1. Introduction

Solar-blind photodetectors have a growing number of potential applications in the fields of military and civil application, such as missile tracking, fire detection, short-range communications security, chemical/biological analysis, and so on [1–3]. Recently, β-Ga₂O₃ has attracted tremendous interests as a promising candidate for solar-blind photodetectors because of its wide bandgap (E_G), the excellent chemical and thermal stability and the high transparency at the wavelength λ longer than 300 nm [4–9]. The β-Ga₂O₃ thin films can be grown by pulsed laser deposition (PLD) [10,11], metalorganic chemical vapor deposition (MOCVD) [12], molecular beam epitaxy (MBE) [13] and magnetron sputtering [14]. So far, the experimental results have indicated that the point defects could act as trapping and recombination centers reducing the collection efficiency of the photogenerated carriers and the photocurrent I_photo [15]. However, thermal annealing treatment is an effective technique to improve the structural and the optical properties of films [16,17]. Therefore, it is necessary to investigate the effects of post annealing on the properties of β-Ga₂O₃ epilayers and the performance of solar-blind photodetectors.

In this paper, we fabricated metal-semiconductor-metal (MSM) photodetectors using the as-grown β-Ga₂O₃ film and the epilayers annealed in various atmosphere. The effects of post annealing atmosphere on the structural and optical properties of β-Ga₂O₃ films and the photodetectors performance are investigated.

2. Experiments

The β-Ga₂O₃ thin films with $(\overline{2}01)$ orientation were epitaxially deposited on sapphire (0001) substrate at 600 °C utilizing laser molecular beam epitaxy (LMBE) with the oxygen pressure of 0.01 mbar. The base pressure of the growth chamber was 1.5 × 10⁻⁸ Torr. The cylindrical Ga₂O₃ ceramic target (99.99% purity) was irradiated by the KrF excimer laser beam with the laser energy density of 2 J/cm² at the repetition rate of 5 Hz. The distance between the sapphire substrate and Ga₂O₃ target was fixed at 5 cm. The Ga₂O₃ films of 132 nm were grown with 8000 pulses. Then, the Ga₂O₃ epilayers were annealed at different temperature for various durations in air and oxygen atmosphere (details shown in Table 1). In the comparison with other annealing condition, samples processed under 800°C for 30 min in air and oxygen atmosphere show the best behavior. The photodetectors with interdigital Ti/Au (20 nm/100 nm) contacts were fabricated on the as-grown, air-annealed and oxygen-annealed Ga₂O₃ films. The electrode fingers were 10 μm wide and 2000 μm long with a 10 μm spacing gap. Figure 1 shows the schematic diagram of the fabricated device.

Table 1. Comparison of interplanar spacing obtained from XRD θ-2θ curves and FWHM obtained from XRD rocking curves.

View Table

 

Fig. 1 Schematic diagram of the β-Ga₂O₃ MSM structure photodetector.

Download Full Size | PDF

The crystal structure of Ga₂O₃ samples was studied by the high-resolution X-ray diffraction (HRXRD) using Cu Kα radiation with a wavelength of 1.5406 Å. Transmittance measurements were taken for a spectral range between 200 and 1000 nm in increments of 1 nm using the Lambda 900 double-beam UV-vis-NIR spectrophotometer. The stoichiometries of Ga₂O₃ films were evaluated by X-ray photoelectron spectroscopy (XPS). Characterization of the dark current I_dark, I_photo, responsivity R and time-dependent photoresponse of the Ga₂O₃-based photodetectors were carried out utilizing a low-pressure mercury lamp U3900 with various power densities P_light and the illumination wavelength λ from 200 nm to 365 nm.

3. Results and discussion

3.1 Material properties of β-Ga₂O₃ films

Figure 2(a) presents the typical X-ray diffraction (XRD) θ-2θ scans with only (01), (02) and (03) diffraction peaks observed for all the samples, demonstrating the single crystallinity of β-Ga₂O₃ phase. Figure 2(b) depicts the (02) HRXRD ω-scan rocking curves (RC), normalized for comparison. According to the XRD results(Fig. 2(a)), peaks of samples annealed under 800 and 900°C for 60 and 30 min, respectively, are close to the standard XRD peaks position, but the high-temperature and long-time annealing also destroy the crystal quality according to the FWHM of (02) peak with the value of 0.930° and 0.973°. On the contrary, samples processed at 800°C for 30 min in air and oxygen atmosphere show a better crystallinity with the RC FWHM of 0.770° and 0.733° in the comparison with the as-grown sample, which is also confirmed by TEM results (shown in Fig. 2(c)). Furthermore, samples annealed at 700 and 800°C for 30 and 15 min, respectively, have the FWHM similar to the as-grown sample. From the discussion above, we choose the samples annealed at 800°C for 30 min in air and oxygen atmosphere (which are called air annealed sample and O₂ annealed sample in the following) to conduct further investigation with the as-grown sample.

