Room temperature fabrication and post-annealing treatment of amorphous Ga₂O₃ photodetectors for deep-ultraviolet light detection

Deep-ultraviolet (DUV) photodetectors (PDs) based on wide-bandgap (WBG) semiconductors have received increasing interest due to a broad range of promising applications. ^1–5) Among these materials, Ga₂O₃ is more suitable for DUV light detection (especially monoclinic phase β-Ga₂O₃) due to the proper bandgap (4.5–4.9 eV of Eg), good chemical and thermal stability. ^6–12) According to previous reports, the DUV light detection performance of β-Ga₂O₃ films largely depends on the crystal quality, and good crystal quality requires matched growth substrates and high growth temperature, which limits its practical application. ^13–16) In recent years, amorphous Ga₂O₃ has received increasing attention owing to its low growth temperature, non-strict requirements for lattice matching substrate, and large-area preparation capacity. ^17–19) In an effort to balance the responsivity and response time of Ga₂O₃-based PDs, recent experimental and theoretical studies have revealed a profound role of the oxygen vacancy (V_O) in Ga₂O₃ films on tuning the performance of DUV PDs. ²⁰⁾ For example, Guo et al. ²¹⁾ and Feng et al. ²²⁾ reported that V_O can significantly improve the conductivity of Ga₂O₃ films and the response speed of Ga₂O₃ PDs. Therefore, how to effectively control the V_O concentration is the key factor for obtaining PDs with a balanced performance. Annealing is an effective way to regulate the concentration of V_O, but current work lacks systematic research on annealing effects in different atmospheres.

In this work, we fabricate DUV PDs with a metal-semiconductor-metal (MSM) structure based on radio-frequency sputtered amorphous Ga₂O₃ films at room temperature, which exhibits a good responsivity of 1.77 A W⁻¹ and a fast response time of 114 ms. A series of annealing treatments with different atmospheres have been found effective to reduce the V_O concentration, which exhibits a trade-off effect on the responsivity and response time. These results demonstrate a cost-effective approach for fabricating amorphous Ga₂O₃ DUV PDs and explore potential ways for further performance enhancement for practical device application.

Ga₂O₃ films were deposited on c-sapphire substrates by radio-frequency magnetron sputtering at room temperature. The Ar flow rate, power and work pressure were fixed at 20 sccm, 100 W and 0.5 Pa, respectively. The surface morphology of the as-synthesized Ga₂O₃ film, labeled as S1, is characterized with atomic force microscopy (AFM), which shows a smooth surface with a roughness of about 1 nm [Fig. 1(a)]. The S1 film consists of densely packed grains, suggesting the characteristic 3D island mode. X-ray diffraction (XRD) scan shows no observable diffraction peak from the S1 film, reflecting the amorphous nature [Fig. 1(b)]. The transmittance spectrum is measured by UV–vis spectroscopy to determine its optical bandgap [Fig. 1(c)]. The S1 film exhibits high transmittance of over 80% in the near-ultraviolet to visible region (300–800 nm), followed by a sharp absorption edge as the excitation wavelength extends through the DUV regime (< 300 nm). The bandgap of S1 is obtained from the (αhν)² versus hν diagram in the inset in Fig. 1(c), where α is the absorption coefficient and hν the photon energy. ^23,24) Fitting the linear region of the curve of (αhν)² versus hν determines the S1 optical bandgap to be ∼5.09 eV.

