₂O₃ [27,28]. These studies have consistently verified that plasma treatment enables the repair of lattice defects and the suppression of carrier traps via mechanisms, including ion bombardment and active oxygen supplementation, thereby yielding significant improvements in device performance. Nevertheless, most of the existing literature centers on performance comparisons between “plasma-treated” and “untreated” samples, with insufficient in-depth investigation into “treatment duration”—a critical process parameter. This research gap leads to a lack of quantitative insights for guiding process optimization, particularly in the context of RFMS-grown β-Ga₂O₃ film systems, where the correlation between plasma treatment duration and photodetector performance remains underexplored. The present study is conducting systematic work around this problem.

Building on these findings and addressing the aforementioned challenges, we deposited β-Ga₂O₃ films on sapphire substrates via RFMS and treated them with oxygen plasma to reduce oxygen vacancies. The results reveal substantial reductions in dark current and response time, accompanied by a notable enhancement in the overall performance of the photodetectors.

2. Materials and Methods

2.1. Fabrication of Oxygen Plasma-Treated β-Ga₂O₃ Photodetectors

The fabrication process of oxygen plasma-treated β-Ga₂O₃ photodetectors is illustrated in Figure 1. All processes were conducted following standard semiconductor manufacturing protocols, with key steps detailed as follows.

Figure 1. Preparation process of β-Ga₂O₃ photodetectors with oxygen plasma treatment.

First, substrate and target preparation was performed. Double-side polished c-plane sapphire (0001) was used as the substrate, and a 99.99% pure Ga₂O₃ ceramic disk (60 mm in diameter, 5 mm in thickness) served as the sputtering target. Prior to deposition, the sapphire substrate was ultrasonically cleaned sequentially in acetone, ethanol, and deionized water for 15 min per step, following standard semiconductor cleaning procedures, to remove surface contaminants.

Second, β-Ga₂O₃ film deposition was performed. A magnetron sputtering system (Model MSP-300B, Beijing Chuangshi Weina Technology Co., Ltd., Beijing, China) was utilized for thin-film growth. Prior to deposition, the system was evacuated to a base pressure of 4.5 × 10⁻⁴ Pa. Pre-sputtering was conducted for 5 min to remove surface contaminants from the Ga₂O₃ target, under the following conditions: sputtering pressure of 0.3 Pa, sputtering power of 100 W, argon (Ar) flow rate of 50 sccm, and oxygen (O₂) flow rate of 1 sccm. Subsequently, a 200 nm thick β-Ga₂O₃ film was deposited onto the cleaned sapphire substrate using the same sputtering conditions. To enhance crystallinity, the as-deposited film was then annealed in a high-temperature tube furnace at 1000 °C for 60 min under an oxygen atmosphere.

Third, oxygen plasma treatment was conducted. To investigate the effect of the treatment duration on the photodetector performance, five sample groups were prepared with the following different oxygen plasma exposure times: 0 s (pristine sample), 15 s (Plasma-15 s), 30 s (Plasma-30 s), 1 min (Plasma-1 min), and 5 min (Plasma-5 min). The specific meaning of the samples is shown in Table 1. Plasma treatment was carried out using an inductively coupled plasma cleaning machine (Model CY-P21-300W, Zhengzhou CY Scientific Instruments Co., Ltd., Zhengzhou, China) at a fixed output power of 40 W and a chamber pressure of 30 Pa.

Table 1. Sample name definition.

Definition of Sample Name	Plasma Treatment Time
Pristine	0 s
Plasma-15 s	15 s
Plasma-30 s	30 s
Plasma-1 min	1 min
Plasma-5 min	5 min

Fourth, electrode fabrication and post-annealing was performed. Ti/Au (40 nm/90 nm) interdigital electrodes were deposited on all treated samples via DC magnetron sputtering, using a metal mask to ensure precise patterning. The electrodes had a width of 9000 μm, length of 9000 μm, and a finger spacing of 300 μm, and the effective light-irradiated area of each photodetector was calculated to be 2.19 × 10⁻⁵ m². Finally, all samples were annealed in a muffle furnace at 500 °C for 10 min under ambient air to form reliable ohmic contacts.