 

Fig. 2 (a) XRD curves for all Ga₂O₃ samples, (b) Rocking curve of all Ga₂O₃ samples, (c) TEM image of as-grown sample and samples annealed under 800°C for 30 min in air and oxygen atmosphere, respectively.

Download Full Size | PDF

The vertical (01), (02) and (03) interplanar spacing of all the Ga₂O₃ samples are calculated and summarized in Table 1 on the basis of Eq. (1). As annealing temperature and duration increase, the interplanar spacing gradually approach to the standard value and the results obtained from each plane are in good agreement with each other. So, there is an out-of-plane tensile strain in all the epitaxially grown Ga₂O₃ films on sapphire, especially the as-grown sample. The reason for the as-grown Ga₂O₃ film under an obvious out-of-plane tensile strain (in-plane compressive strain) [18] is believed to be lattice mismatch [19] and thermal expansion mismatch [20,21] between the sapphire substrate and Ga₂O₃ epilayer, while the smaller compressive strain in the other samples can be attributed to the strain relief during annealing at high temperature [22], and it can be confirmed by the Raman measurement results.

Figure 3 shows the Raman measurement results of the as-grown, air annealed and O₂ annealed sample. The peaks from the (−201) β-Ga₂O₃ bulk was presented as the reference, including 114 cm⁻¹, 144 cm⁻¹, 169 cm⁻¹, 200 cm⁻¹, 320 cm⁻¹, 346 cm⁻¹, 416 cm⁻¹, 475 cm⁻¹, 629 cm⁻¹ and 653 cm⁻¹ [23,24]. While the peak at 416 cm⁻¹ originates from the sapphire substrate and β-Ga₂O₃ epilayer, Taken the intensity into account, the shift of the peak 200 cm⁻¹ was used to characterize the strain in the epilayer, as shown in the inset of Fig. 3.There is a blue-shift relative to the reference for all the three samples, about 2.74 cm⁻¹, 1.24 cm⁻¹ and 0.54 cm⁻¹ for the as-grown, air annealed and O₂ annealed sample, respectively. In other words, the compressive strain was partially released after high temperature annealing [25], which is consistent with the XRD measurement results.

 

Fig. 3 Raman spectra of the as-grown, air annealed, O₂ annealed and Ga₂O₃ bulk sample.

Download Full Size | PDF

The transmittance spectra of Ga₂O₃ samples are shown in Fig. 4. A red-shift of absorption edge is observed in annealed epilayers compared to the as-grown sample. The inset shows the relationship between (αhν)² and hν, where hν is the energy of the incident photon and α is the absorption coefficient, which is evaluated using $\alpha =\ln (100/T)/d$, where T is the transmittance and d is the film thickness [26]. By extrapolating the linear part of (αhν)²-hν curves to (αhν)² = 0, bandgaps are obtained to be 4.93, 4.84 and 4.81 eV for the as-grown, air-annealed and oxygen-annealed samples, respectively. The bandgap differences can be ascribed to the in-plane compressive strain relaxation during thermal annealing, which is consistent with the results of XRD.

 

Fig. 4 (a) Transmittance spectra and αhν versus hν of Ga₂O₃ samples.

Download Full Size | PDF

Figure 5 depicts the XPS spectra of Ga 3d and O 1s peaks. Before XPS measurement, a few nanometers of surface were removed using in situ plasma etching, and the peaks were calibrated using the adventitious C 1s of 284.6 eV. On the basis of Ga 3d and O 1s XPS peaks area and the corresponding sensitivity factors, the atomic ratios of O to Ga of the three samples, as-grown, air and oxygen- annealed epilayers, were estimated to be 50.96:49.04, 52.38:47.62 and 53.83:46.17, respectively. Furthermore, the O 1s peaks were fitted using two Gaussian peaks, corresponding to GaO_x and Ga₂O₃ phase. After high temperature annealing in air and O₂ atmosphere, the intensities and areas of GaO_x phase are reduced and it may be converted into Ga₂O₃ phase, indicating the decrease of oxygen deficiency in Ga₂O₃ films and also the oxygen vacancy (V_O) concentration.

 

Fig. 5 XPS Ga 3d and O 1s spectra of Ga₂O₃ samples. Ga 3d raw data were fitted using two Gaussian peaks, corresponding to O 2s and Ga 3d, respectively. O 1s raw data were also fitted with two Gaussian peaks, corresponding to GaO_x and Ga₂O₃, respectively.

