1. Introduction

Ultraviolet (UV) photodetectors have received extensive attention for applications in flame sensors, radiation detectors, and analysis in chemical, environmental, and biological fields [1–6]. In particular, deep-UV photodetectors with solar blindness, defined as having a cut-off wavelength of 280 nm, are preferred for applications in fire detection and military surveillance. Several research groups have studied the development of novel wide-bandgap materials for solar-blind photodetectors with high responsivity, selectivity, and stability. Si- and GaAs-based deep-UV photodetectors are not truly solar-blind, as additional visible-light blocking filters are required due to the narrow bandgaps of the semiconducting materials. Therefore, commercially available solar-blind optical devices are bulky, fragile, and only operational under large bias conditions [1,2]. These limitations have triggered research into the development of alternate wide-bandgap materials to replace Si and GaAs. Promising alternative wide-bandgap materials include SiC, diamond, and III-nitrides, especially Al_xGa_1-xN, ZnO, and β-Ga₂O₃ [1,2]. Among these, Al_xGa_1-xN is suitable for solar-blind photodetectors because of the direct and wide bandgap, which ranges from 3.4 to 6.2 eV as the concentration of Al increases. However, Al_xGa_1-xN-based devices have a persistent photoconductivity effect, a light-induced enhancement in the conductivity that persists for a long period after the termination of the light source [2]. Additionally, the growth of Al_xGa_1-xN with a high concentration of Al is quite challenging [7,8]. ZnO also has a direct and wide bandgap of ~3.4 eV, which makes it suitable for photodetection in the UV-A region of 320 nm to 400 nm. MgZnO alloy which has a tunable bandgap (3.4-7.8 eV) also has been investigated [9,10].

Monoclinic β-Ga₂O₃ is one of the most promising candidates for use in practical solar-blind photodetectors, having a wide bandgap of ~4.9 eV, good chemical and thermal stability, high optical transparency in both UV and visible regions, and high thermal conductivity [11,12]. Several types of β-Ga₂O₃-based UV photodetectors, including Schottky photodiodes, metal–semiconductor–metal (MSM) photodiodes, and photoconductors, using various growth techniques to synthesize high-quality β-Ga₂O₃, have been reported [13–18]. Li et al. demonstrated the β-Ga₂O₃ nanowire-based deep-UV photodetectors with MSM structures. The devices were fabricated by a single-step chemical vapor deposition (CVD) process on Au-patterned Al₂O₃ [13]. Yu et al. reported the growth of β-Ga₂O₃ films by pulsed laser deposition method and the detection performance of the fabricated devices with MSM structures [14]. Guo et al. showed the growth of β-Ga₂O₃ thin films by laser-assisted molecular beam epitaxy; the β-Ga₂O₃ thin-film-based solar-blind photodetectors with MSM structures had good photoconductive behaviors [15]. Nakagomi et al. reported the performance of deep-UV photodiodes based on a heterojunction of n-type β-Ga₂O₃/p-type SiC [16].

Herein, we report the fabrication of solar-blind photodetectors with a planar photoconductor structure from metalorganic CVD (MOCVD)-grown β-Ga₂O₃ thin films. The photoconductor structure has several advantages over other structures, including the ease of the fabrication process and high photoconductive gain [1,2]. A particularly important aspect is the formation of a low-resistance Ohmic contact between the β-Ga₂O₃ film and the contact electrode, as the resistivity of this interface controls the flow of photo-induced carriers, and consequentially affects various photoconductive performance metrics such as spectral selectivity and responsivity [19]. Several techniques to improve the Ohmic behavior between β-Ga₂O₃ and metal electrodes have been reported previously. Sasaki et al. reported on Si-ion implantation doping to fabricate Ohmic electrodes with low contact resistance [20]. Kim et al. demonstrated that the electrical properties of the layers composed of un-doped Ga₂O₃ nanoparticles were improved by the introduction of single-walled carbon nanotubes [21]. Orita et al. obtained deep-UV transparent conductive films by the addition of Sn to the Ga₂O₃ target using pulsed laser deposition [22]. Takakura et al. reported on the effect of Si doping on the conductivity of β-Ga₂O₃ films fabricated by co-sputtering [23]. In this study, we deposited a β-Ga₂O₃ layer on an Al₂O₃ substrate by MOCVD technique. Subsequently Si ions were implanted into the film to improve the contact resistance between the β-Ga₂O₃ films and the electrodes. Solar-blind deep-UV photodetectors, based on the Si-doped β-Ga₂O₃ films and using a planar photoconductor structure, were then fabricated and characterized.