2.2. Characterization and Performance Measurement

X-ray diffraction (XRD): Crystalline phase and structural quality were analyzed using a Smartlab XRD system (Rigaku Corporation, Tokyo, Japan) with Cu Kα₁ radiation (λ = 0.15406 nm).

X-ray photoelectron spectroscopy (XPS): The chemical valence states and relative content of oxygen in the films were determined using an ESCALAB 250Xi XPS system (Thermo Scientific, Waltham, MA, USA).

Photoluminescence (PL) spectroscopy: Room-temperature (298 K) PL spectra were recorded using a steady-state/transient fluorescence spectrometer (Model FLS2000, Edinburgh Instruments, Livingston, Scotland, UK) with a 254 nm excitation wavelength.

Photodetector performance testing: The photoresponse characteristics (dark current, photocurrent, and response time) were measured at room temperature and ambient air using a Keithley 4200-SCS semiconductor parameter analyzer (Tektronix, Beaverton, OR, USA) coupled with a UV mercury lamp (254 nm, as the excitation source). The Keshengda Light Shutter-S19CA (Chengdu Keshengda Optoelectronic Technology Co., Ltd. Chengdu, China) is used to control the light-on time and light-off time. Subsequently, the optoelectronic properties of the photodetectors were discussed based on the collected data.

3. Results and Discussion

Figure 2a shows the XRD patterns of Ga₂O₃ films grown on (0001) sapphire substrates. For all five samples, three distinct and invariant diffraction peaks near 2θ = 18.8°, 38.4°, and 59.2°, corresponding to the (−201), (−402), and (−603) crystal planes of β-Ga₂O₃ (JCPDS No. 43-1012), respectively, indicate the excellent crystallographic orientation of these films [29] and signify that the oxygen plasma treatment did not induce any significant alterations to the crystal structures of the Ga₂O₃ thin films.

Figure 2. Performance characterization results of all samples, as follows: (a) XRD patterns, (b) UV-Vis absorption spectra, (c) full XPS spectra, and (d) XPS spectra of Ga2p.

Figure 2b illustrates the transmission spectra of Ga₂O₃ films before and after oxygen plasma treatment. All samples exhibit a maximum absorption peak centered at ~254 nm, accompanied by sharp absorption edges—this spectral feature is consistent with the wavelength requirement for solar-blind UV detection, indicating highly suitability for solar-blind detection [30]. For plasma-treated samples, the average transmittance exceeds 90% in the visible wavelength range of 250–800 nm, reflecting excellent optical transparency. Furthermore, the absorption edge of the films remains unchanged after plasma treatment, with a moderate enhancement in optical transmittance; this improvement is attributed to the surface cleaning and modification induced by oxygen plasma, which eliminates surface contaminants and reduces light absorption. The optical bandgap of the films was calculated using the Tauc plot method, and the results reveal a consistent value of 4.96 eV across all samples, a finding that further confirms that oxygen plasma treatment neither alters the optical bandgap of β-Ga₂O₃ films, nor induces changes in their crystal structure.

To further investigate the elemental composition of β-Ga₂O₃ films before and after oxygen plasma treatment, X-ray photoelectron spectroscopy (XPS) characterization was conducted. All XPS spectra were calibrated using the C 1s peak at a binding energy of 284.8 eV.

Figure 2c presents the full-scan XPS spectra of the five samples (pristine and plasma 15 s–5 min). Distinct peaks corresponding to Ga 3d, Ga 3p, Ga 3s, O 1s, C 1s, Ga 2p, and Ga LMM, as well as O KLL Auger peaks, are observed for all samples. The absence of impurity-related peaks confirms that the films consist exclusively of gallium (Ga) and oxygen (O), consistent with the target composition of β-Ga₂O₃ [31]. Figure 2d displays the high-resolution Ga 2p X-ray photoelectron spectroscopy (XPS) spectra of β-Ga₂O₃ films before and after oxygen plasma treatment, showing two distinct characteristic peaks for all samples, the Ga 2p_1/2 peak centered at ~1145.2 eV and the Ga 2p_3/2 peak at ~1118.3 eV, with a peak separation of ~26.9 eV that confirms the presence of Ga₂O₃ [32]. In the study of samples (Plasma-15 s–Plasma-5 min), XPS analysis showed a redshift in Ga 2p and Ga 3d peaks. Oxygen anions adsorbed to formed chemisorbed oxygen, causing electrons to transfer to Ga, increasing its electron-cloud density and reducing the binding energy.

