Nanoscale Single-Element Color Filters - ACS Publications - American

Aug 3, 2015 - ABSTRACT: Visible-light filters constructed from nanostructured materials typically consist ... KEYWORDS: color filter, ZnO, optical fil...
1 downloads 19 Views 3MB Size
Letter pubs.acs.org/NanoLett

Nanoscale Single-Element Color Filters Jerome K. Hyun,† Taehee Kang,‡ Hyeonjun Baek,‡ Dai-sik Kim,‡ and Gyu-chul Yi*,‡ †

Department of Chemistry and Nano Science, Ewha Womans University, Seoul 120-750, Korea Department of Physics and Astronomy, Seoul National University, Seoul 151-742, Korea



S Supporting Information *

ABSTRACT: Visible-light filters constructed from nanostructured materials typically consist of a metallic grating and rely on the excitation of surface plasmon polaritons (SPPs). In order to operate at full efficiency, the number of grating elements needs to be maximized such that light can couple more efficiently to the SPPs through improved diffraction. Such conditions impose a limitation on the compactness of the filter since a larger number of grating elements represents a larger effective size. For emerging applications involving nanoscale transmitters or receivers, a device that can filter localized excitations is highly anticipated but is challenging to realize through grating-type filters. In this work, we present the design of an optical filter operating with a single element, marking a departure from diffractive plasmonic coupling. Our device consists of a ZnO nanorod enclosed by two layers of Ag film. For diffraction-limited light focused on the nanorod, narrow passbands can be realized and tuned via variation of the nanorod diameter across the visible spectrum. The spectral and spatial filtering originates from scattering cancellation localized at the nanorod due to the cavity and nanorod exhibiting opposite effective dipole moments. This ability to realize high-performance optical filtering at the ultimate size introduces intriguing possibilities for nanoscale near-field communication or ultrahigh resolution imaging pixels. KEYWORDS: color filter, ZnO, optical filter, nanoscale filter

A

occurring at the metal interface accumulated along the length of the waveguide. Herein, we design a high-performance, metal− semiconductor hybrid filter with a tunable response in the form of a ZnO nanorod sandwiched between two layers of Ag film and probe its operation in the far-field by exciting the device with focused light. Unlike plasmonic gratings, our design does not rely on the excitation of surface plasmons but rather on scattering cancellation realized by the dipole moment of the Ag coating and ZnO nanorod being out of phase with each other, effectively increasing the transparency of the device10 with respect to the neighboring Ag film. As such, the coupling of light to the device need not be diffractive, enabling the filter to perform reliably even at the single-element level. The transmission efficiency is primarily controlled by the thickness of the two Ag films and can deliver filtered light with excellent spectral purity, whereas its center wavelength can be tuned through the diameter of the ZnO nanorod. Such attributes allow the hybrid device to perform as a tunable optical filter in the visible wavelengths at lateral sizes less than the diffraction limit (∼λ/2). Figure 1a describes the schematic of the hybrid optical filter, consisting of a single ZnO nanorod positioned in between two layers of Ag film. Ag was selected because of the lower loss component (i.e., imag{ε}) it exhibits relative to other metals such as Au, Al, or Pt in the visible range. ZnO was chosen because of its large bandgap energy of ∼3.29 eV,11 enabling

s the drive to miniaturize integrated devices continues, novel designs for compact and high-resolution visible filters have been investigated in order to replace traditional pigment or dye-based filters.1 One popular alternative is the plasmonic grating, in which optical passbands are sculpted by controlling the interference between surface plasmon polaritons (SPPs).2−6 Various designs relying on this concept have led to an assortment of visible color-filters, whose types of grating element range from metal−insulator−metal stacks5 to single Al nanorods.4 In all cases, a larger number of grating elements results in improved transmission efficiencies and narrower line widths due to improved momentum matching between light and SPPs. Although these types of filters offer several key advantages over traditional filters, their response is spatially delocalized, resulting in poor transmission of localized nearfield excitations. For instance, emission from a single molecule or nanoscale antenna attached to a grating-type filter would not be filtered effectively because of its limited spatial extension. In this context, a nanoscale single-element device that does not rely on diffractive SPP coupling and that can locally filter light with high efficiency is an attractive and interesting concept, especially for near-field communication in the nanoscale. Several examples of filters that do not rely on gratings have been reported. A 3D-structured hole in a metal film7 and a biaxial metallic element8 were shown to exhibit plasmonically induced spectral filtering properties, albeit with broad spectral widths due to challenges in overcoming metallic losses with single structures. Metal−insulator−metal waveguide structures based on the interference between cavity modes also showed spectral filtering properties.9 However, in this design, the transmission intensity can suffer due to propagation losses © XXXX American Chemical Society

