Hong, Wang, Yang, and Sang (Plasmonics, 2017) proposed suppressing the scattering of surface plasmon polaritons (SPPs) at sharp bends, waveguide-width adapters, and surface defects by covering a silver surface with a roughly 50-nm-thick isotropic gradient refractive index medium (GRIM) whose in-plane permittivity grade concentrates the SPP against the metal through the perturbed region and then symmetrically releases it; finite-element simulations with lossless silver predicted above 92–94% transmission at 632 nm. The study, however, neglected ohmic absorption in the metal, modeled a single wavelength, assumed a perfect graded profile, and included no experiment. The open problem is to determine the realistic performance envelope of these designs: the net insertion loss when the complex (lossy) Ag permittivity is included, the sensitivity to deviations of the graded profile and to fabrication imperfections, and the usable wavelength bandwidth in the visible. No published follow-up quantifies any of these points; adjacent experimental work on graded dielectric overlays on real metals establishes the feasibility of the general platform but not the performance of this design class.

Background

A surface plasmon polariton (SPP) is an electromagnetic surface wave bound to a metal–dielectric interface, in which photons couple to the quasi-free electrons of the metal. Because an SPP can be confined to lateral dimensions far below the free-space wavelength, plasmonic waveguides have long been studied as a route to compact optical circuitry. The obstacle is that guiding light along a metal at visible frequencies suffers both scattering and absorption: at a sharp bend, an abrupt width transition, or a surface bump, the weakly confined SPP mode radiates into free space and partially reflects, and even on a perfect surface the field loses energy to ohmic (absorption) dissipation in the metal, which at visible wavelengths limits SPP propagation lengths to micrometer scales. Prior theoretical remedies for the scattering part—designs based on transformation optics (the reshaping of field distributions through spatially engineered material tensors) or on zero-index metamaterials (artificial media with near-zero refractive index, in which waves accumulate negligible phase) — typically require anisotropic, spatially dispersive, or resonant metamaterial layers that are difficult to fabricate and integrate at visible wavelengths. The work of Hong, Wang, Yang, and Sang (Plasmonics, 2017) took a different route: it covered the metal surface with a thin isotropic gradient refractive index medium (GRIM) — a dielectric overlay about 50 nm thick whose permittivity varies smoothly in the propagation plane. In their design, two graded sections of half-length L=7.5 μmL=7.5\,\mathrm{\mu m} flank the perturbation, with linear permittivity profiles ε1(x)=1+r(x+L)\varepsilon_1(x)=1+r(x+L) on −L<x<0-L<x<0 and ε2(x)=1+r(L−x)\varepsilon_2(x)=1+r(L-x) on 0<x<L0<x<L, gradient ratio r=1 μm−1r=1\,\mathrm{\mu m}^{-1}, and maximum permittivity εmax=8.5\varepsilon_\mathrm{max}=8.5 at the center; the corner or connection region itself is filled with dielectric of permittivity 8.5. The mechanism rests on the standard SPP evanescent decay depth into the dielectric, d=−(εm+εd)/(εdk0)d=\sqrt{-(\varepsilon_m+\varepsilon_d)}/(\varepsilon_d k_0) with k0=2π/λk_0=2\pi/\lambda: at λ=632 nm\lambda=632\,\mathrm{nm} with silver permittivity εm=−16.24\varepsilon_m=-16.24, dd shrinks from about 390 nm at εd=1\varepsilon_d=1 to a few tens of nanometres at εd≈8.5\varepsilon_d\approx 8.5, comparable to the 50-nm overlay thickness. The rising permittivity therefore pulls the SPP tightly against the metal as it approaches the perturbation ('concentration'), the slow spatial grade approximates impedance matching and suppresses back-reflection, and the falling grade on the far side symmetrically 'releases' the mode. Finite element method (FEM) simulations — numerical solutions of Maxwell's equations on a discretized geometry — with lossless silver then predicted transmission above 92% through bends of 90°, 240°, 270°, and 300° (versus below 20% uncoated), 94% through a 250-nm-long adapter from a 400-nm-wide to a 100-nm-wide plasmonic waveguide (versus about 60% uncoated), and about 94% past a 240-nm bump or groove (versus about 35% uncoated). Three idealizations define the open question. First, silver was modeled by the real permittivity εm=−16.24\varepsilon_m=-16.24 with the imaginary (absorptive) part dropped; at 632.8 nm tabulated optical constants (Johnson and Christy, 1972) give εm≈−16+1.1i\varepsilon_m\approx-16+1.1i, so realistic ohmic loss was never included, even though the total graded path is 2L=15 μm2L=15\,\mathrm{\mu m} and the concentration stage pushes the mode into high-permittivity regions where SPP propagation lengths are shortest — the same field concentration that suppresses scattering increases the fraction of modal energy dissipated in the metal, so the net insertion loss is not obvious a priori. Second, only the single wavelength 632 nm was simulated, leaving the operational bandwidth and the sensitivity to the design parameters (rr, LL, hh, εmax\varepsilon_\mathrm{max}) and to graded-profile fabrication imperfections unquantified. Third, the work is purely numerical: no device was fabricated or measured. For context, the individual ingredients do exist separately: graded dielectric overlays on real, lossy metals have been fabricated by grey-scale lithography (a lithographic process producing a continuously varying film thickness) and demonstrated experimentally as plasmonic Luneburg and Eaton lenses that focus and bend SPPs (Zentgraf et al., 2011), graded-index plasmonic optics has been pursued on chip (Grajower et al., 2013), SPP bends have been demonstrated with real metals in other confinement platforms such as dielectric stripes on gold and V-groove channel plasmon waveguides (Steinberger et al., 2007; Bozhevolnyi et al., 2006), and loss-inclusive bend analyses have been published for other subwavelength plasmonic waveguide classes (Shi and Kochergin, 2009). No published work combines realistic metallic absorption, fabrication tolerances, and bandwidth analysis for the GRIM concentration-and-release designs themselves.

