The term solar-blind ultraviolet (UV) is due to the radiation in this region (220 nm∼280 nm) being almost completely absorbed by the ozone layer and thus the background radiation in the atmosphere is close to zero. The black background makes the solar-blind UV photodetector (PD) have low noise and high sensitivity simultaneously. Owing to the huge application prospects in civil and military areas, such as flame monitoring, space communication, biological medicine, and missile guidance, solar-blind UV PDs have been attracting extensive attention.^[1,2] The high photon energy of solar-blind UV light enables the wide band gap semiconductors to serve as detection materials, including diamond,^[3,4] AlGaN,^[5,6] MgZnO,^[7–9] Ga₂O₃, etc.^[10–17] Compared with other wide band gap materials, Ga₂O₃ has a direct band gap (∼4.9 eV) and a high absorption coefficient (> 10⁵ cm⁻¹ near band edge) in the solar-blind UV spectral region, which avoids the complex band gap engineering problem and makes Ga₂O₃ a particularly suitable candidate for solar-blind UV photodetection.^[18,19] To date, bulk,^[11,12] thin-film,^[13–15] and nanostructure^[16,17] Ga₂O₃ solar-blind UV PDs have been proposed. The bulk and nanostructured Ga₂O₃ PDs usually exhibit superior sensitivity and photoresponsivity to the solar-blind UV light in contrast to their thin-film counterpart.^[11–17] However, the complex synthesis technologies and high cost of the bulk Ga₂O₃ crystals and the low repeatability of Ga₂O₃ nanostructures greatly restrict their practical applications. Therefore, thin-film-type PDs are still the main research focus. On the other hand, the reported thin-film-type PDs, which have good photoresponse performances (> 1 A/W), are based on high-quality Ga₂O₃ single crystalline films.^[14,15] The growth methods for these high-quality thin-films include molecular beam epitaxy, metal-organic chemical vapor deposition, and pulsed laser deposition, in which usually there are the needs of high-vacuum instruments, high growth temperatures, and complicated processes.^[14,15,20] The expensive costs and rigorous conditions required by these synthesis techniques significantly restrict the industrial development of large-area films. Obviously, the current Ga₂O₃ PDs are far away from the industrialization demands, and it is urgently needed to explore a cost-effective and practical method to acquire a unique Ga₂O₃ material with improved solar-blind photoresponse performance.

Surface plasmon (SP) resonance is the phenomenon of collective electron oscillations on the surfaces of metal nanoparticles driven by incident light.^[21,22] Under resonance state, the local electromagnetic field around the metal nanoparticles can be greatly enhanced,^[23] and the scattering cross section of incident light increases at the same time, which greatly improve the interaction between the light and the material.^[24] This principle ensures the SP effect, which is widely used in the fields of optoelectronics, such as detectors,^[25–28] solar cells,^[24] photocatalysts,^[29] and Raman detection.^[30] It is well established that the resonance peak of gallium (Ga) based plasmonic platform is adjustable from the visible to solar-blind UV spectral range.^[31–33] Recently, we have achieved a Ga/Ga₂O₃ nanocomposite film through post-annealing the Ga₂O₃/Ga/Ga₂O₃ laminated film, and the corresponding solar-blind PD has remarkably enhanced responsivity due to the SP coupling effect, which will be published elsewhere.^[34] It should be noted that the thickness of the Ga interlayer plays an important role in the photoresponse properties of Ga/Ga₂O₃ nanocomposite PDs, and the dependence of the PD performance on this key parameter has to be explored in detail.

Here in this work, we present a series of Ga/Ga₂O₃ nanocomposite solar-blind PDs with adjustable photoresponse enhancement factors caused by the SP coupling effect. The Ga/Ga₂O₃ nanocomposite films are obtained by annealing the Ga₂O₃/Ga/Ga₂O₃ triple-layer films in a tube furnace. By changing the thickness of the pre-buried Ga interlayer, the enhancement factor of the responsivity can be increased by about 4.5–220 times. The detector achieves the best solar-blind UV photoresponse performance when the interlayer Ga is 20-nm thick. The low dark current of 8.52 pA at 10-V bias along with the high peak responsivity of 2.85 A/W at 15-V bias indicates that the SP coupled PD is likely to have prospects of practical applications.

