Hot electron generation by aluminum oligomers in plasmonic ultraviolet photodetectors

We report on an integrated plasmonic ultraviolet (UV) photodetector composed of aluminum Fano-resonant heptamer nanoantennas deposited on a Gallium Nitride (GaN) active layer which is grown on a sapphire substrate to generate significant photocurrent via formation of hot electrons by nanoclusters upon the decay of nonequilibrium plasmons. Using the plasmon hybridization theory and finite-difference time-domain (FDTD) method, it is shown that the generation of hot carriers by metallic clusters illuminated by UV beam leads to a large photocurrent. The induced Fano resonance (FR) minimum across the UV spectrum allows for noticeable enhancement in the absorption of optical power yielding a plasmonic UV photodetector with a high responsivity. It is also shown that varying the thickness of the oxide layer (Al2O3) around the nanodisks (tox) in a heptamer assembly adjusted the generated photocurrent and responsivity. The proposed plasmonic structure opens new horizons for designing and fabricating efficient opto-electronics devices with high gain and responsivity. ©2016 Optical Society of America OCIS codes: (040.5160) Photodetectors; (250.5403) Plasmonics; (250.0250) Optoelectronics. References and links 1. B. Heshmat, H. Pahlevaninezhad, Y. Pang, M. Masnadi-Shirazi, R. Burton Lewis, T. Tiedje, R. Gordon, and T. E. Darcie, “Nanoplasmonic terahertz photoconductive switch on GaAs,” Nano Lett. 12(12), 6255–6259 (2012). 2. C. C. Chang, Y. D. Sharma, Y. S. Kim, J. A. Bur, R. V. Shenoi, S. Krishna, D. Huang, and S.-Y. Lin, “A surface plasmon enhanced infrared photodetector based on InAs quantum dots,” Nano Lett. 10(5), 1704–1709 (2010). 3. J. Hetterich, B. Bastian, N. A. Gippius, S. G. Tikhodeev, G. von Plessen, and U. Lemmer, “Optimized design of plasmonic MSM photodetector,” IEEE Quantum Electron. 43(10), 855–859 (2007). 4. X. Wang, Z. Cheng, K. Xu, H. K. Tsang, and J. B. Xu, “High-responsivity graphene/silicon-heterostructure waveguide detectors,” Nat. Photonics 7(11), 888–891 (2013). 5. D. M. Schaadt, B. Feng, and E. T. Yu, “Enhanced semiconductor optical absorption via surface plasmon excitation in metal nanoparticles,” Appl. Phys. Lett. 86(6), 063106 (2005). 6. R. Sundararaman, P. Narang, A. S. Jermyn, W. A. Goddard 3rd, and H. A. Atwater, “Theoretical predictions for hot-carrier generation from surface plasmon decay,” Nat. Commun. 5, 5788 (2014). 7. H. Morkoc, A. Di Carlo, and R. Cingolani, “GaN-based modulation doped FETs and UV detectors,” Solid-State Electron. 46(2), 157–202 (2002). 8. W. Zhang, J. Xu, W. Ye, Y. Li, Z. Qi, J. Dai, Z. Wu, C. Chen, J. Yin, J. Li, H. Jiang, and Y. Fang, “Highperformance AlGaN metal-semiconductor-metal solar blind ultraviolet photodetectors by localized surface plasmon enhancement,” Appl. Phys. Lett. 106(2), 021112 (2015). 9. D. Gedamu, I. Paulowicz, S. Kaps, O. Lupan, S. Wille, G. Haidarschin, Y. K. Mishra, and R. Adelung, “Rapid fabrication technique for interpenetrated ZnO nanotetrapod networks for fast UV sensors,” Adv. Mater. 26(10), 1541–1550 (2014). 10. D. Li, X. Sun, H. Song, Z. Li, Y. Chen, H. Jiang, and G. Miao, “Realization of a high-performance GaN UV detector by nanoplasmonic enhancement,” Adv. Mater. 24(6), 845–849 (2012). 11. G. C. Hu, C. X. Shan, N. Zhang, M. M. Jiang, S. P. Wang, and D. Z. Shen, “High gain Ga2O3 solar-blind photodetectors realized via a carrier multiplication process,” Opt. Express 23(10), 13554–13561 (2015). 12. B. Zhao, F. Wang, H. Chen, Y. Wang, M. Jiang, X. Fang, and D. Zhao, “Solar-blind Avalanche photodetector based on single ZnO-Ga2O3 core-shell microwire,” Nano Lett. 15(6), 3988–3993 (2015). 13. B. Mallampati, S. V. Nair, H. E. Ruda, and U. Philipose, “Role of surface in high photoconductive gain measured in ZnO nanowire-based photodetector,” J. Nanopart. Res. 17(4), 176 (2015). #261332 Received 17 Mar 2016; revised 21 May 2016; accepted 25 May 2016; published 10 Jun 2016 © 2016 OSA 13 Jun 2016 | Vol. 24, No. 12 | DOI:10.1364/OE.24.013665 | OPTICS EXPRESS 13665 14. J. Yu, C. X. Shan, X. M. Huang, X. W. Zhang, S. P. Wang, and D. Z. Shen, “ZnO-based ultraviolet avalanche photodetectors,” J. Phys. D Appl. Phys. 46(30), 305105 (2013). 15. Y. Yu, Y. Jiang, K. Zheng, Z. Zhu, X. Lan, Y. Zhang, Y. Zhang, and X. Xuan, “Ultralow-voltage and high gain photoconductor based on ZnS:Ga nanoribbons for the detection of low-intensity ultraviolet light,” J. Mater. Chem. C Mater. Opt. Electron. Devices 2(18), 3583–3588 (2014). 16. W. Y. Weng, T. J. Hsueh, S. J. Chang, S. B. Wang, H. T. Hsueh, and G. J. Huang, “A high-responsivity GaN nanowire UV photodetector,” IEEE Sel. Top. Quantum Electron. 17(4), 996–1001 (2011). 17. Y. Q. Bie, Z.-M. Liao, H.-Z. Zhang, G.-R. Li, Y. Ye, Y.-B. Zhou, J. Xu, Z.-X. Qin, L. Dai, and D.