Received: May 25, 2015 Revised: July 29, 2015

A

DOI: 10.1021/acs.nanolett.5b02049 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

illustrated in Figure 1c. Visible light (400−700 nm) sourced by a supercontinuum laser (Fianium Inc.) and polarized along the nanorod axis is focused onto the device through a 0.8NA objective lens, producing an excitation spot size ranging between 1 and 2 μm in size. Our choice of polarization along the nanorod axis precludes the excitation of surface plasmons because of the absence of electric field components normal to the Ag surface. The transmitted light is then collected through a separate 0.8NA objective lens positioned after the device and is monitored by a spectrometer. The theoretical far field transmission efficiency generated from our setup was predicted using a commercial finite difference time domain (FDTD) solver (Lumerical, Inc.). A 2D or 3D model was employed to describe the system. Although both 2D and 3D models produce equivalent transmission results for a plane wave incident on an infinitely long nanowire, the 2D model overestimates the 3D transmission intensity by roughly ∼27% for a Gaussian beam (Supporting Information Figure S3). Despite this overestimate, the two simulations share near-equivalent spectral responses, enabling relevant calculations to be performed in 2D to avoid the large computational demands accompanying fine 3D meshing. For comparisons to experimentally obtained intensity values, we employ 3D simulations to accurately model the system. Figure 1d presents the far-field transmission efficiency calculated in 3D as a function of wavelength and nominal core diameter. The plot demonstrates that the transmission is both filtered and tunable across the visible wavelengths as a function of core diameters. The transmission efficiencies generated and collected by the two 0.8NA objective lenses are shown to reach values between 25 and 40%. As the transmission efficiency is variable upon the optics of both the excitation and collection, we present the dependence of the maximum transmission on numerical aperture of the exciting and collecting lens in Supporting Information Figure S4, where maximum transmission values can reach between 30 to 40% for different hybrid nanorods. Other type of illuminating optics such as a cylindrical lens can produce a 2D Gaussian beam, often used in commercial Raman systems, resulting in higher transmission efficiencies up to 50% (Supporting Information Figure S3). We note that these theoretical values represent upper limits to the achievable efficiencies as they do not include the contribution of the Ge layer. Because Ge absorbs strongly, a 1 nm-thick layer of Ge can in fact reduce the maximum far-field efficiencies by 10% when using focused illumination and collection through 0.8NA objective lenses (Supporting Information Figure S5). Different Ag wetting materials such as Al-doped Ag alloys have been reported to exhibit negligible absorption in the visible range, suggesting alternative pathways toward achieving smooth Ag films without sacrificing efficiency.14,15 The role of the top and bottom Ag layers is further emphasized by observing the power transfer into the underlying quartz without either one of the layers, as displayed in Supporting Information Figure S6. In both cases, the filtering functionality is largely absent or significantly degraded compared to when both layers are present as shown in Figure 1d. From this, we can deduce that each Ag layer serves two main purposes that permit the hybrid nanorod to perform as a spatial and spectral filter. One is the shielding of light leaking through all areas of the Ag film that excludes the hybrid nanorod. The other is the formation of an effective Ag-cavity encompassing the ZnO nanorod, which gives rise to a spatially

Figure 1. (a) Schematic of the single-element optical filter consisting of a ZnO nanorod positioned between two Ag layers whose top and bottom thicknesses are 30 and 15 ± 5 nm, respectively. (b) SEM image of a representative hybrid nanorod filter. Scale bar is 1 μm. (c) Setup for measuring the far-field transmission through the hybrid nanorod. (d) Numerically simulated transmission as a function of diameter and wavelength by modeling the measurement setup in 3D.