Figures 1(a)–1(f) show the AFM images of S0–S4. For S0 sample without Ga interlayer, the surface is relatively smooth, with a root mean square (RMS) roughness value of 2.5 nm [Fig. 1(a)]. For samples S1–S4 with pre-deposited metal Ga layer, dense and uniform crystalline grains are dominated on the surface. The grains tend to grow and to coalesce with the thickness of Ga layer increasing as shown in Figs. 1(b)–1(e). The specific RMS values for S1, S2, S3, and S4 are 11.5 nm, 22.3 nm, 27.3 nm, and 35.3 nm, respectively. The average diameters of the grains on the surfaces are 150 nm (S1), 185 nm (S2), 245 nm (S3), and 400 nm (S4), respectively. Figure 1(f) shows the detailed three-dimensional surface feature of S2 as an example, where the grains clearly protrude from the surface. Considering the above results, it is supposed that annealing of the pre-buried metal gallium layer contributes to the rough surface. In fact, the surface protrusion in Fig. 1(f) is composed of the interior discrete metal Ga NSs and the surrounding β-Ga₂O₃ matrix, which has been confirmed by TEM observations (not shown here). The Ga NSs in S2 is about 10 nm–20 nm in diameter with an adjacent distance of 5 nm–20 nm. Ga NSs with this size and distribution can couple with the energy band of Ga₂O₃. Formation of the Ga/Ga₂O₃ nanocomposite film is caused by the Ga atoms migrating from the interlayer upwards to the surface and their partial oxidation in the thermal annealing process. Therefore, the sample without Ga interlayer (S0) is much smoother.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) AFM images of the samples S0–S4: (a) S0 (without Ga layer), (b) S1 (with 10-nm Ga), (c) S2 (with 20-nm Ga), (d) S3 (with 30-nm Ga), (e) S4 (with 40-nm Ga), and (f) three-dimensional feature of S2 (with 20-nm Ga).

To further investigate the crystal structures of all the samples, XRD measurements are carried out. Figure 2(a) reveals that all the diffraction peaks of the bare Ga₂O₃ sample (S0) can be indexed and belong to the monoclinic structure, in good consistence with those of β-Ga₂O₃ (a = 12.214 Å, b = 3.037 Å, c = 5.798 Å, β = 103.8○, JCPDS Card No. 87-1901). The four dominant peaks located at 18.9○, 30.2○, 31.8○, and 64.7○ are assigned to , (400), , and planes respectively, indicating that all the samples are polycrystalline. After incorporating Ga, two remarkable differences occur, in contrast to S0. Firstly, the intensities of and planes decrease. Secondly, some new diffraction peaks appear and all of them belong to the planes of monoclinic Ga₂O₃. Notably, the intensity of the new peak at 35.3^○, coming from β-Ga₂O₃ (111) plane, increases as the thickness of Ga interlayer increases. The structural evolutions and their influence on PD performance are not clear and need further investigating.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) (a) XRD patterns and (b) optical transmittance spectra of samples S0–S4. Inset shows the magnified curves around the absorption edges.

Transmittance spectra of all the samples are exhibited in Fig. 2(b), where each of all the samples demonstrates transparency in the visible spectrum range with a high transmittance over 75%. The curves of S1–S4 samples with different thickness values of pre-buried Ga layer show various degrees of red shifts compared with that of the bare Ga₂O₃ sample (S0). A close view of their absorption edges presents the red shift tendencies of these nanocomposite samples (see the inset of Fig. 2(b)). In addition to the red shift, the transmittance within a region from 300 nm to 500 nm also shows a little degree of gradual decrease with Ga thickness increasing. All the changes in the transmittance spectra are presumably ascribed to the increasing of residual Ga NSs and their sizes in the Ga/Ga₂O₃ nanocomposite films.

Prototype solar-blind PDs are fabricated on these samples each with an MSM structure, and their photoresponse performance is shown in Fig. 3. Figure 3(a) presents the I–V characteristics of the devices in the dark. As the thickness of the Ga interlayer increases, the dark current first increases and then decreases. The S0 illustrates the lowest dark current (only 2.76 pA at 10 V), and S2 has the largest dark current (8.52 pA at 10 V) among all the samples. Actually, the dark currents for all PDs are relatively low compared with previous results.^[14,35–37] The evolution of photocurrent under the 254-nm UV illumination shown in Fig. 3(b) follows the same trend as the dark current, i.e., S2 exhibits the largest current increasment compared with S0. Table 1 shows the photoelectric parameters of all the samples. It can be seen that 10-nm pre-buried Ga can enhance the photocurrent by about two orders of magnitude. Further increasing the thickness of Ga metal layer, the photocurrent can be increased about 300 times for S2. The enhancement ratio of the photocurrent is much larger than that of the dark current whose enhancement is only about 3 times. Therefore, a much higher light-to-dark ratio (I_poto/I_dark, ∼8 × 10⁵ for S2) is achieved when the Ga NSs/Ga₂O₃ nanocomposite film is employed as a photo active layer. In order to obtain the information about the response speed and repeatability of the devices, time-dependent deep UV response tests are performed with a UV 254-nm hand-held lamp on and off periodically. As shown in Fig. 3(c), each of all the detectors presents a very stable and repeatable photoresponse behavior. Moreover, there is no obvious change in the rise time nor in the decay time of the device when changing the thickness of the Ga layer. The spectral response characteristics under 15-V bias of the PDs are shown in Fig. 3(d). All the response peaks are located at ∼ 260 nm, which is consistent with previous report on the Ga₂O₃ detector.^[38] As summarized in Table 1, the peak responsivity for S0 is only 0.013 A/W. When the samples are embedded with discrete Ga NSs, the responsivities increase dramatically. Especially for sample S2, the responsivity reaches 2.85 A/W, which is an extraordinary promotion compared with the responsivity for S0. The photoresponse enhancement can be adjusted by simply varying the thickness of the metal layer. The big enhancement of the solar-blind UV photoresponse performance results from the SP coupling at the Ga NSs/Ga₂O₃ interface. Specifically speaking, the SP facilitates the plasmon-induced resonant energy transfer from metal Ga to Ga₂O₃ in the nanocomposite film. In addition, direct hot electron transfer to Ga₂O₃ is also probable due to the intimate contact between Ga NSs and Ga₂O₃, which further increases the photocurrent collected. The detailed analysis of the coupling mechanism can be found in our previous report.^[34] The plasmon peak energy will experience a blueshift with the diameter of the metal particles decreasing.^[31,39–41] The efficiency of the SP enhancement will reach a maximum when the energy of plasmon is exactly consistent with the band gap of the semiconductor. In these PD devices, the resonance enhancement effect varies with pristine Ga thickness, the biggest enhancement factor for S2. As the thickness of the middle metal Ga layer increases, the diameter and the density of the Ga NSs gradually increase correspondingly. The sizes and distribution of Ga NSs in S2 provide the best energy coupling between SP and Ga₂O₃ energy band. When the NS diameter is too small or too large, the plasmon resonance peak will deviate from the band gap of Ga₂O₃, leading to a reduced coupling efficiency. That is why a controllable SP resonance enhancement can be realized by changing the thickness of the Ga layer.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) Photoelectric properties of the PDs: I–V curves (a) in the dark and (b) under 254-nm UV light illumination, (c) time-dependent photoresponse with the 254-nm UV light periodically on and off, and (d) photoresponsivity spectra of the PDs biased at 15 V.