Zoom In Zoom Out Reset image size

To characterize the photodetection response, interdigital Au (50 nm)/Ti (20 nm) electrodes are deposited on the S1 film [Fig. 1(d)]. The length, width, and spacing of the electrode fingers were 1250 μm, 150 μm, and 150 μm, respectively. The S1 shows a low dark current (I_dark) of 1.41 × 10⁻¹¹ A at a bias of 20 V as seen from the current–voltage (I–V) characteristic in Fig. 1(e). Upon illumination with a 254 nm light at 5.45 mW cm⁻², the S1 delivers a photocurrent (I_photo) of 1.45 × 10⁻⁴ A, thus achieving a photo-to-dark current ratio (PDCR) as high as 1.03 × 10⁷ and a responsivity (R) of 1.77 A W⁻¹. Detectivity (D*) as a standard criterion for evaluating the lowest detectable signal. The S1-based PD delivers a high D* of 7.68 × 10¹³ Jones which shows its good ability to detect weak signals. The external quantum efficiency (EQE) represents the ratio of the number of electron-hole pairs collected to the number of incident photons per second, and the S1 has an EQE of 864.67%.

In order to verify the response time of S1, current–time (I–t) characteristics were collected by alternately switching on/off the light source under a fixed bias (20 V). The response time is usually obtained from the fitting resulted based on a double exponential relaxation equation [supplementary data Eq. 1 (available online at stacks.iop.org/APEX/15/022007/mmedia)]. Here the two exponential terms include fast/slow response components, which τ_r1/τ_r2 for 10%–90% of the rising process and τ_d1 /τ_d2 for 90%–10% of the decay process. Generally, the fast response component corresponds to carrier generation or recombination between intrinsic band edges, and the slow response component corresponds to the carrier trap or detrap process by interband defect states. ^15,25,26) The time constants for fast and slow response components of the S1 rising edge are 114 ms (τ_r1) and 2180 ms (τ_r2), respectively. The large τ_r2 value is likely due to the carrier trap process by existing V_O defects in S1, ^27,28) since no oxygen source is provided during the S1 film growth. Compared to the rising edge, the falling edge of S1 is steeper, and the falling time is 37 ms. Supplementary data Table Ⅰ summarizes the key performance of the PD in this work and other Ga₂O₃ PDs. Compared with other PDs, the S1 PD exhibits several competitive device characteristics including low current, large PDCR and fast τ_r1.

Post-annealing has been broadly exploited to modulate the PD performance owing to their strong relation with the film microstructure evolution and defect formation. ^20,29) To fully examine the annealing impact, we adopt a medium-temperature (600 ℃) annealing treatment (2 h) to amorphous films in different atmospheres, including neutral atmosphere (air and N₂), oxidizing atmosphere (O₃) and reducing atmosphere (gas mixture of 95%Ar and 5%H₂ volume ratio). The flow of each atmosphere is 100–120 sccm. Among them, the neutral atmosphere is used as a comparative experiment, oxidizing atmosphere and reducing atmosphere are used to explore the regulation of the concentration of V_O. The annealed films are labeled as S2 for air, S3 for N₂, S4 for O₃ and S5 for Ar/H₂ (Fig. 2(a)). The morphology of the annealed films showed similar close-packed grains and surface roughness of ∼1 nm to the S1 [Figs. 2(b)–(e)], and the amorphous feature as determined from the XRD patterns [Fig. 2(f)]. All the samples exhibit high transmittance of over 80% in the near-ultraviolet to visible region [Fig. 2(g)]. Their optical bandgaps are close to 5.0 eV with small variation within 0.1 eV [Figs. 2(h) and 2(i)].

Zoom In Zoom Out Reset image size

Despite the negligible influence on surface, structure and optical properties, the annealing treatments have an apparent impact on the carrier transport and photoresponse properties of PDs. As seen from the I–V characteristics in Fig. 3(a), the I_dark decreases to a pA range regardless of the annealing atmosphere, suggesting suppression of available free carriers. ²⁸⁾ With the same illumination source, the I_photo of annealed samples decreases slightly within an order of magnitude [Fig. 3(b)]. More detailed device performance has summarized in Table Ⅰ. Despite the reduced responsivity after annealing, the response time for the fast and slow components simultaneously reduces for all annealed samples [Figs. 3(c) and 3(d)]. The decay edges are very sharp for PDs before and after annealing, which is described by a single time constant (τ_d). Figure 3(e) summarizes the τ_r1, τ_r2 and τ_d constants of S1-S5 PDs. It is clearly seen that the O₃-annealed S4 PD exhibits a maximum reduction of τ_r1 and τ_r2 by 55% and 83%, respectively. In addition, the response performance is highly reproducible as seen for the repetition test [Fig. 3(f)].