To identify the elemental oxygen species in Ga₂O₃ films, the O 1s XPS spectra were analyzed and deconvoluted into three peaks via Gaussian fitting (Figure 3). These peaks correspond to lattice oxygen (O_L) at 529.8 eV, vacancy oxygen (O_v) at 530.7 eV, and adsorbed oxygen (O_a) at 532.2 eV. The O_v contents of the five samples were 66.5%, 54.9%, 51%, 46.8%, and 52.1%, while the O_a contents were 17.61%, 19.86%, 21.62%, 28.18%, and 27.68%, respectively. These results show that oxygen plasma treatment effectively modulates oxygen vacancies and recovers the crystalline surface. However, prolonged treatment (Plasma-5 min) disrupts the lattice structure, causing an increase in oxygen vacancies. All data variations observed during the characterization and testing processes are summarized in Table 2, and specific differences in relevant parameters can be referred to in this table.

Figure 3. Fine-sweeping and split-peak fitting of the peaks of Ga₂O₃ thin films (a) pristine film, and (b–e) oxygen plasma-treated samples (Plasma-15 s–Plasma-5 min).

Table 2. Device characterization and performance parameter comparison.

Sample	OA(adsorbed oxygen)	OV(vacancy oxygen)	OL(lattice oxygen)	τr/τd	Peak II/Peak I	D*	R	PDCR
Pristine	17.61%	66.6%	15.89%	2.92 s/1.35 s	3.00	7.63 × 109 cm·Hz1/2W−1	1.4 mA/W	1.06 × 105
Plasma-15 s	19.86%	54.9%	25.24%	1.17 s/0.98 s	1.91	-	-	1.63 × 105
Plasma-30 s	21.62%	51%	27.38%	0.76 s/0.96 s	1.51	-	-	3.4 × 105
Plasma-1 min	28.18%	46.8%	25.02%	0.41 s/0.17 s	1.02	2.62 × 1010 cm·Hz1/2W−1	1.95 mA/W	9.18 × 105
Plasma-5 min	27.68%	52.1%	20.22%	1.18 s/1.57 s	1.41	-	-	4.7 × 105

To elucidate the impact of defects on the optical properties of pristine and oxygen plasma-treated β-Ga₂O₃ thin films, PL spectra of five samples were characterized under 254 nm UV excitation. As shown in Figure 4a–e, all samples exhibited asymmetric ultraviolet–green emission peaks (300–630 nm). After normalization, the peaks were assigned as follows. Peak I represents intrinsic luminescence, as the ultraviolet (UV) luminescence of gallium oxide is caused by self-trapped exciton recombination. Peak II arises from donor-acceptor pair (DAP) recombination involving oxygen vacancies (V_O), gallium vacancies (V_Ga), and their complexes. Here, V_Ga/V_Ga-V_O capture electrons as acceptor levels, while V_O/Gai (gallium interstitials) supply carriers as donor levels, generating a blue emission band via radiative recombination. The origin of the green emission (Peak III) remains unclear, though V Pati et al. [33] propose that it may result from DAP transitions between V_O-related donors and V_Ga or V_Ga + V_O acceptors. The area ratios of Peak II/Peak I for the five samples were 3.00, 1.91, 1.51, 1.02, and 1.41. Sample Plasma-1 min, with the lowest ratio (1.02), indicated the lowest oxygen vacancy concentration, consistent with prior XPS results. This confirms that 1 min oxygen plasma treatment effectively reduces oxygen vacancies in β-Ga₂O₃ thin films.

Figure 4. PL spectrum and split-peak fitting of peaks of Ga₂O₃ thin films (a) pristine film, and (b–e) oxygen plasma-treated samples (Plasma-15 s–Plasma-5 min) (the normalized intensity).