2. Experimental details

Figure 1 shows the device fabrication process, including the growth of β-Ga₂O₃ films, Si-ion implantation, activation annealing, and device fabrication. MOCVD was conducted in an impinging-flow cold-wall radio frequency-heated system. The c-plane Al₂O₃ substrates were heated under flowing H₂ at 1200°C to remove residual contaminates on the substrate surfaces. The chamber temperature was reduced to 650°C and held under a N₂ purge for 30 min. Subsequently, growth was conducted for 2 h at 650°C and 50 Torr with N₂ push gas and trimethylgallium as the Ga source. The crystal structure and quality of the β-Ga₂O₃ films were investigated by X-ray diffraction (XRD) and micro-Raman spectroscopy. The XRD (ATX-G, D/max 2500-PC, Rigaku, Japan) spectra were obtained in the 2θ mode using Cu K_α radiation (λ = 1.5406 Å). The Raman spectrum was recorded with a Horiba Jobin Yvon LabRam Aramis system in backscattering geometry. The 514.5 nm line of an Ar-ion laser was used as the excitation source, with a beam spot size of approximately 1 μm in diameter. The laser power on the sample was approximately 1 mW, and the spectrum was accumulated for 120 s.

 

Fig. 1 Schematic of the fabrication process: (a) β-Ga₂O₃ films grown by MOCVD technique on Al₂O₃, (b) Si-ion implantation process, (c) the fabricated photodetector device, and (d) Si-ion depth profile in β-Ga₂O₃ films by SRIM simulation.

Download Full Size | PDF

Si ions were implanted at 30 keV energy with a dose of 1 × 10¹⁵ cm^–2. A simulation using the Stopping and Range of Ions in Matter (SRIM) software package was used to obtain a penetration depth profile. Post-implantation annealing to activate the implanted Si atoms was performed under Ar ambient at 900°C by rapid thermal annealing (RTA, MILA3000-P-N, ULVAC-RIKO). After the RTA, the sheet resistances (R_S) of the samples were measured at more than ten positions per sample using a four-point probe setup (Desk 205, MS Tech) connected to a source meter (Keithley 2400). The contact structure was fabricated using conventional photolithography by depositing Ti/Au (20 nm/80 nm) on the Si-implanted β-Ga₂O₃/Al₂O₃ substrate. A schematic of the fabricated photodetectors is shown in Fig. 1(c). The Ti/Au metallizations were deposited by electron-beam evaporation technique. The current–voltage (I–V) characteristics and time-dependent photoresponse of the β-Ga₂O₃-based photodetector were obtained under vacuum condition (2 × 10⁻² Torr) using an Agilent 4155C semiconductor parameter analyzer connected to the probe station. The time-dependent photoresponse measurement was performed at a constant voltage of 5 V. A 15 W UV lamp (UVItec LTD.) with emission wavelengths of 254 and 365 nm was fixed at a distance of ~9 cm from the fabricated photodetectors. A laser power meter (FieldMaxII-TO, Coherent Inc.) was used to obtain the intensity of UV lamp.

3. Results and discussion

Figure 2(a) shows the XRD spectrum of the Si-doped β-Ga₂O₃ photoconductor devices with Ti/Au electrodes. The obtained peaks include information from β-Ga₂O₃, the Al₂O₃ substrate, and the electrodes. Five peaks located at 19.1, 30.2, 48.7, 59.3, and 62.8°, corresponding to the ($\overline{2}$01), (110), (510), ($\overline{6}$03), and (710) planes, respectively, are observed, which indicate that the prepared films consist primarily of monoclinic β-Ga₂O₃, coinciding with JCPDS card #04-002-2603. Figure 2(b) shows the micro-Raman spectrum of the Si-doped β-Ga₂O₃ film grown on Al₂O₃. The three peaks located near 200 cm^–1 correlate to the low-frequency vibration and translation of tetrahedron–octahedron chains within the β-Ga₂O₃ structure. The peaks located in the range from 300 to 500 cm⁻¹ are attributed to the mid-frequency deformation of Ga₂O₆ octahedra; the ~765 cm⁻¹ peak is attributed to the high-frequency stretching and bending of GaO₄ tetrahedra [24,25]. Several peaks related to β-Ga₂O₃ do not appear because they overlap with some high-intensity peaks of Al₂O₃ [26]. The Raman spectrum confirms the high quality of the Ga₂O₃ films grown by the MOCVD process.

 

Fig. 2 (a) X-ray diffraction results and (b) Raman spectrum of Si-implanted β-Ga₂O₃ films grown on Al₂O₃ substrate. (c) Sheet resistance (R_S) of Si-implanted β-Ga₂O₃ films after the RTA process and (d) optical microscope image of the fabricated device.

