1. Introduction

AlGaN-based solar-blind ultraviolet (SB-UV) detectors have attracted much attention because they are in a solid state and small in size, with good chemical and thermal stability, thereby requiring less energy and having a long lifetime. These detectors have wide potential applications in flame detection, environmental monitoring, medical diagnostics and even UV astronomy [1]. To date, AlGaN-based SB-UV detectors have been fabricated with metal-semiconductor-metal (MSM), Schottky and PIN structures. However, the performance of these detectors remains lower than expected; a much higher responsivity and a much lower dark current are needed before the detectors can be widely used for the detection of very weak UV signals. Due to the lack of suitable homosubstrates, the dislocation density in AlGaN is still as high as 10⁹-10¹⁰ cm⁻². This problem cannot be resolved without the use of more suitable substrates. Thus, greater attention has recently been given to optimizing the structure of AlGaN-based SB-UV detectors. High-gain AlGaN-based SB-UV detectors have been obtained with an avalanche multiplication structure [2]. However, the stringent requirements on crystalline quality, device structure and processing techniques hinder the wide application of avalanche-type SB-UV detectors. Moreover, these detectors work at high voltage and usually suffer from noise. Recently, the interaction of light with metallic nanoparticles has led to a diversity of profound effects that have improved the performance of optoelectronic devices [3–8]. Plasmonic phenomena and related effects offer new opportunities for tuning the performance of photodetectors, photovoltaics, and light-emitting structures via improved coupling between optical processes in the device active region and incident or emitted light.

The localized surface plasmon (LSP) enhancement effect results from an incident light field exerted on the free electrons of a metal, leading to collective electron-photon oscillations. When the frequency ω_sp of this collective electron-photon oscillation is close to the frequency of the excitation light wave, resonance occurs, which enables a deep subwavelength localization of incident electromagnetic fields [9–11]. Based on this principle, high responsivity in GaN UV detectors has been realized by enhancement with silver (Ag) NPs [12]. However, because the light frequency in the SB-UV region is higher than the plasmon frequencies of metals such as Ag and gold (Au), the light propagating inside the metals is strongly absorbed. Therefore, these metals are not suitable for the excitation of surface plasmon resonance (SPR) in the SB-UV region. Instead, Al is the best plasmonic material in the SB-UV region because of its high plasmon resonance energy [13]. However, the properties associated with the plasmonic resonances of Al used in AlGaN-based optoelectronic devices must be further understood and exploited, especially for use in AlGaN-based SB-UV detectors. In this study, Al NPs were prepared to enhance the spectral response of AlGaN-based SB-UV detectors. The enhancement mechanism is explored and discussed by means of the finite-difference time-domain (FDTD) method based on the quasi-particle approximation.

2. Experimental section

An undoped AlGaN epilayer was grown on a (0001) sapphire substrate by metal-organic chemical vapor deposition (MOCVD). Trimethylgallium, trimethylaluminum and ammonia were used as Ga, Al and N precursors, respectively. An AlN template was first grown on a sapphire substrate by a two-step growth method, and the temperature was then ramped to 1200 °C to produce an AlGaN layer [14]. Nickel (Ni) (50 nm) Schottky contacts were fabricated for AlGaN MSM interdigitated contacts, and the samples were then treated by rapid thermal annealing at 450 °C for 180 s. The fingers were 100 μm long and 5 μm wide, with a spacing of 10 μm. The devices were fabricated using standard photolithography and lift-off technology. Al nanoparticles were then deposited on the active area of AlGaN detectors by electron beam evaporation through the surface of the detectors parallel to the Al evaporant. The deposition rate was approximately 0.2 nm per second under a pressure of 4 × 10⁻⁶ mbar. The deposition time determined the size of the Al particles. A schematic diagram of an AlGaN-based MSM detector with Al NPs is shown in Fig. 1(a).

 

Fig. 1 (a) Schematic illustration of Al NPs on an AlGaN-based deep-UV detector. (b-d) Top-view SEM images of Al NPs with diameters of 20 nm, 40 nm, and 60 nm, respectively, deposited on the detectors. (e-f) 3D morphology and corresponding cross-sectional profile of the Al NPs.

Download Full Size | PDF

Scanning electron microscopy (SEM) was used to characterize the size and distribution of the Al NPs. Atomic force microscopy (AFM) was employed to measure the height of the Al NPs. In measuring the optical spectral response of the devices, a UV-enhanced Xe arc lamp was used as the optical source for spectral responsivity studies. A mechanical chopper modulated the incident light, and a lock-in amplifier recorded the photocurrent from the MSM barrier photodetector. The system was calibrated with a UV-enhanced Si detector. A Keithley 6487 electrometer was employed to test the current-voltage (I-V) performance of the AlGaN-based SB-UV detectors. All measurements were carried out at room temperature.