Table 1.

Table 1.

Table 1. Summary of the photoelectric properties of PDs. . Samples Idark@10 V/A Iphoto@10 V/A Iphoto/Idark Responsivity/(A/W) S0 2.76 × 10−12 2.23 × 10−8 8.08 × 103 0.013 S1 6.65 × 10−12 2.93 × 10−6 4.47 × 105 1.080 S2 8.52 × 10−12 6.86 × 10−6 8.05 × 105 2.854 S3 4.88 × 10−12 5.62 × 10−7 1.15 × 105 0.199 S4 3.79 × 10−12 1.27 × 10−7 3.35 × 104 0.059

Table 1. Summary of the photoelectric properties of PDs. .

To further explore the SP performance of the Ga NS array in the surrounding Ga₂O₃ matrix, the spatial electric field intensity distribution is simulated by the FDTD method. In order to match the results of S2, the diameters of the Ga NSs in the simulation are 10 nm and 20 nm, and the gaps between two NSs are 10 nm and 20 nm. Figures 4(a) and 4(b) show the simulated results with the diameter of 10 nm, and with the gaps of 20 nm and 10 nm, respectively. In the case of Fig. 4(a), the electric field intensity in the middle of the two NSs is slightly larger than that of the background, indicating the occurrence of a weak inter-sphere interaction. When the gap is reduced to 10 nm, the interaction between two Ga NSs becomes stronger and the area with the enhanced electric field runs through the entire range between the two NSs. Figures 4(c) and 4(d) show the typical interactions of the Ga NSs with d = 20 nm and gaps of 20 nm and 10 nm, respectively. The trend of the electric field changing from Fig. 4(c) to Fig. 4(d) is the same as that for the gap of 10 nm. The electric field distribution and the inter-sphere interaction are similar to those of other metal NSs.^[23,41–43] As the diameter of the metal NS increases, the electric field further strengthens and the inter-sphere interaction becomes stronger. The FDTD simulated results indicate that the local electromagnetic field around metal Ga NSs in the nanocomposite film can be greatly increased by the SP. The enhanced electric field around the metal NSs can effectively accelerate the separation of the photo-induced electron-hole pairs and thus a higher UV response can be achieved.^[24,41]

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) Electric field distributions of Ga NSs in Ga2O3 matrix with different diameters and gaps: (a) diameter = 10 nm and gap = 20 nm; (b) diameter = 10 nm and gap = 10 nm; (c) diameter = 20 nm and gap = 20 nm; (d) diameter = 20 nm and gap = 10 nm.

A series of Ga NSs/Ga₂O₃ nanocomposite films is acquired through post-annealing the Ga₂O₃/Ga/Ga₂O₃ triple-layer films. The nanocomposite films are each composed of polycrystalline β-Ga₂O₃ matrix and discrete metal Ga NSs embedded therein. The sizes of the metal Ga NSs in the nanocomposite films vary when changing the metal Ga layer thickness. The corresponding PDs demonstrate adjustable solar-blind UV photoresponse enhancement factors due to the coupling efficiency varying by tuning the morphology of Ga NSs. The maximum peak responsivity reaches 2.85 A/W at 260 nm, two orders of magnitude higher than the maximum peak responsivity of the bare Ga₂O₃ PD. This work reveals the remarkable and tunable plasmonic coupling capability in Ga/Ga₂O₃ systems. The results presented here are of profound significance for studying plasmonics and detecting solar-blind UV light. Meanwhile, the simple, repeatable, and cost-effective method of fabricating the high-response Ga₂O₃ solar-blind UV PDs promises to have practical applications.