-P. Yu, “Selfpowered, ultrafast, visible-blind UV detection and optical logical operation based on ZnO/GaN nanoscale p-n junctions,” Adv. Mater. 23(5), 649–653 (2011). 18. M. W. Knight, N. S. King, L. Liu, H. O. Everitt, P. Nordlander, and N. J. Halas, “Aluminum for plasmonics,” ACS Nano 8(1), 834–840 (2014). 19. M. W. Knight, L. Liu, Y. Wang, L. Brown, S. Mukherjee, N. S. King, H. O. Everitt, P. Nordlander, and N. J. Halas, “Aluminum plasmonic nanoantennas,” Nano Lett. 12(11), 6000–6004 (2012). 20. V. S. Kortov, S. V. Zvonarev, and A. Medvedev, “Pulsed cathodoluminescence of nanoscale aluminum oxide with different phase compositions,” J. Lumin. 131(9), 1904–1907 (2011). 21. Q. Xu, F. Liu, Y. Liu, W. Meng, K. Cui, X. Feng, W. Zhang, and Y. Huang, “Aluminum plasmonic nanoparticles enhanced dye sensitized solar cells,” Opt. Express 22(S2), A301–A310 (2014). 22. J. Becker, Plasmons as Sensors (Springer, 2012). 23. J. A. Fan, C. Wu, K. Bao, J. Bao, R. Bardhan, N. J. Halas, V. N. Manoharan, P. Nordlander, G. Shvets, and F. Capasso, “Self-assembled plasmonic nanoparticle clusters,” Science 328(5982), 1135–1138 (2010). 24. P. Nordlander, C. Oubre, E. Prodan, K. Li, and M. I. Stockman, “Plasmon hybridization in nanoparticle dimers,” Nano Lett. 4(5), 899–903 (2004). 25. B. Luk’yanchuk, N. I. Zheludev, S. A. Maier, N. J. Halas, P. Nordlander, H. Giessen, and C. T. Chong, “The Fano resonance in plasmonic nanostructures and metamaterials,” Nat. Mater. 9(9), 707–715 (2010). 26. J. B. Lassiter, H. Sobhani, J. A. Fan, J. Kundu, F. Capasso, P. Nordlander, and N. J. Halas, “Fano resonances in plasmonic nanoclusters: geometrical and chemical tunability,” Nano Lett. 10(8), 3184–3189 (2010). 27. Z. Fang, Y. Wang, Z. Liu, A. Schlather, P. M. Ajayan, F. H. L. Koppens, P. Nordlander, and N. J. Halas, “Plasmon-induced doping of graphene,” ACS Nano 6(11), 10222–10228 (2012). 28. S. Golmohammadi and A. Ahmadivand, “Fano resonances in compositional clusters of aluminum nanodisks at the UV spectrum: A route to design efficient and precise biochemical sensing,” Plasmonics 9(6), 1447–1456 (2014). 29. V. Giannini, A. I. Fernández-Domínguez, Y. Sonnefraud, T. Roschuk, R. Fernández-García, and S. A. Maier, “Controlling light localization and light-matter interactions with nanoplasmonics,” Small 6(22), 2498–2507 (2010). 30. V. Giannini, A. I. Fernández-Domínguez, S. C. Heck, and S. A. Maier, “Plasmonic nanoantennas: fundamentals and their use in controlling the radiative properties of nanoemitters,” Chem. Rev. 111(6), 3888–3912 (2011). 31. P. Reineck, G. P. Lee, D. Brick, M. Karg, P. Mulvaney, and U. Bach, “A solid-state plasmonic solar cell via metal nanoparticle self-assembly,” Adv. Mater. 24(35), 4750–4755 (2012). 32. A. Manjavacas, J. G. Liu, V. Kulkarni, and P. Nordlander, “Plasmon-induced hot carriers in metallic nanoparticles,” ACS Nano 8(8), 7630–7638 (2014). 33. M. W. Knight, H. Sobhani, P. Nordlander, and N. J. Halas, “Photodetection with active optical antennas,” Science 332(6030), 702–704 (2011). 34. F. Wang and N. A. Melosh, “Plasmonic energy collection through hot carrier extraction,” Nano Lett. 11(12),


Introduction
In recent years, there has been growing interest for plasmonic photodetectors for a wide spectral range covering terahertz (THz) to visible frequencies [1][2][3].In all these works, several strategies have been used to enhance the absorption, responsivity, and quantum efficiency of the photodetectors.For instance, graphene plasmonics has been introduced as a promising platform to enhance the absorption of metal-semiconductor-metal (MSM) photodetectors including Schottky contacts at the optical and THz frequencies [4].Moreover, plasmonic nanoparticles with absorptive characteristics (ohmic losses) have widely been utilized to improve the spectral response of detectors [5].In the previously reported works, several techniques were used to enhance the detection performance of plasmonic photodetectors such as enhancing of the Schottky barrier height at the metal-semiconductor interface, which provides a wider depletion region [6,7], and excitation of surface plasmon resonances based on collective, coherent hot electron oscillations [8,9].The ultraviolet (UV) detectors are very useful for applications in UV astronomy, environmental monitoring, missile warning, and biotechnology and medicine.However, in spite of the extensive researches, the UV photodetectors suffer from dissipative losses, large dark currents, limited responsivity and quantum efficiency [10,11].To address these challenges and improve the performance of the UV detectors, two major methods have been proposed: (1) Avalanche multiplication [12], and (2) photoconductive gain [13].However, high responsive GaN-based avalanche detectors suffer from an increased noise [14].On the other hand, the photoconductive UV