Download Full Size | PDF

3.2. Electrical characteristics of Ga₂O₃ photodetectors

Figure 6 shows the characteristics of I_dark and I_photo versus the bias voltage V_bias at incident P_light of 300 and 600 μW/cm² for the three photodetector samples. It can be seen clearly that the I_dark and I_photo under 254 nm normal incidence illumination increase linearly with increasing V_bias. What’s more, larger I_photo and lower I_dark is obtained in the oxygen-annealed device, which is probably due to the fact that the high temperature annealing in oxygen atmosphere is more beneficial for the recrystallization of Ga₂O₃ epilayer, resulting in a reduction of V_O. Although there is no accurate report on the energy level of oxygen vacancy defect, but according to the Reference [27], it may be approximately (0.7 ± 0.2) eV below the bottom of conduction band and the Fermi level was pinned close to the oxygen vacancies defect level. After the metal deposition, a few atomic layers thick depletion region is formed and the electrons are susceptible to tunneling, which leads to an Ohmic contact. After the air and O₂ annealing at 800°C for 30 min, the decrease of the V_O concentration results in the dark current reduction. The superlinear I_photo is observed in the I_photo- P_light curves for the photodetectors under the various V_bias of 10, 20 and 30 V, as presented in Fig. 7, illustrating an optical gain in all three samples.

 

Fig. 6 Comparison of (a) dark currents and (b) photocurrents for three photodetector samples.

Download Full Size | PDF

 

Fig. 7 The linear photocurrent I_photo versus P_light at the various bias voltage V_bias for devices.

Download Full Size | PDF

The time-dependent photoresponse characteristics of the photodetectors were measured by a 254 nm illumination square-wave light under a V_bias of 10 V and a P_light of 600 μW/cm², as shown in After multiple illumination cycles, the devices still achieve the stable on-state photo current I_on, indicating that the photodetectors have high robustness and good reproducibility. The rise and decay processes consist of two components: fast response and slow response. Generally, the fast-response, τ_r1 and τ_d1, is associated with the interband optical transition, while the slow-response component is caused by the carrier trapping/releasing owing to the existence of oxygen vacancies defects in β-Ga₂O₃ thin films. As τ_d2 > τ_r2, when the illumination is OFF, besides the recombination of the carriers through the defect bands, some captured by oxygen vacancies might be released and recombine, which leads to a more pronounced persistent photoconductivity (PPC). In order to further analyze the response time, a quantitative analysis of the current rise and decay processes involves the fitting of the photoresponse curve with a biexponential relaxation equation of the following type [28]

 

Fig. 8 (a) Time-dependent photoresponse of the Ga₂O₃ photodetectors under a V_bias of 10 V; (b), (c), (d) Experimental curves and fitted curves of the rise and decay processes of the photodetectors.

Download Full Size | PDF

Figure 9 depicts the responsivity characteristics versus the illumination optical λ under V_bias of 10 and 20 V. It is obviously noted that the oxygen-annealed Ga₂O₃ photodetector achieves an improved responsivity over the other two samples. The reason for this lies in much more V_O in the as-grown and air-annealed Ga₂O₃ films than in oxygen-annealed Ga₂O₃ sample and V_O can trap the carriers or act as the recombination centers for photogenerated electron-hole pairs, causing the reduction of device responsivity and I_photo. The values of cutoff λ defined as λ at $\sqrt{1/2}$R_max are about 252 nm, 256 nm, and 259 nm, indicating that the direct E_G of as-grown, air-annealed and oxygen-annealed device is 4.92 eV, 4.84 eV and 4.80 eV, respectively. The cut off λ of air-annealed and oxygen-annealed Ga₂O₃ devices shows a significant redshift compared with the as-grown sample, in excellent agreement with the bandgap results in Fig. 4.

 

Fig. 9 Responsivity as a function of λ for the Ga₂O₃ photodetectors at various V_bias.

Download Full Size | PDF

4. Conclusion

β-Ga₂O₃ films were deposited on sapphire (0001) substrate using LMBE and post-annealing was carried out in air and oxygen atmosphere, respectively. The XRD peaks shift to high-angle side, indicating the in-plane compressive strain relaxation in β-Ga₂O₃ epilayers after thermal annealing. In addition, the MSM structure photodetectors were fabricated from the β-Ga₂O₃ thin films. According to the responsivity results, the E_g decreased from 4.92 eV, to 4.84 eV and 4.80 eV for the air-annealed and oxygen-annealed sample, which is consistent with the absorption spectra. Moreover, the enhanced photocurrent, responsivity and time-dependent photoresponse in oxygen-annealed photodetector were also obtained due to the reduction of oxygen vacancies during annealing in oxygen atmosphere.