3. Results and discussion

AlGaN-based SB-UV detectors with different Al NP sizes and distributions were fabricated. Figures 1(b)-1(d) show the SEM morphologies of three samples with Al NPs deposited for times of 40 s, 60 s, and 80 s, respectively. The sizes of the Al NPs were measured by SEM; longer deposition times resulted in larger Al NPs. The diameters of the Al NPs were approximately 20 nm at 40 s, 40 nm at 60 s and 60 nm at 80 s. The spacing distance of the randomly distributed Al NPs ranged from 5 nm to 40 nm.

The responsivity of the AlGaN-based SB-UV detectors was investigated. Figure 2 shows the spectral responses of the detectors without NPs and with different sizes of Al NPs, where sample A is the detector without Al NPs and samples B, C and D include Al NPs with diameters of 20 nm, 40 nm and 60 nm, respectively. All detectors were measured under a 5-V bias. All samples show high responsivity, with a peak response at 288 nm related to the 40% Al content of the AlGaN and a sharp cut-off above the band gap. Meanwhile, the responsivity for all detectors decreases with decreasing wavelength due to the reduced photo-penetration depth and increasing recombination of surface photo-generated carriers. However, the peak responsivity is enhanced for all three samples with Al NPs compared to that without Al NPs, although the enhancement factor is different. The peak responsivity is the highest for sample D, with an Al NP diameter of approximately 60 nm; the responsivity increased from 0.141 A/W to 0.288 A/W, greater than a two-fold enhancement. Meanwhile, the enhancements for samples B and C are 1.5 and 1.7 times, respectively, compared with that of sample A. More than 10 detectors, corresponding to each type of diameter, were measured, and all exhibited the same phenomena. In addition, after a series of repeated measurements at 5 V, the detectors showed the same spectral response properties, indicating that the Al NPs on the surface are stable.

 

Fig. 2 Spectral response of AlGaN detectors with and without Al nanoparticles under a 5-V bias.

Download Full Size | PDF

Because the only difference between these detectors is the presence and size of the Al NPs, the spectral response enhancement may arise from the localized surface plasmon resonance effects of the Al nanoparticles. One possible mechanism (Fig. 3) is that incident light on the Al nanoparticles causes collective oscillations of the conducting electrons due to the oscillating electromagnetic field of the light. The subsequent polarization effects and restoring forces lead to resonance behavior. The localized electric field enhancement effect and related scattering effect generate more electron-hole pairs and thus a higher responsivity [12, 15].

 

Fig. 3 Schematic mechanism of AlGaN-based detectors with Al NPs.

Download Full Size | PDF

To confirm the above hypothesis, the extinction spectra (scattering + absorption) of single and dimer Al NPs were simulated via the FDTD method. Circular and cylindrical Al NPs were used, based on our experiment, and a 3-nm Al₂O₃ shell was considered in our calculation [16]. The extinction spectra for single Al NPs are shown in Fig. 4(a). All spectral positions of the plasmon resonance for isolated Al NPs whose diameters change from 20 nm to 60 nm occur in the UV region. Therefore, we easily determined that the spectral response enhancement in the AlGaN-based SB-UV detectors arose from the Al NPs in our experiment. Moreover, the Al NP size has a very important influence on the localized surface plasmon resonances; the spectral position of the plasmon resonance shifts from 230 nm to 290 nm as the diameter of an isolated Al NP increases from 20 nm to 60 nm. In addition, the resonance amplitude is enhanced with increasing Al NP size. Furthermore, larger Al NPs can gather more electromagnetic energy due to their large absorption cross sections, as well as higher polarizabilities and larger radiative rates [17], thus leading to higher resonance amplitudes. However, a larger size corresponds to a longer coupling wavelength, resulting in a red shift of the spectral position of the plasmon resonance [18]. According to the simulation, the spectral position of the plasmon resonance for an Al NP with a diameter of 60 nm (290 nm) is the nearest to the peak responsivity position of our AlGaN-based SB-UV detectors (288 nm), which may be the main reason that the highest peak responsivity occurs in the detector with 60-nm-diameter Al NPs.