detectors are slow and noisy [15].As another solution, GaN-based UV photodetectors with silver (Ag) plasmonic nanoparticles have been introduced to enhance the responsivity [6,16,17].The major problem correlating with this method is the performance of utilized metals for UV bandwidth.The plasmon resonances in the subwavelength structures based on conventional noble metals (e.g.Au, Ag, and Cu) can be tuned across the visible wavelengths to the near infrared region (NIR).However, extending these plasmonic properties into the UV spectrum is highly challenging due to the intrinsic limitations in the chemical characteristics of the used metals.For instance, silver shows a dramatic degradation in plasmonic properties because of rapid oxidation and gold suffers from the interband transitions in the UV band [18].Lately, Aluminum (Al), Rhodium (Rh), Gallium (Ga), Chromium (Cr), and Indium (In) have been introduced as potential plasmonic materials for the UV spectrum [19].Aluminum has widely been employed in designing light harvesting devices, nanoantennas, cathodoluminescence spectroscopy, and antireflective surfaces [18][19][20][21], in spite of the inherent and rapid oxidation.
Aluminum also shows significant EM field localization because of its low screening ( ε ∞ ≈ 1) in comparison to gold ( ε ∞ ≈ 9) and silver ( ε ∞ ≈ 5).In addition, aluminum has high electron density since a single aluminum atom contributes three electrons compared to a single electron per atom for gold and silver [22].Due to the negligible influence of interband transitions in aluminum across the UV spectrum, therefore, the geometry of nanoscale structure plays a major role in decaying plasmons and generation of photoexcited hot carriers during light-matter interactions.
Closely packed and strongly coupled plasmonic nanoparticle assemblies in symmetric and antisymmetric orientations, known as plasmonic oligomers, can be tailored to support strong resonances across the visible to the NIR [23].These nanoparticle clusters show significant absorption cross-sections in the visible and NIR ranges, including strong plasmon resonance hybridization in the offset gaps between proximal particles [24].Depending on their shape and orientation, nanoparticle clusters are able to show unique spectral lineshapes, called "Fano resonances (FR)" that can be characterized by narrow spectral windows, where scattering maxima are suppressed and absorption peaks are enhanced [23][24][25].The physical mechanism behind formation of the plasmonic FR is a weak and destructive coupling between a spectrally broad superradiant mode and a narrow subradiant mode.When a plasmonic oligomer is excited at the frequency of the bonding mode, the incident light directly couples into the bonding mode via light-matter interaction resulting a robust indirect excitation of the antibonding resonant mode.In the nonretarded limit, the antibonding mode is dark without a net dipole moment and, hence, cannot be coupled directly to the incident beam.In contrast, in the retarded limit, the bonding mode becomes bright and a weak coupling mediated by the strong near-field coupling gives rise to the interaction between bonding and antibonding resonant modes, inducing a FR dip in the bonding continuum at the energy level of the dark mode [26].Excitation of a plasmonic FR mode leads to significant absorption compared to excitation of usual bright resonant mode which could be used for hot electron generation.This feature of plasmonic FR mode can be exploited to enhance the photocurrent in plasmonic photodetectors [27].However, inducing FR dips in the UV band is challenging due to limitations correlating with the chemical properties of conventional noble metals at this domain.Recently, it is proved that Al/Al 2 O 3 nanodisk clusters in symmetric and antisymmetric orientations are able to support strong FR modes with excellent absorption coefficient at the UV spectrum [28].Low-cost, CMOS compatibility, and supporting strong plasmon resonances at the UV spectrum are some of unique features of aluminum-based molecular clusters that make these nanoscale assemblies suitable for designing efficient nanoplasmonic devices.
In this paper, we propose a novel device based on plasmonic Al/Al 2 O 3 nanoparticle assemblies integrated into a GaN UV photodetector.To this end, we utilized seven-member heptamers with the symmetry of a benzene molecule as Fano-resonant plasmonic nanoclusters.All of the aluminum particles are deposited between Ni/Au fingers on a GaN active layer grown on a sapphire substrate.The presented results show that nanoplasmonic aluminum assemblies could generate hot electrons to enhance the absorption via inducing the FR modes across the UV spectrum.The proposed structure could realize the UV photodetectors with a significantly improved responsivity.