visible light to be efficiently confined through the real part of its dielectric function, and loss to be minimized through the small imaginary part. We note that other large bandgap material such as GaN can also be used, but due to the smaller loss component, ZnO was found to be a superior material for this study. The hybrid system was constructed by first drop-casting ultrahigh purity ZnO nanorods, grown through the MOVPE process11 and exhibiting diameters between 100 and 200 nm, onto a Ag surface prepared on top of a thin Ge layer wetting a quartz substrate. The Ge wetting layer serves to reduce the roughness of the Ag layer.12,13 Once the geometrical dimensions of the ZnO nanorods were measured through SEM, a separate layer of Ag was deposited on top of the ZnO nanorods, forming the top Ag-layer (See Supporting Information Figure S1). Herein, the top and bottom Ag-layer thicknesses were numerically determined by finding the optimal combination which maximized the filtered transmission efficiency. As excessively thin layers of Ag with a total thickness less than the skin depth result in light leaking through the background, whereas thick layers prohibit light from entering and leaving the nanorod (See Supporting Information Figure S2), we targeted an optimum thickness combination of 30 and 15 ± 5 nm for the top and bottom layers, respectively. Figure 1b displays an SEM image of a representative Agcoated ZnO nanorod. Due to the top half of the nanorod shielding the bottom half from the metal evaporation source, the Ag coverage does not fully encompass the ZnO nanorod. Yet, this half-cavity enables incoming light to undergo a strong resonance within the ZnO nanorod and transmit through the bottom Ag layer by providing sufficient scattering cancellation and optical confinement as will be discussed further below. We gauge the performance of our hybrid device in the far-field by employing a spatially localized excitation using the setup B

DOI: 10.1021/acs.nanolett.5b02049 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 2. Mie scattering for core−shell (top) and bare (bottom) infinite cylinders (a) Schematic of sections of a core−shell ZnO−Ag cylinder (top) and a bare ZnO cylinder (bottom). The Ag shell thickness is fixed at 30 nm. (b) Analytically calculated scattering cross sections as a function of diameter and wavelength (c) Scattering (black curve) and absorption cross section (blue curve) as a function of wavelength for a core-diameter of 180 nm. (d) Normalized angular scattering distribution for a core-diameter of 180 nm and at a wavelength of 591 nm, corresponding to the fundamental absorption resonance in the core−shell cylinder. The angular variable, θ, is defined in (a).

localized filtering mechanism based on generation of Mie resonances, detailed further below. In order to explore the operational mechanism of our optical filter, we isolate the hybrid nanorod from the surrounding Ag film by considering a simplified model in the form of a core− shell ZnO-Ag cylinder extending infinitely into and out of the page. Such a model allows us to treat the problem analytically using Mie scattering theory and achieve a more intuitive guide to understanding the interaction between light and the hybrid nanorod. Similar approaches have been used to describe enhanced Raman signals in dielectric-coated Ge nanowires.16 Figure 2a illustrates the schematic diagram of the model where the top panel shows a ZnO cylinder enclosed by a 30 nm-thick Ag shell and the bottom panel describes a bare ZnO cylinder. Light is incident normal to the nanorod axis and is polarized along the NR axis. We first calculate analytically the scattering cross sections for the core−shell and bare cylinder over a range of core diameters and wavelengths as displayed in Figure 2b. For both core−shell and bare cases, the scattering is found to be governed by the dipolar (m = 0) scattering modes in the wavelength and diameter range of interest. For the core−shell cylinder, the scattering undergoes a noticeable decrease at specific diameters and wavelengths, shown as dark bands in the plot. These features exhibit trends similar to the transmission characteristics shown in Figure 1d, but with inverted intensities. On the other hand, for the bare cylinder, no such bands are observed and the scattering cross sections tend to increase for larger diameters. The contrasting behaviors between bare and core−shell cylinders suggest that the Ag−shell assists in reducing the scattering at specific wavelengths and diameters, thereby allowing light to transmit undisturbed into the forward direction. This phenomenon was elucidated in a previous study by Alu et al., in which the metal shell was presented as a path to achieving transparency in small-diameter nanowires.10 The scheme was further demonstrated by Fan et al., in the form of an invisible photodetector comprising a Au-coated Si nanowire.17 In line with these previous studies, the reduced scattering in the hybrid ZnO nanorod can be traced to the presence of opposing dipoles in the Ag shell and ZnO core,