Zoom In Zoom Out Reset image size

The mechanism of the annealing impact on the device performance is further analyzed from the V_O modulation among Ga₂O₃ films by X-ray photoelectron spectroscopy (XPS). Figures 4(a) and 4(d) show the O1s core level spectra of S1 and S4, respectively. The O1s core level spectra can be fitted with three components by Gaussian fitting. These components are centered at (1) 530.8 ± 0.1 eV (O_Ⅰ), representing O–Ga bonds of Ga₂O₃; (2) 532.2 ± 0.1 eV (O_Ⅱ), characterizing the O²⁻ ions in the oxygen-deficient regions and generally considered to be V_O defects; (3) 533.8 ± 0.1 eV (O_Ⅲ), corresponding to the chemisorbed species on the surface of films. The peak area is used to characterize the relative concentration of each component and the ratio of O_Ⅱ / (O_Ⅰ + O_Ⅱ) can reflect the concentration of V_O. ²⁴⁾ The O_Ⅱ peak of the S2–S5 is suppressed, and the calculated O_Ⅱ / (O_Ⅰ + O_Ⅱ) ratio gradually decreases (supplementary data Fig. 1), indicating that the annealing treatment can effectively reduce the V_O concentration in the films.

Zoom In Zoom Out Reset image size

The energy band diagram of the Ga₂O₃ PD under dark and illumination conditions are presented in Fig. 4 for discussing the role of V_O in the carrier transport process in the PD. From the XPS analysis, S1 has more V_O, which could promote electron transport and interface tunneling (Fig. 4b), so the PD exhibits a higher dark current. Upon illumination, the photocarriers trapped by V_O at the barrier region are extracted to the external circuit that leads to a higher photoconductive gain and a larger R (Fig. 4c). Such a de-trapping process enabled by a large concentration of V_O usually results in persistent photoconductivity (PPC) effect and inevitably makes the response time longer. ^28,30) In contrast, the concentration of V_O is effectively suppressed after annealing, thereby reducing the capture of photogenerated carriers and increasing the junction barrier height of the devices [Figs. 4(e) and 4(f)], which weakens the internal gain and suppresses the PPC effect. Hence, the annealed devices exhibit lower R while faster response speed. Note that we have regulated the V_O concentration in the films by annealing in different atmospheres while the performance of the PDs is not fully optimized, it can be further improved from the aspects of the device structure, crystallinity and annealing temperature in follow-up research. The key function of V_O should be carefully handled when optimizing the device performance, since V_O could provide free carriers to improve the responsivity, and they can also become deep traps for photocarriers so as to increase the response time. In addition, the following work remains to be explored for amorphous Ga₂O₃ photodetectors, such as integration of low-cost substrates, carrier modulation with intentional doping and device structure optimization.

In conclusion, we have demonstrated DUV PDs with an MSM structure based on radio-frequency sputtered amorphous Ga₂O₃ thin films. The Ga₂O₃ PD exhibits a responsivity of 1.77 A W⁻¹ and a fast rise response time of 114 ms. We further explored a series of annealing treatments to modify the performance of Ga₂O₃ PDs. It is found that the annealing treatment effectively reduces the response time at the expense of partial responsivity. Such change is related to the V_O modulation during the post-annealing treatment. While V_O serves as effective donors to provide charge carriers for improving responsivity, they also become trap centers for photogenerated carriers that increase the response time. Our results demonstrate the room-temperature fabrication of high-performance amorphous Ga₂O₃-based DUV PDs and highlight potential ways for balancing the device performance for DUV light detection.

Acknowledgments