The proposed device
Radiative and non-radiative excitation of plasmons in metallic components and their decay leads to generation of hot carriers at the metal-semiconductor interfaces [6,[29][30][31].The surface modes are important to achieve the plasmon resonant behavior and hot carrier distribution at the metal-semiconductor interfaces [8,36,37].Plasmonic photovoltaic devices [32,33], and photodetectors [34,35] employ the decay of plasmons to generate hot carriers.Aluminum with the ability of generating continuous energy distribution for electrons is one of the most suitable metals for this process [6].In our proposed system, the confined plasmons also lead to the hotspot formation with extremely intense local fields in the capacitive regions between the proximal particles.Assuming electrons have an isotropic momentum distribution, approximately half of the photoexcited electrons are expected to be transported to the aluminum-GaN interface.Due to the continuous distribution of highly energetic hot electrons in the aluminum nanostructures [6,38], hence, we expect large number of charges to reach the metal/semiconductor interface compared to the conventional noble metals.Figure 1(a) shows a three-dimensional schematic of the proposed plasmonic UV detector (not to scale).The device comprises arrays of Al/Al 2 O 3 heptamer antennas between two Ni/Au fingers (electrodes) deposited on an undoped n-type GaN epilayer with the thickness of 4 μm which is grown on a sapphire substrate.The inset figure shows the geometry of the heptamer assembly.The space between two neighboring heptamers is set to 250 nm to prevent any destructive optical interference between the scattered fields associating with hybridized modes arising from the nearby antennas.In Fig. 1(b), we show the important geometrical dimensions for the metallic electrodes, the overall size of the proposed photodetector, and the distance between two fingers.Using the plasmon hybridization theory to analyze closely packed nanoscale assemblies, the plasmon responses of various types of aluminum-based nanodisk oligomers and monomers have already been investigated numerically and analytically [28,39].It is also shown that aluminum nanodisk heptamer antennas with a thin oxide layer (2-25 nm, depends on the size of consisting particles) can be tailored to support strong plasmonic FR mode across the near-UV (λ~350 nm) band [39].However, this wavelength is not unique and the position of FR minimum can be tuned via modifications in the geometrical, chemical and environmental parameters of the assembly.In the proposed UV detector, we used nanodisks with the geometrical dimensions that were calculated by Golmohammadi et al. [28], and accordingly, the radius of nanodisks is R = 70 nm with the thickness of t = 35 nm separated with the offset gap of D 7h = 12 nm.It should be noted that while the thickness of the oxide layer around nanoparticles is varied the size of offset gap is kept fixed to satisfy the required near-field coupling strength.To provide a detailed study and compare the effect of aluminum heptamer arrays on the responsivity and performance of the structure, we also demonstrate the spectral response of the structure without presence of antennas on GaN as the non-plasmonic regime.To this end, we used empirically measured values for a GaN-based UV detectors reported by Li et al. [6].
The frequency-dependent absorption mechanism of the proposed plasmonic UV detector is based on the hybridized plasmon resonant modes due to the interaction of an incident beam with the metallic antenna.The maximum absorption can be achieved at the spectral position of the antisymmetric FR dip, because of the suppression of the scattering bright resonant dipolar peak by narrow antibonding dark mode.Such a significant absorption leads to generation of large number of hot carriers at the metal-dielectric interface, which are transferred to semiconductor surmounting the Schottky barrier and collected by the electrodes resulting a remarkable photocurrent and hence, high responsivity.Figure 1(c) exhibits a twodimensional (xz-view) cross-sectional schematic of the proposed UV detector displaying the hot electron transport in the GaN layer to the adjacent electrodes.It is well-understood that in a metal-semiconductor system, reduced electron-electron scattering in the metallic part of the nanoantenna increases the number of hot electrons transferred to the semiconductor layer [33,41].This ultrafast transition of plasmonic charges leads to accumulation of more hot electrons and sweeping them before immediate recombination.Figure 1(d) demonstrates the schematic band diagram profile for the proposed plasmonic device, showing the carrier formation and transition mechanisms and sweeping opposite charges to the nearby electrodes.When a bias is applied (here 5.0 V) between the metallic contacts one forward and one reverse biased Schottky junctions are formed.The large electric field in between, results in sweeping of the photogenerated hot electrons to the positive electrode and thereby producing a photocurrent.However, due to losses via back-scattering, inelastic collisions, and heat energy conversion (internal damping), not all of the photoexcited electrons are injected to the semiconductor [44,45].Thus we have to consider only the hot carriers within the mean-free path (MFP) length (l p ) distance from the interface for transferring to the semiconductor over the Schottky barrier [45][46][47].In this approach, photoexcited electrons are excited from the energy states below the Fermi level (d-band) to the higher energy levels, and once they arrive at the interface with an energy larger than the Schottky barrier height get injected to the GaN.Experimental results show that MFP for electrons is strongly energy-dependent and minor perturbations in energy level results significant changes [48][49][50].In our analysis we assumed l p = 25 nm for electrons 5 eV above Fermi level energy as reported in the literature [49].Considering depicted band diagram for the Aluminum-GaN-Ni/Au structure, the electrical simulation results verify formation of a Schottky barrier with the height of Φ B = 0.87 eV.In this regime, the decayed plasmons result hot electrons that are arrived to the interface with higher energies more than ~0.87 eV are able to pass the barrier and transit to reach the biased electrode.In the examined device, hot electron generation rate (G he ) by aluminum nanoantennas at the Fano dip wavelength due to photoexcitation can be calculated using [27]: G he = PC abs (λ)/ħωA h , where P is the incident light power (20 μW), C abs (λ) is the absorption cross section as a function of resonant wavelength, and A h is the metallic nanoantenna area and found to be G he = 5 × 10 17 s −1 .Then, the electron concentration can be calculated as [27]: n e = τG he /A h , where τ is the relaxation time, which estimated to be 0.825 × 10 −6 s (see methods).The approximate electron concentration at the aluminum-GaN interface is defined as n e = 1.04 × 10 17 cm −2 .Comparing hot electron generation rate and the associated concentration in the proposed system with gold nonamers and gratings for the same purpose [27,33,40], we realized a significant enhancement due to inherent and remarkable absorptive behavior of aluminum across the UV spectrum as well as continuous electron energy distribution [6].