resulting in enhanced transparency. We note that to achieve stronger scattering cancellation, a thinner Ag shell with a 10 nm thickness suffices (See Supporting Information Figure S7). However, when applied to the device platform illustrated in Figure 1a, this choice results in an effective background thickness of ∼20 nm, which is insufficient to shield transmission through the background. Furthermore, a thin shell results in a wider spectral line width, producing poorer spectral purity. By increasing the shell thickness to 30 nm, the spectral purity improves, but with an associated reduction in absolute scattering cancellation. An equivalent interpretation for the enhanced transmission in the hybrid device can be found by monitoring the absorption cross sections for the core−shell cylinder. Supporting Information Figure S8 illustrates the analytically calculated absorption cross section for the core−shell and bare cylinders as a function of wavelength and diameter. We first observe that the absorption resonances, also known as leaky-mode resonances,18 are highly enhanced by the Ag shell and share nearly identical trends with the scattering valleys of Figure 2b. Figure 2c presents a clearer picture by displaying simultaneously the analytical scattering and absorption cross sections of a core−shell and bare cylinder from a 180 nm-diameter core as a function of wavelength. For the core−shell cylinder, the scattering is observed to be reduced when the absorption is peaked near a resonance of 590 nm, whereas such behavior is absent for the case of the solid cylinder. The close correlation between enhanced transmission and absorption is akin to behavior found in dye-filled holes in metal films, termed absorption-induced transparency,19 whose mechanism was theoretically clarified by S.G. Rodrigo et al.20 Here, the authors20 demonstrated that the simultaneous peaking of absorption and transmission was caused by the reduction in evanescent light propagating through the dye-filled hole as a consequence of the change in effective permittivity at absorption resonance. Although our hybrid geometry in Figure 1a does not represent an absorbent-filled hole as described in the aforementioned studies, the mechanism is similar in that the transmission is peaked spatially at the absorbing target and spectrally at its absorption resonance. Figure 2d presents the C

DOI: 10.1021/acs.nanolett.5b02049 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 3. (a) Calculated power transfer efficiency from a 2D model across the Ag film for focused light incident on a hybrid ZnO nanorod (red curve) and a bare Ag film (black curve). The hybrid ZnO nanorod consists of a 180 nm-sized core diameter, 30 nm-thick top Ag layer and 15 nmthick bottom Ag layer. The bare Ag film is 45 nm thick. (b−c) Real part of the Poynting vector along the y direction describing the interaction between light focused on the hybrid nanorod (b) and bare Ag film (c). Panels 1, 2, and 3 represent the interaction occurring at 475, 612, and 750 nm, respectively, as denoted by the arrows in (a).

Figure 4. (a) SEM images of three bare ZnO nanorods on Ag film, before deposition of the second Ag layer. Left and right panels represent magnified and full-view images of each nanorod. Nominal diameters correspond to 186, 156, and 125 ± 6 nm, from top to bottom, respectively. Left and right scale bars are 250 and 500 nm, respectively. (b) Optical microscopy images of the three nanorods after depositing the second layer of Ag. Scale bar is 2.5 μm. (c) Measured (top) and numerically simulated (bottom) transmission through the hybrid device normalized by transmission through the neighboring Ag film for the three hybrid nanorods. The calculation is generated from a 3D model. (d) Experimental and calculated wavelength-diameter dispersion from different hybrid nanorods. Exp1 and Exp2 correspond to two sets of hybrid nanorods each prepared separately. (e) (top) SEM image of a highly tapered ZnO nanorod on Ag film, before deposition of the second Ag layer. Scale bar is 500 nm. (bottom) Optical microscopy image of the same ZnO nanorod after depositing the second layer of Ag. Scale bar is 2.5 μm.