 

Fig. 4 FDTD simulations. (a) Normalized extinction spectra (scattering + absorption) of isolated Al/Al₂O₃ core/shell nanocylinders. (b)-(d) Normalized extinction spectra of Al/Al₂O₃ core/shell nanocylinders with Al NP diameters D of 20 nm, 40 nm, and 60 nm, respectively, and gaps ranging from 5 to 100 nm.

Download Full Size | PDF

Considering the random Al NP distribution, interparticle coupling between NPs was also explored. Figures 4(b)-4(d) show the extinction spectra for dimers with diameters of 20 nm, 40 nm and 60 nm, respectively. For all three cases, the spectral position of the plasmon resonance of the Al NPs appears to redshift when the gap decreases due to the interaction coupling between the Al NPs [19–22]. This interaction effect is sufficiently weak to be neglected when the gap is 100 nm, where the spectral position of the plasmon resonance of the Al NPs is similar to that of isolated NPs [21, 23]. This redshift phenomenon may also enhance the performance of our detectors with Al NPs, especially for those with small Al NPs. The spectral positions of the plasmon resonance of the Al NPs shift to approximately 280 nm for a diameter (D) of 40 nm and 250 nm for D = 20 nm when the gap is 5 nm, as shown in Fig. 4(b) and 4(c). After this redshift, the spectral positions of the plasmon resonance of the dimer Al NPs are closer to the peak responsivity position of our experimental results than that of isolated Al NPs with the same size. Thus, the plasmon resonance effect will become stronger and cause a higher responsivity in these detectors.

To further explore the enhanced responsivity of AlGaN-based SB-UV detectors with Al NPs, the spatial distributions of the electric field intensity were simulated for both isolated and dimer Al NPs. Figure 5(a) shows the calculation results for an isolated Al NP with a diameter of 20 nm, 40 nm and 60 nm. The electric field intensity increases with increasing Al NP diameter, which is consistent with the above analysis, explaining why the peak responsivity for sample D is the highest. Moreover, there is an observable electric field in the Al₂O₃ shell for Al NPs with D = 20 nm, explaining the quadrupole resonance at 210 nm for these NPs, as shown in Fig. 4(b). The interparticle interaction of the Al NPs was also calculated, and Fig. 5(b) shows the typical interaction of Al NPs with D = 40 nm and a gap of 10 nm, 20 nm, 40 nm. The interaction effect becomes stronger as the gap decreases from 40 nm to 10 nm, resulting in a redshift of the spectral positions of the plasmon resonance from 250 nm to 275 nm, which enhances the responsivity of the AlGaN-based SB-UV detectors in the short-wavelength region.

 

Fig. 5 (a) Electric field amplitude distribution of a single Al nanocylinder with a diameter of 20/40/60 nm and a 3-nm Al₂O₃ shell. (b) Electric field amplitude distribution of Al NPs with a diameter of 40 nm and a gap g of 10/20/40 nm.

Download Full Size | PDF

Because the dark current is important in the performance of AlGaN-based SB-UV detectors, the effect of Al NPs on the dark current was also studied. As shown in Fig. 6, the dark current of the detectors with Al NPs is not increased but slightly reduced compared with that of the detectors without Al NPs. Dislocations, especially screw dislocations, are the main leakage path in AlGaN-based optoelectronic devices, and dielectric materials, such as Al₂O₃ and SiO₂, can inhibit leakage by terminating screw dislocations [24, 25]. Thus, the reduced dark current caused by the Al NPs may result from the passivation effect of the Al₂O₃ shell.

 

Fig. 6 Dark current-voltage (I-V) curves of AlGaN detectors without NPs and with 20-nm, 40-nm, and 60-nm Al NPs.

Download Full Size | PDF

4. Conclusion

In conclusion, an enhanced spectral response of AlGaN-based SB-UV detectors has been obtained by using Al NPs of 20-60 nm. The size of the Al NPs determines the enhancement factor, and the peak responsivity (about 288 nm) was more than 2 times larger when using Al NPs with a diameter D of 60 nm, reaching 0.288 A/W. Based on theoretical calculations and analysis, the dipole surface plasmon resonance wavelength of the Al NPs is localized at approximately 288 nm for a diameter of 60 nm. Thus, the localized surface plasmon resonance effect results in the improvement of AlGaN-based solar-blind detectors. The localized electric field enhancement generates more electron-hole pairs and thus a higher responsivity. In addition, the related scattering effect enhances the absorption of light, leading to further responsivity enhancement. It is difficult to improve the performance of AlGaN-based SB-UV detectors by improving the crystallinity of the AlGaN; therefore, our results may greatly promote the development and application of AlGaN-based SB-UV detectors.