For conventional semiconductor layers that have broadly been utilized for photocurrent generation in designing plasmonic devices (e.g.silicon, cadmium selenide, etc.), the carrier lifetime is in the range of ~100 μs.Long carrier lifetime prevents immediate recombination and facilitates generation of large photocurrent.In contrast, the carrier lifetime and recombination process for UV-compatible GaN around is in the range of a few nanoseconds.In this regime, extremely short transition time (in the range of a few picoseconds) is required to overcome the immediate recombination of the carriers.Using experimentally and theoretically obtained values for the saturation velocity (V sat ) in n-type GaN [42,43], the transition time (t tr ) can be determined by: t tr = L/V sat , where L is the pitch (see methods).For the saturation velocity of 10 5 cm/s, with the pitch of 500 nm, the transition time is calculated as 5 × 10 −12 s (5 ps) which is extremely short (t tr <<τ n ) compared to the carrier lifetime in GaN which is around 6.5 ns (see methods).Therefore, large number of electrons can be collected before they recombine resulting photocurrent with gain.Additionally, for the uncovered parts of the photodetector, the incident photons with the energies larger than the bandgap of GaN can also be absorbed and generate electron-hole pairs.These electron-hole pairs in the uncovered parts will be added to the hot electron pairs of clusters and contribute to the photocurrent [see Fig. 1(d)].The fast relaxation and transition times constitute the base for very fast temporal response for the proposed devices.We estimated that the rise and fall times are in the sub-microsecond range using the standard methods [51,52].

Results and discussion
Figure 2(a) represents the scattering and absorption cross-sections for the aluminum heptamer.Clearly, an antisymmetric, narrow, and tunable plasmonic FR mode is induced around λ~325 nm, which is between two distinct shoulders correlating with the bonding and antibonding plasmon modes at λ~250 nm and λ~385 nm, respectively.These spectral responses are calculated by selecting the geometrical dimensions of the molecular heptamer to tune the UV detector frequency close to the FR dip frequency.Using previously discussed geometries for the heptamer clusters, we set the oxide (Al 2 O 3 ) layer thickness to t ox = 2 nm.In our analysis, we detected a distinct absorption extreme at the FR position with a couple of absorption shoulders in the vicinity of the bonding and antibonding modes in both of the examined heptamers with two different aluminum types [see Fig. 2(a)].However, the absorption at these wavelengths is not as high as the one at the FR dip position.When the incident UV beam is resonant with the induced absorption window, the strong near-field coupling of light and cluster gives rise to generation of hot electron-hole pairs.Figure 2(b) exhibits the normalized E-field map of the plasmon resonance excitation and hybridization corresponding to the plasmonic Fano dip mode wavelength in an isolated aluminum heptamer.In the plotted snapshot, formation of the hotspots at the offset spots between nanodisks is clearly visible.The full optical responses of simple and complex aluminum antennas such as scattering profiles, and extinction spectra are discussed with details in previous studies [28,53], hence, here we just considered the absorption properties of the proposed heptamer assemblies.Since the oxidation of aluminum in the subwavelength regime also affects the dielectric permittivity of the particles, the thickness of the oxide layer has an important role in hybridization of plasmon resonances in closely spaced nanoparticle assemblies.This effect is due to the variations in both chemical and geometrical features of the oxidized nanoparticle assembly.The chemical influence of the oxide layer is discussed in numerical setup part of the Methods section.For the geometrical dimensions, increasing the oxide layer directly increases the overall size of the cluster, hence, we expect noticeable variations in the spectral response of the cluster.The effect of oxide layer on the scattering efficiency profile for an isolated nanodisk with an oxide thickness of t ox ~2-3 nm have been examined, theoretically and experimentally [18].Besides, it is shown that increasing the thickness of the oxide layer results in a red-shift of scattering dipolar resonance peak to the longer wavelengths (from UV to the visible band) [53].This red-shift also contains dramatic decrements in the scattering efficiency peak.Therefore, we have to find an acceptable trade-off between scattering and absorption efficiencies by finding appropriate dimensions for the oxide layer around nanodisks in the heptamer assembly.To this end, we investigated the effect of oxide coverage on the plasmon response of the proposed heptamer.The absorption spectra for an isolated aluminum nanodisk heptamer on a glass host is plotted in Fig. 3(a).Increasing the thickness of the oxide layer leads to enhancements in the absorption efficiency due to formation of narrower FR minimum resulted by the suppression of the scattering extreme by antibonding dark resonant mode [26,39,53].This phenomenon includes a red-shift in the position of the peaks to the longer wavelengths.The reason originates from the strong EM field hybridization of plasmonic resonances in large size heptamer clusters.As a result, deeper Fano minimum in the extinction profile can be induced, including a significant enhancement in the ratio of the absorbed power.It is well-accepted