curve) and that of the bare Ag film constituting the background (black curve). A clear peak is observed at 612 nm for the hybrid nanorod whereas a monotonic decay is observed for the bare Ag film. Figure 3b and 3c monitor the real part of the Poynting vector along the y direction in the hybrid nanorod and bare Ag film, respectively, at selected wavelengths indicated by the numbered arrows in Figure 3a. In the case of the bare Ag film, Figure 3c shows that power cannot be transferred across the Ag film at all three wavelengths of interest. When the ZnO nanorod is inserted in between the two Ag layers, the power distribution is significantly altered as shown in Figure 3b. For panels 1 and 3, light is unable to penetrate efficiently through the hybrid device. However, near resonance, labeled 2, light can leak through the nanorod as the nanorod appears more transparent compared to the background,

normalized angular distribution of the scattered light at the absorption resonance. Comparison of the angular scattering between the core−shell and bare cylinder shown in the top and bottom panels, respectively, show that the presence of the Ag shell indeed helps with directing scattered light into the forward direction. Having established the basic mechanism for the enhanced transmission at resonance through the simple core−shell model, we proceed to describe the filtering in our actual hybrid device. To calculate the fraction of power transferred across the Ag film, we employ a 2D description of the model in Figure 1a and inject a Gaussian beam generated through a 0.8NA thin lens. Figure 3a presents the numerically simulated 2D power transfer efficiency for the hybrid device consisting of a ZnO core exhibiting a nominal diameter of 180 nm (red D

DOI: 10.1021/acs.nanolett.5b02049 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

size at 1 to 2 μm was close to 3 to 4 times larger than the Abbe diffraction limit (∼0.6λ/NA) used in the model. This near 3- to 4-fold increase in spot-size results in a 3- to 4-fold decrease in transmission intensity, whereas the remaining difference in intensity between calculation and measurement can be accounted by inhomogeneities along the nanorod axis and surface roughness of the Ag. We also suggest the possibility of encoding different colors into a single element by exploiting differences in diameter of the nanorod. Figure 4e shows a highly tapered ZnO nanorod with a tip diameter of 40 and base of 169 ± 6 nm. When covered with a second Ag layer and excited by light polarized parallel to the nanorod, the device filters an assortment of colors according to its wavelength-diameter dispersion. For incident light polarized perpendicular to the nanorod axis, the visible color contrast is reduced for each nanorod due to the shift in the peak position toward the near-infrared regime (See Supporting Information Figure S12). The appearance of a response in the near-IR is expected from consideration of an isolated core−shell cylinder where scattering is reduced, as shown in Supporting Information Figure S13. The absolute spectral positions of the response are inconsistent with the measurement due to differences in shell geometry between model and actual system. The change in polarization also enhances the sensitivity of the metal−nanorod interface, which can lead to increased loss21 and further reduction of the transmission. For practical applications, we suggest the possibility of designing the nanowires through lithographic processes. An important criterion for successful implementation of such hybrid filters is the robustness of center wavelength against fabrication errors in width or height of the nanorod. One way to reduce the filter’s sensitivity to such variations is to increase the slope of the wavelength-diameter dispersion relation shown in Figure 4d. A smaller real part of the dielectric function permits an increase in slope, albeit at the expense of lower optical confinement, and can be achieved through doping the ZnO nanorods with Al or Ga.22,23 We finally note that although nanorods were used in our demonstration, single nanodiscs could also be implemented, ensuring improved resolution in both lateral directions. However, this comes at the expense of reduced transmission because the nanodisc presents an effective cross section that is limited in both lateral dimensions. In summary, we have demonstrated a simple design in which light can be filtered and tuned over wavelength through the use of a single nanoscale element in the form of a ZnO nanorod integrated with a Ag cavity. The principle behind the transmission originates from the ability of the Ag cavity to highly concentrate light into the nanorod and create strong absorption resonances, or equivalently, the reduction in scattered light due to the Ag cavity. Because the Ag coverage only needs to occur above and below the nanorod and need not be uniformly concentric around the nanorod, fabrication constraints can be relaxed and integration of the device into on-chip platforms can be facilitated. In contrast to plasmonic grating type filters whose spectral response degrades significantly when the number of grating elements reduces to unity, the hybrid nanorod exhibits excellent spectral purity at the single element level. This allows the hybrid nanorod to perform at the ultimate size of optical filters and opens up a myriad of potential applications such as the construction of nanoscale communication devices and ultrahigh resolution pixel arrays.