The I-V characteristics of the five β-Ga₂O₃ solar-blind UV photodetectors were measured under dark and illuminated conditions, with the results shown in Figure 5a,b, respectively. The light source utilized for these measurements has a central wavelength of 254 nm and a power density of 799 μW/cm². Combining the results in Figure 3 and Figure 4, it can be concluded that oxygen plasma treatment significantly reduces the dark current. This effect is attributed to reducing oxygen vacancies in β-Ga₂O₃ films and suppressing carrier recombination induced by oxygen-related defects [34]. Notably, sample Plasma-1 min exhibits the lowest oxygen vacancy concentration, consistent with its minimal dark current among all samples. Specifically, at a bias voltage of 10 V, the dark current of sample Plasma-1 min is as low as 0.378 pA—approximately one order of magnitude lower than that of the pristine sample.

Figure 5. Pristine and oxygen-plasma-treated Ga₂O₃ films (a) under UV irradiation as well as PD schematic, and (b) in dark and (c) time-dependent light response I-T curves during periodic turn-on and turn-off of 254 nm UV light at 799 μW/cm² with 10 V bias. (d) Relation between R, D* and light intensity.

Compared to the untreated samples, the photocurrent of samples Plasma-15 s and Plasma-30 s decreased (Figure 5b,c). This trend can be attributed to the fact that energy-sufficient reactive oxygen species dissociate into oxygen anions, which overcome the energy barrier to occupy oxygen vacancy sites—reducing both the oxygen vacancy content and the corresponding carrier density, thereby leading to lower photocurrents [28]. Extending the treatment to 1 min not only further reduces oxygen vacancies concentration but also eliminates surface defects and suppresses the defect-related recombination of photogenerated carriers. Specifically, the sample treated with 1 min plasma (Plasma-1 min) exhibits a significant 41.3% increase in photocurrent. This improvement occurs because the reduction in defects, which include both oxygen vacancies and surface defects, facilitates the transport and collection of photogenerated carriers, decreases the likelihood of carrier recombination, and, thus, enhances the photocurrent. Moreover, the dark current is reduced as the defect-mediated recombination is mitigated by the 1 min oxygen plasma treatment [26]. Excessive removal of oxygen vacancies depletes free electrons from donor defects, thereby disrupting the balance of carrier concentration. This disruption reduces the density of photogenerated carriers and the corresponding photocurrent, ultimately degrading optoelectronic performance. Furthermore, the β-Ga₂O₃ film was subjected to prolonged treatment with oxygen plasma. For instance, the treatment used in the Plasma-5 min sample results in a sudden drop in the photocurrent; on one hand, excess chemical adsorbates act as centers for carrier scattering and trapping, and on the other hand, prolonged ion bombardment damages the crystal lattice. Together, these two factors lead to reduced photocurrent and overall electrical performance.

The β-Ga₂O₃ photodetector treated for 1 min achieved the highest photo-to-dark current ratio (PDCR) of 9.18 × 10⁵. Figure 5c shows the response curves of the photocurrents over time at 10 V bias, demonstrating excellent on/off characteristics and reproducibility.

Responsivity (R) and specific detectivity (D*) are crucial parameters for evaluating the photodetector photoelectric performance [35].

R can be expressed as follows:

$$R={\displaystyle \frac{I_{photo}-I_{dark}}{P_{\mathsf{\lambda }}•S}}$$

(1)

where P_λ is the optical power density of incident UV light and S is the effective light area of the device.