that Fano dips are very sensitive to the minor alterations in the structural properties of nanoparticle clusters [25,26].Noticing in the absorption profile in Fig. 3(a), for the heptamer assembly with thicker oxide layer, the peak of the absorption is shifted to the visible spectrum that is not desired for our UV photodetector.In addition, for the ideal case, for an entirely aluminum cluster without oxide layer (t ox ~0 nm), a noticeable extreme is appeared at the short wavelengths around λ~280 nm, close to deep-UV band.For #261332 t ox = 2 nm and 4 nm two absorption extremes are obtained at λ~320 nm and 345 nm, respectively, which have almost equal amplitude.This profile also shows the absorption spectra for the UV detector without presence of nanoparticle clusters.Noticing in the corresponding curve, due to the absence of metallic components and plasmonic effects, we observe only the natural absorption of incoming UV beam by GaN substrate, which is reduced dramatically after UV band λ>400 nm.electric field is observed at the offset spots between aluminum nanodisks because of hybridization and strong confinement of the plasmonic resonant modes.Comparing two types of antennas with different oxide thicknesses, a slight difference in the enhancement is noticed.This effect can be described by the effect of oxide layer thickness on the Bruggeman dielectric function of the entire composite aluminum antenna [18], yielding different real and imaginary permittivities at different wavelengths.Hence, modifying the oxide thickness can lead to severe changes in the spectral response of the structure, as shown in the preceding profiles.This plot also shows that no distinct shoulder is observed in the electric field profile at the illumination spots for the absence of metallic heptamers and therefore the absence of the plasmonic effects.For the case without the heptamers, a thin layer of electric field appears at the surface of the GaN (with the magnitude of 1.15 × 10 5 V/cm).While for the plasmonic case, a much larger electric field is monitored below the cluster due to hybridization of plasmons (with the magnitude of 3.95 × 10 6 V/cm).Fig. 4. a,b) carrier concentration for the detector system without heptamers with bias (5.0 V), while the UV light is in OFF and ON states, respectively c,d) carrier concentration for the system with heptamers with bias (5.0 V), while the UV light is in OFF and ON states, respectively, e,f) E-field enhancement map for the device with and without clusters, while the UV light is ON and bias is 0.0 V.
Figures 4(a) and 4(b) display the electron concentration for the proposed UV photodetector device without presence of aluminum heptamers, while the bias is applied (5.0 V) and the light source is OFF and ON, respectively [these states are indicated inside the corresponding profile in Fig. 4].By applying both bias and UV beam, comparing to the absence of the beam [Fig.4(a)], a noticeable electron concentration is obtained under the electrodes [Fig.4(b)], resulting a photocurrent.On the other hand, by adding metallic nanoscale heptamers between electrodes, we observed a dramatic enhancement in the concentration of carriers resulted by the metallic clusters, as shown in Figs.4(c) and 4(d), respectively.To show the effect of plasmonic clusters on carrier generation, we used aluminum nanoparticles with the oxide coverage of t ox = 2 nm.In this regime, the generated #261332 large carrier concentration causes Schottky barrier lowering which could contribute to the enhancement of the photocurrent [54,55].The above comparison between non-plasmonic and plasmonic UV detectors can be further illustrated by plotting corresponding E-field maps for the excited electric field at the surface of the GaN, below the antennas, as shown in Figs.4(e) and 4(f), where the effect of plasmonic antennas in formation of a large field at the GaNaluminum interface is obvious.
Further, we study the electrical response of the plasmonic GaN photodetector. Figure 5(a) illustrates the current-voltage (I-V) characteristic calculated for the peak of the absorption profile along with the one for the device without plasmonic assemblies.In the calculated response, the voltage is changed between 0 V to 5.0 V.For a bias of 5.0V the plasmonic detector with the heptamer arrays with t ox = 2 nm and 4 nm, yield the photocurrents of 88.56 μA and 90.25 μA, respectively.For the non-plasmonic case (absence of metallic nanoparticle clusters), the photocurrent is found as 1.72 μA under 5.0 V bias.Dramatic enhancement in the photocurrent due to the plasmonic heptamers is clearly visible.The inset of Fig. 5(a) shows the extracted dark current as a function of the bias voltage, which reaches to 47.95 nA, 52.5 nA and 55.25 nA for the non-plasmonic case without the heptamers, t ox = 2 nm and 4 nm, respectively, under 5.0 V bias. Figure 5(b) shows the photocurrent as a function of the polarization angle of the incoming light.In the plotted figure, hallow and solid circles represent the calculated photocurrents for different incident polarization modes in two types of heptamers with different oxide thicknesses.It is observed that the response of the proposed structure is insensitive to the variations of the polarization angle of the incident EM energy due to the inherent symmetry of the molecular heptamer cluster.In addition, besides the ability to support pronounced Fano dip at the UV spectrum, it should be noted that antisymmetric structures with more complex geometries cannot provide such a high and polarization-independent absorption spectra [56].