resulting in the effective transfer of power across the Ag barrier. Such transfer is mediated by the confinement of electric field at resonance, as displayed in Supporting Information Figure S9. By observing the concentrated field distribution at resonance, one can also confirm that the mechanism is not governed by localized surface plasmons as the fields do not occur at the metal interface. In order to verify the ability to filter light, and the tunability of the filtered response, we measured the transmission through three hybrid nanorods with distinct diameters that correspond to outputs in the blue, green and red wavelengths. Figure 4a shows SEM images of the three ZnO nanorods in the order of decreasing nominal diameter from top to bottom. The images were taken before depositing the second Ag layer in order to obtain an accurate measure of the nominal diameters. The left panel corresponds to a magnified scan of the respective nanorod whose full form is shown in the right panel, thereby providing a clearer view of the change in diameters from top to bottom panel. Although some nanorods exhibited inhomogeneities in diameter along the length, the optical characteristics were largely dominated by regions exhibiting smallest variation, as shown in the left panel images. The nanorods showed diameters of 186, 156, and 125 ± 6 nm, from top to bottom, respectively. From transmission measurements, the top and bottom Ag film thicknesses were deduced to be near 38.5 ± 2.5 and 17 nm, respectively (See Supporting Information Figure S10). Figure 4b displays optical microscopy images of the three hybrid nanorods illuminated with light polarized along the nanorod axis. Indeed, each hybrid nanorod transmits at distinct wavelengths, corresponding to red, green, and blue outputs, from top to bottom, respectively. Figure 4c displays the measured and calculated transmission response for each nanorod, wherein the measurement was conducted using the setup shown in Figure 1d. Here, we present the transmission through the hybrid device normalized by the transmission through the neighboring bare Ag film. In all cases, peaks with values above unity are observed, confirming that the hybrid nanorod appears more transparent at its absorption resonance. Also, an increase in nominal diameter is manifested in the redshifting of the transmitted peak, whose position is in excellent agreement with the numerically simulated profile, shown in the bottom panel, generated with a 3D model involving the experimentally obtained core diameter, Ag thicknesses and Ge film thicknesses as input parameters. The tunability of the transmission was further confirmed by identifying distinct peak wavelengths from hybrid nanorods of varying diameters and observing their agreement with the wavelength-diameter dispersion relation extracted from simulation as shown in Figure 4d. Herein, two sets of hybrid filters each prepared with a different top Ag thickness are plotted. The first set corresponds to the three measurements shown in Figure 4c. The second set exhibited a top Ag thickness that was under our targeted value, resulting in red-shifted peak wavelengths, consistent with predicted and experimentally confirmed trends in the wavelength−Ag thickness relations described in Supporting Information Figure S2 and S11. The calculated dispersion is represented as a band to portray the uncertainty (i.e., ± 2.5 nm) in thickness of the top Ag layer. We note that although the peak positions follow the predicted linear trend, the measured and predicted peak intensities are largely different from one another as can be verified in Figure 4c. The large discrepancy is mostly attributed to the larger spot size of the focused light used in the measurement. Our focused beam spot E

DOI: 10.1021/acs.nanolett.5b02049 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters



(14) Lee, K. T.; Lee, J. Y.; Seo, S. Y.; Guo, L. J. Light: Sci. Appl. 2014, 3, e215. (15) Zhang, C.; Zhao, D. W.; Gu, D. E.; Kim, H.; Ling, T.; Wu, Y. K. R.; Guo, L. J. Adv. Mater. 2014, 26, 5696−5701. (16) Hyun, J. K.; Kim, I. S.; Connell, J. G.; Lauhon, L. J. Opt. Express 2012, 20, 5127−5132. (17) Fan, P. Y.; Chettiar, U. K.; Cao, L. Y.; Afshinmanesh, F.; Engheta, N.; Brongersma, M. L. Nat. Photonics 2012, 6, 380−385. (18) Cao, L. Y.; White, J. S.; Park, J. S.; Schuller, J. A.; Clemens, B. M.; Brongersma, M. L. Nat. Mater. 2009, 8, 643−647. (19) Hutchison, J. A.; O’Carroll, D. M.; Schwartz, T.; Genet, C.; Ebbesen, T. W. Angew. Chem., Int. Ed. 2011, 50, 2085−2089. (20) Rodrigo, S. G.; Garcia-Vidal, F. J.; Martin-Moreno, L. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 88, 155126. (21) Hyun, J. K.; Lauhon, L. J. Nano Lett. 2011, 11, 2731−2734. (22) Kim, J.; Naik, G. V.; Gavrilenko, A. V.; Dondapati, K.; Gavrilenko, V. I.; Prokes, S. M.; Glembocki, O. J.; Shalaev, V. M.; Boltasseva, A. Phys. Rev. X 2013, 3, 041037. (23) Naik, G. V.; Liu, J. J.; Kildishev, A. V.; Shalaev, V. M.; Boltasseva, A. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 8834−8838.

ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.5b02049. Details on sample fabrication, calculations, transmission measurement, dependence of power transfer on Ag thickness, effect of Ge wetting layer, far-field transmission as a function of excitation and collection NA, performance of individual Ag layers, Mie absorption cross sections of ZnO core−shell and core cylinders, electricfield distribution inside hybrid filter, experimental transmission spectra of the first and two layers of Ag, experimentally determined dispersion for different top Ag layer thicknesses, and transmission measured with light polarized perpendicular to nanorod axis. (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was financially supported by Future-based Technology Development Program (Nano Fields) through the National Research Foundation (NRF) of Korea funded by the Ministry of Education, Science and Technology (MEST) (No. 2014M3A7B4051589) and the SNU-Yonsei Research Cooperation Program through Seoul National University (SNU) in 2014. Taehee Kang and Daisik Kim acknowledge the support by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (No. 2008-0061906, 2005-0093838, 2008-00580) and the Brain Korea 21 Plus Project in 2014. Jerome Hyun acknowledges support from the TJ Park Science Fellowship through the POSCO TJ Park Foundation and the Ewha Womans University Research Grant of 2015.



REFERENCES

(1) Sabnis, R. W. Displays 1999, 20, 119−129. (2) Barnes, W. L.; Dereux, A.; Ebbesen, T. W. Nature 2003, 424, 824−830. (3) Lee, H. S.; Yoon, Y. T.; Lee, S. S.; Kim, S. H.; Lee, K. D. Opt. Express 2007, 15, 15457−15463. (4) Olson, J.; Manjavacas, A.; Liu, L. F.; Chang, W. S.; Foerster, B.; King, N. S.; Knight, M. W.; Nordlander, P.; Halas, N. J.; Link, S. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 14348−14353. (5) Xu, T.; Wu, Y. K.; Luo, X. G.; Guo, L. J. Nat. Commun. 2010, 1, 59. (6) Zeng, B. B.; Gao, Y. K.; Bartoli, F. J. Sci. Rep. 2013, 3, 2840. (7) Li, W. D.; Hu, J.; Chou, S. Y. Opt. Express 2011, 19, 21098− 21108. (8) Goh, X. M.; Zheng, Y. H.; Tan, S. J.; Zhang, L.; Kumar, K.; Qiu, C. W.; Yang, J. K. W. Nat. Commun. 2014, 5, 5361. (9) Diest, K.; Dionne, J. A.; Spain, M.; Atwater, H. A. Nano Lett. 2009, 9, 2579−2583. (10) Alu, A.; Engheta, N. Phys. Rev. E 2005, 72, 016623. (11) Park, W. I.; Kim, D. H.; Jung, S. W.; Yi, G. C. Appl. Phys. Lett. 2002, 80, 4232−4234. (12) Chen, W. Q.; Thoreson, M. D.; Ishii, S.; Kildishev, A. V.; Shalaev, V. M. Opt. Express 2010, 18, 5124−5134. (13) Vj, L.; Kobayashi, N. P.; Islam, M. S.; Wu, W.; Chaturvedi, P.; Fang, N. X.; Wang, S. Y.; Williams, R. S. Nano Lett. 2009, 9, 178−182. F

DOI: 10.1021/acs.nanolett.5b02049 Nano Lett. XXXX, XXX, XXX−XXX