D* is used to describe the ability of the photodetector to detect weak light and can be expressed as follows:

$$D^{*}={\displaystyle \frac{R_{\mathsf{\lambda }}}{\sqrt{\frac{2q{I}_{dark}}{S}}}}$$

(2)

Here, noise measurement was not conducted due to the limitations in experimental conditions, and the bandwidth was set to a default value of ∆f = 1 Hz. q denotes the charge of an electron. The calculations of R and D* at a bias voltage of 10 V are presented in Figure 5d. At a 10 V bias and 799 μW/cm² light intensity, the untreated sample exhibited R = 1.4 mA/W and D* = 7.63 × 10⁹ cm·Hz^1/2W⁻¹, while sample Plasma-1 min showed the significantly enhanced R = 1.95 mA/W and D* = 2.62 × 10¹⁰ cm·Hz^1/2W⁻¹. Most photodetectors exhibit decreasing responsivity as incident light intensity increases. However, in Figure 5d of this manuscript, responsivity (R) and detectivity (D*) show an upward trend with rising light intensity. Through detailed analysis, we propose two possible explanations for this phenomenon. First, the number of photogenerated charge carriers is relatively higher than that of incident photons, coupled with a lower carrier recombination rate. This allows the density of photogenerated carriers to increase continuously with light intensity, ultimately leading to the observed growth of R and D* with light intensity in Figure 5d [36,37]. Second, it may also stem from an insufficient incident light intensity, which keeps the photoresponse in an unsaturated state.

Figure 6a–e presents the response/recovery time test results for the five samples. Response time τ_r is defined as the time for photocurrent to rise from 10% to 90% of the peak value upon light illumination, while recovery time τ_d is the time for photocurrent to decrease from 90% to 10% after light cessation. The response/recovery times for the samples were 2.92 s/1.35 s, 1.17 s/0.98 s, 0.76 s/0.96 s, 0.41 s/0.17 s, and 1.18 s/1.57 s, respectively. Devices without oxygen plasma treatment have higher oxygen vacancy concentrations, which trap electrons and impede carrier migration, resulting in longer response times. After 1 min plasma treatment, τ_r and τ_d are significantly reduced due to fewer oxygen vacancies acting as recombination centers. This minimizes electron capture, enhances photogenerated carrier separation and collection, and suppresses the persistent photoconductivity (PPC) effect, thus shortening response times. Conversely, sample Plasma-5 min shows increased τ_r/τ_d due to prolonged plasma treatment causing excessive surface chemical adsorption. These adsorbates act as scattering and trapping centers, reducing electron mobility and recombination efficiency, leading to longer response times. Therefore, optimizing plasma treatment time is crucial for device performance.

Figure 6. I-T curves of normalized rise and decay processes for (a) pristine and (b–e) plasma-treated Ga₂O₃ films with different treatment times (Plasma-15 s–Plasma-5 min) under 799 μW/cm² illumination at 10 V (pink represents the time of ascent and descent).

To evaluate the long-term operational stability of the β-Ga₂O₃ photodetectors, the Plasma-1 min sample (optimally treated oxygen plasma) was irradiated with UV light for 100+ cycles under a 10 V bias, with each cycle lasting 10 s (total test duration ~1500 s). The corresponding transient photoresponse curves are presented in Figure 7a,b. After over 1500 s of continuous cyclic testing, the Plasma-1 min photodetector maintained a consistent and excellent photocurrent response, with no obvious degradation in signal amplitude—confirming its repeatability and long-term reliability. This result demonstrates that the optimal oxygen plasma treatment (1 min) not only effectively modulates oxygen vacancy defects to enhance photodetection performance but also preserves the photodetector’s long-term operational stability, a critical requirement for its practical application in solar-blind UV detection scenarios.

Figure 7. (a,b) Long-term stability test of the transient optical response characteristics of the device at 10 V bias.

Oxygen vacancies (O_V) are the primary intrinsic defects in Ga₂O₃, which originate from an insufficient oxygen source, excessively high growth temperature, or lattice stress during thin film growth. These factors cause the detachment of partial O atoms from the lattice, forming positively charged vacancies (predominantly O_V²⁺). Oxygen plasma treatment can introduce strongly oxidizing oxygen ions (O²⁻) and highly reactive oxygen radicals (O) [38]. Through oxygen plasma treatment, the following occurs:

Ga₂O₃(O_V²⁺) + 2O + 2e⁻ → Ga₂O₃

The reactive oxygen species overcome the energy barrier to occupy O_V sites, thereby reducing the oxygen vacancy concentration in the film. Short-duration treatments (15 s and 30 s) lead to a decrease in the photocurrent due to the reduced oxygen vacancy content and a corresponding drop in carrier density [28]. Extending the treatment to 1 min not only further reduces oxygen vacancies, but also eliminates surface defects and suppresses the defect-related recombination of photogenerated carriers, resulting in a slight recovery of photocurrent; meanwhile, the dark current is reduced as the defect-mediated recombination is mitigated by the oxygen plasma treatment [26]. However, excessive treatment time (e.g., sample Plasmap-5 min) results in a sudden drop in photocurrent because excessive chemical adsorbates act as carrier scattering/trapping centers, and prolonged ion bombardment damages the crystal lattice, ultimately degrading both the photocurrent and overall electrical performance.

Table 3 presents a comparison of the performances of Ga₂O₃ thin films treated with oxygen plasma via different preparation techniques reported in the literature, to the best of our knowledge. Our work exhibits remarkable advantages in τ_r/τ_d, along with a low dark current and relatively strong detection capability. Furthermore, by employing the RFMS process that is more conducive to industrialization, a favorable balance between performance and production feasibility has been attained.

Table 3. Recent comparison of Ga₂O₃ thin film performances treated with oxygen plasma via different preparation technologies.

Device	Method	Idark	PDCR	τr/τd	D*	R	Ref
β-Ga2O3 film@3 V	MOCVD	0.005 pA	5.7 × 106	0.063 s/0.089 s	3.1 × 1014 cm·Hz1/2W−1	17.7 A/W	[38]
β-Ga2O3 film@10 V	MOCVD	2.9 × 10−2 nA	-	1.17 s/0.98 s	9 × 1015 cm·Hz1/2W−1	38.82 A/W	[26]
β-Ga2O3 film@10 V	RFMS	0.0032 pA	2.7 × 107	0.54 s/0.92 s	-	-	[28]
α-Ga2O3 film@10 V	RFMS	3.6 μA	19.4	0.89 s/1.08 s	3.41 × 1010 cm·Hz1/2W−1	9.63 A/W	[34]
β-Ga2O3 film@10 V	RFMS	0.378 pA	9.18 × 105	0.41 s/0.17 s	2.62 × 1010 cm·Hz1/2W−1	1.95 mA/W	This work

4. Conclusions

In summary, this study demonstrates that appropriate oxygen plasma treatment of sputtered β-Ga₂O₃ films effectively reduces their oxygen vacancy content, which not only weakens the electron capture capability of defects, but also significantly suppresses the persistent photoconductivity (PPC) effect, thereby enhancing the photodetection performance. Specifically, the β-Ga₂O₃ photodetector treated with oxygen plasma for 1 min (the optimal duration) exhibited the following exceptional optoelectronic properties: at a 10 V bias, it achieved an extremely low dark current of 0.378 pA; under 254 nm UV illumination (power intensity: 799 μW/cm²) and 10 V bias, it delivered a high photo-to-dark current ratio of 9.18 × 10⁵, a responsivity of 1.95 mA/W, and an elevated specific detectivity of 2.62 × 10¹⁰ cm·Hz^1/2W⁻¹; and compared to the untreated (pristine) sample, its UV response performance was drastically improved, with the response rise time and recovery time shortened from 2.92 s and 1.35 s, to 0.41 s and 0.17 s, respectively, and the photocurrent intensity increased from 245 nA to 347 nA. Given its efficiency, low cost, and compatibility with β-Ga₂O₃ film fabrication processes, oxygen plasma treatment emerges as a highly promising strategy for advancing the performance of β-Ga₂O₃ solar-blind UV photodetectors, with significant potential for practical applications in fields such as ozone monitoring and missile warning. Given that the oxygen plasma treatment technique with an appropriate processing duration not only combines the advantages of having a high efficiency and low cost, but it also exhibits excellent compatibility with the magnetron sputtering-based β-Ga₂O₃ thin film fabrication process, which enables industrial-scale and large-volume production. Thus, this technique has emerged as a highly promising strategy for enhancing the performance of β-Ga₂O₃ SBPDs. Particularly in fields such as ozone monitoring and missile early warning systems, this method further demonstrates significant potential in practical applications.