Download Full Size | PDF

To enhance the electrical conductivity of β-Ga₂O₃ films, many techniques such as co-sputtering, impurity doping, and the incorporation of conductive materials have been attempted, because β-Ga₂O₃ is intrinsically highly resistive [20,21,27,28]. Among these, ion implantation has many advantages, such as the precise control of doping concentrations and high reproducibility. Therefore, the Si-ion implantation method was employed to improve the electrical conductivity of the β-Ga₂O₃ films and eventually to form an Ohmic contact with the Ti/Au electrode. Figure 1(d) shows the SRIM simulated depth profile of Si ions implanted at an energy of 30 keV with a dose of 1 × 10¹⁵ cm^–2. The projected range is approximately 25 nm. The profile can be affected by the activation annealing process due to the thermal diffusion of Si atoms. The as-implanted β-Ga₂O₃ layer was highly resistive, although a precise number is not available as the resistance was above the detection limit of the available instruments. After the first RTA process (30 s at 900°C), the R_S of the implanted films is measured at 206.7 kΩ/□ as shown in Fig. 2(c). After extended RTA, the R_S of implanted films is reduced to 164.1 kΩ/□. After activation annealing at 900 °C, the electrical activation efficiency of Si atoms implanted into Ga₂O₃ films was reported to be approximately 75% [20]. The decrease in R_S is attributed to the increase of the effective donor concentration through the Si-ion implantation and subsequent activation annealing process. After RTA for 120 s at 900°C, the R_S of the Si-implanted film increases to 199.2 kΩ/□, which can be explained by the creation of compensating defects in the oxide or the decreased number of oxygen vacancy sites [29].

In order to investigate the UV photoresponse of the Si-doped β-Ga₂O₃ films, photoconductor-type solar-blind photodetectors were fabricated. An optical microscope image of the fabricated device is shown in Fig. 2(d). The UV lamp emitting at wavelengths of 254 nm and 365 nm was used as the light source. Figure 3(a) shows the I–V characteristics under dark conditions and UV illuminations of 254 nm and 365 nm wavelengths. We can observe that an Ohmic contact was formed based on the linear I–V characteristics. Under UV light exposure, the increase in the photocurrent can be explained by the band-to-band transition and a transition between a band and impurity (or defect) levels. Under 254 nm deep-UV irradiation, the photocurrents are 218 nA at 5 V. The photocurrent measured under exposure to 365 nm light increases slightly in comparison with the dark current, indicating that the prepared β-Ga₂O₃ films have high spectral selectivity between the UV-A and UV-C spectral regions. The defect levels within the bandgap can explain the slight increase of the photocurrents under 365 nm illumination. The ratio of the photocurrents under 254 nm and 365 nm illumination is 9.43, as shown in Fig. 3(b), where the photocurrents at 150 s are 206 and 21.9 nA at the wavelengths of 254 and 365 nm, respectively.

 

Fig. 3 (a) I–V characteristics and (b) time-dependent photoresponse of the Ga₂O₃ solar-blind photodetectors under dark conditions and illumination with 254 nm and 365 nm UV light. Experimental data and fitted curve of the rise and decay process in response to (c) 254 nm and (d) 365 nm illumination, respectively.

Download Full Size | PDF

Figure 3(b) shows the time-dependent photoresponse of the fabricated photodetector device to 254 nm and 365 nm illumination at an applied bias of 5 V with a weak persistent photoconductivity. For a more detailed investigation, the rise and decay time constant were analyzed by using the following equation: I = I₀ + A·exp(−t/τ₁) + B·exp(−t/τ₂), where I₀ is the steady-state photocurrent, t is the time, A and B are constants, and τ is the relaxation time constant. τ_r and τ_d denote the rise and decay time constants, respectively. Figures 3(c) and 3(d) show that the photoresponse processes are well fitted with the above equation. When the device is exposed to 254 nm light, the rise time constants (τ_r1, τ_r2) and the decay time constants (τ_d1, τ_d2) are estimated to be 0.58 s, 32.93 s and 1.2 s, 32.86 s, respectively. The response to 254 nm photons has two components (fast and slow), which are attributed to band-to-band transition and band-to-deep-level transition. In the case of 365 nm light, τ_r and τ_d are estimated to be 18.35 s and 42 s, respectively (Fig. 3(d)). The slow response and decay processes can be attributed to the deep levels within the energy bandgap. The responsivity at 254 nm was 1.45 A/W, which is comparable to the reported values (4.3 ~17 A/W) for β-Ga₂O₃ photodetectors [30–33]. Higher responsivity was demonstrated by using a carrier multiplication process or a high resistive cap layer [32,33]. The 254-to-365 nm rejection ratio was ~10.8, indicating the fabricated photodetector has the high spectral selectivity. We believe that photodetector devices using MOCVD-grown β-Ga₂O₃ thin films can be further improved by minimizing the defects in β-Ga₂O₃ and optimizing the fabrication process, demonstrating the potential for the development of high-performance solar-blind photodetectors.

4. Conclusion

We demonstrated a MOCVD-grown β-Ga₂O₃ film-based solar-blind photodetector with a planar photoconductor structure, and investigated the photoresponse properties of the fabricated devices. The R_S of the β-Ga₂O₃ films was reduced by Si-ion implantation and RTA activation, which provided a good Ohmic contact between the metal electrodes and β-Ga₂O₃ thin films. The Si-implanted β-Ga₂O₃ thin-film photodetectors showed a fast photoresponse and decay, as well as reproducible responsivity and high spectral selectivity.