Figure 6(a) represents the spectral response of the proposed UV detector with and without aluminum antenna arrays, where the thickness of the oxide layer is changed and the bias is kept as 5.0 V.For the plasmonic regime, the responsivity peaks are at λ~325 nm and λ~330 nm, and the cutoff wavelengths here is at λ~335 nm and λ~345 nm, for t ox = 2 nm and 4 nm, respectively.The peak responsivity (R ph ) corresponds to the position of FR dip of heptamers.At the peaks of the curve for the heptamer with t ox = 2 nm and 4 nm, the responsivity of the proposed plasmonic UV photodetector exceed 20.8 A/W and 21.9 A/W, respectively.This outcome shows the superior responsivity of the examined UV detector in comparison to analogous nanoscale devices [6,17].On the other hand, for the non-plasmonic case, we observed a conventional responsivity with a distinct shoulder at the UV spectra in the range of λ~300 nm to 350 nm, where at the highest peak this parameter is measured as approximately 0.13 A/W [see the inset diagram in Fig. 6(a)].Using the calculated responsivity data for the proposed plasmonic UV photodetector, we extracted the external quantum efficiency (EQE) for the structure in two different regimes by employing the conventional equation [11]: EQE = hcR/eλ, where h is the Planck's constant, c is the velocity of light, e is the electron charge, R is the responsivity of the device, and λ is the wavelength of the incoming optical power.The calculated EQE for the UV detector in non-plasmonic case is 64.5% while EQE is 8065% and 8116% for the devices with the presence of aluminum clusters with t ox = 2 nm and 4 nm, respectively.This dramatic enhancement in the responsivity and efficiency of the device during transition from non-plasmonic to the plasmon regime originates from the generation of hot carriers due to strong hybridization of plasmons at resonant frequencies.As the other important parameter, we estimated the internal quantum efficiency (IQE) of the proposed UV photodetector, which is the number of the produced charge carriers per incident photon and can be calculated using the computed photocurrent profile as well as the incoming photon energy flux on the subwavelength heptamer antennas.ε is the effective permittivity of the semiconductor substrate and metallic heptamer that are contributed in the absorption mechanism.Accordingly, the number of absorbed photon is given by: ( ) On the other hand, using the equation above, we define the IQE [41]: Number of hot electrons/Sec Total absorbed photons/Sec IQE = (2) Figure 6(b) exhibits numerically obtained IQE for the proposed device as a function of the incident UV light, where the peaks for t ox = 2 nm and 4 nm with the values of 38% and 40%, respectively are induced at the FR dip positions.It is also worth noting that at these short wavelengths, generation of hot electrons by metallic nanodisk heptamers has an undeniable impact in having such a large photocurrent as well as a significant IQE.We also calculated corresponding IQE for the UV detector without nanodisk clusters displayed in the profile with dotted curve as 15.6%.A comparison of the performance of all the examined regimes for the proposed UV photodetector shows that inducing the plasmonic effect via aluminum clusters enhances the responsivity and photocurrent of the device with the expense of having a few nano-amperes dark current.Additionally, comparing with the reported IQE for the recent works, 6,11 the proposed plasmonic UV photodetector shows significant efficiency.The response of the plasmonic UV photodetector is comparable with the more complex designs that have been proposed, such as coupling of plasmons between aluminum particles and zinc oxide (ZnO) nanoparticles, or using multilayer substrate to enhance the electron-hole confinement to improve generated photocurrent [6,[57][58][59][60][61][62][63].Finally, we estimated the corresponding gain (Γ ph ) of the investigated photodetector using [64,65]: where q is the elementary charge, and c is the velocity of light.The corresponding gain is found to be Γ ph = 2.1 × 10 2 for the aluminum cluster with the oxide thickness of t ox = 2 nm.

Conclusion
In conclusion, a method is proposed to enhance the photoresponse of a GaN UV detector using plasmon hybridization mechanism.Using aluminum-based symmetric heptamer clusters deposited on a GaN active layer, we developed a structure with enhanced photocurrent due to significant increase in the generation of hot electrons under the heptamer clusters resulted by the decay of the plasmons on the Al disks.We also investigated the effect of oxide layer thickness variations on the characteristics of the UV photodetector.Inducing a pronounced Fano dip in the UV region, we obtained a significant absorption of incoming light power by suppression of the scattering maxima.Calculating the important parameters for the proposed photodetector, we proved its superior performance and quality in comparison to analogous devices without plasmonic structures.Possessing high responsivity, quantum efficiency, internal gain, and significant photocurrent across the UV spectrum make this structure as a potential platform for designing and fabricating optoelectronic UV devices for several sensing applications.

Methods
Definition of the optical response of the proposed device.To extract the optical properties of the proposed UV photodetector, we investigated the excitation of plasmon resonant bright and dark modes and their interference using the finite-difference time-domain (FDTD) method (Lumerical FDTD).In the simulations, to determine the plasmonic responses, following parameters were employed: The spatial cell sizes were set to d x = d y = d z = 0.8 nm, and 48 perfectly matched layers (PMLs) were the boundaries.Additionally, simulation time step was set to the 0.01 fs according to the Courant stability.The light source was a linear plane wave electric source with a pulse length of 2.6533 fs, offset time of 7.5231 fs, and with the illumination power of P = 20 μW/mm 2 .Using the ellipsometric data for Al thin films that were obtained in recent works [18,19], we calculated the dielectric function for the nanostructures composed of a thin coverage of oxide layer using Bruggeman dielectric model for modified Drude model as following: where ɛ ∞ (~2-3) is the high frequency response, ω p (~13.9 eV) is the bulk plasmon frequency, and Γ (~1.2 eV) is the damping constant [19].It should be underlined that our FDTD simulations, the empirical settings and conditions are employed to study the features and plasmon responses of the proposed UV detector [6].Definition of the electrical response of the proposed device.On the other hand, to determine the electrical properties such as responsivity and dark current characteristic (I-V) diagrams, we used fully physics-based Lumerical DEVICE.To this end, we applied finite element mesh (FEM) generation method to solve the electrical properties of the plasmonic UV detector numerically.For the GaN substrate with the presence of metallic electrodes and clusters, the work function is taken as 5.85 eV by following Kim et al. [65] with the dc permittivity of 9.7.Additionally, the bias voltage during the simulations is set as 0.0 V<V G <5.0 V.The majority carriers lifetime (electrons) and diffusion length were gotten from experimentally determined works for n-type GaN epilayers on a sapphire substrate with dislocation density of 10 8 cm −2 , where the recombination lifetime for carriers was set to τ n = 6.5 ns [66][67][68].
Saturation velocity (V sat ) calculation.As a general rule for III-V materials, the bulk saturation velocity in high-field mobility can be defined by modeling as a function of lattice temperature given by [69]: ( ) where V sat is the saturation velocity at the lattice temperature (T = 300 K), and A represents the temperature coefficient, showing the strong dependency of the various materials which are included in the mechanism (Monte Carlo simulations model).Relaxation time estimation.For the examined aluminum antennas, using the following method [70,71]: where R L is the aluminum antenna resistance, and C is the gate-interface capacitance that can be defined as below: ( ) where ɛ 0 is the permittivity of the vacuum., L and W are the pitch and electrode width, respectively.Therefore, for L = 500 nm, W = 200 nm, R L = 250 kΩ, and ɛ GaN = 9.7, the capacitance is calculated as 330 pF.Using the computed values, the relaxation time is defined as 0.825 × 10 −6 s.

Fig. 1 .
Fig. 1. a) Schematic of the plasmonic photodetector composed of aluminum nanodisk clusters deposited on GaN-sapphire substrates.The inset are the definitions for a nanodisk and a heptamer cluster with geometrical parameters, b) a top-view of the photodetector with the geometrical dimensions identification, c) the cross-sectional view of the hot electron generation and transform under the aluminum-based nanodisk clusters at the GaN-metal interface, d) schematic band diagram for the aluminum-GaN interface, showing the carrier formation mechanism in the device.

Fig. 2 .
Fig. 2. a) Scattering and absorption cross-sectional profiles for an aluminum heptamer antenna with the oxide size of 2 nm around nanoparticles for Knight et al. aluminum, b) E-field map of the plasmon resonance excitation and hybridization in the antenna, and formation of energetic hotspots between proximal nanodisks are obvious.

Fig. 3 .
Fig. 3.The plasmon responses for the UV photodetector, a) the absorption spectra for heptamer clusters deposited in GaN epilayer with variant oxide thickness and without metallic heptamers, b) E-field enhancement diagram for the UV device with and without aluminum clusters, c) numerically plotted absorption spectra for the oxide layer thickness as a function of incident UV beam in a heptamer nanocluster.

Figure 3 (
Figure 3(b) represents numerically obtained absorption spectra for Al 2 O 3 thickness variations as a function of incident beam.The absorption ratio increased significantly including a red-shift to the visible spectrum by increasing the thickness and the entire size of the heptamer.The enhancement of the electric field |E| at the gap spots between central nanodisks of the heptamer cluster is shown in Fig. 3(c).Significant enhancements of the

Fig. 5 .
Fig.5.Electrical response for the UV photodetector, A) numerically achieved photocurrentvoltage (I-V) curves for two different oxide thicknesses of a heptamer cluster and without heptamers.The inset is the dark current-voltage (I-V) curves for the non-plasmonic and plasmonic UV photodetector for two different oxide thicknesses at λ = 325 nm and 335 nm for t ox = 2 nm, and 4 nm, respectively, B) polarization-independency of the generated photocurrent (blue-spheres) of the device for the polarization angle variations of the incident UV beam.

Fig. 6 .
Fig.6.The spectral responses for the UV detector in both non-plasmonic and plasmonic regimes, with variant Al 2 O 3 thicknesses of heptamer clusters, A) responsivity profile under 5.0 V applied bias.Inset is the responsivity profile for the non-plasmonic regime, B) internal quantum efficiencies (IQEs) for different regimes of the UV detector.The absorbed power by the structure is given by 261332 Received 17 Mar 2016; revised 21 May 2016; accepted 25 May 2016; published 10 Jun 2016 © 2016 OSA 13 Jun 2016 | Vol. 24, No. 12 | DOI:10.1364/OE.24.013665 | OPTICS EXPRESS 13677