Problem Statement

Determine the realistic performance envelope of the thin-GRIM plasmonic scattering-suppression designs of Hong et al. (2017) at visible wavelengths, for the designs as published: a silver surface covered by an isotropic GRIM overlay of thickness h=50 nmh=50\,\mathrm{nm} with in-plane linear permittivity grades ε1(x)=1+r(x+L)\varepsilon_1(x)=1+r(x+L) on −L<x<0-L<x<0 and ε2(x)=1+r(L−x)\varepsilon_2(x)=1+r(L-x) on 0<x<L0<x<L, with r=1 μm−1r=1\,\mathrm{\mu m}^{-1}, L=7.5 μmL=7.5\,\mathrm{\mu m}, and εmax=8.5\varepsilon_\mathrm{max}=8.5, the corner/connection regions filled with dielectric of permittivity 8.5, for bend angles 90°, 240°, 270°, and 300°, a 400 nm→100 nm width adapter over 250 nm, and a bump/groove of height/depth 240 nm and half-width 350 nm. The problem comprises three linked parts serving one objective — establishing whether the concentration-and-release mechanism provides low net insertion loss in practice rather than only in the lossless idealization: (i) compute the net SPP transmission (insertion loss) of each device when silver is described by realistic complex optical constants (e.g., Johnson–Christy data, εm≈−16+1.1i\varepsilon_m\approx-16+1.1i at 632 nm), with the loss decomposed into ohmic absorption versus residual scattering and reflection, and state whether the above-92–94% lossless-metal transmissions survive; (ii) quantify the sensitivity of the net transmission to deviations of the graded profile (in rr, LL, hh, εmax\varepsilon_\mathrm{max}, and in the profile shape) such as arise from realistic graded-index fabrication, stating the deviation ranges and the profile-shape error model adopted and treating any physically motivated fabrication range as acceptable; and (iii) determine the wavelength range across the visible over which the scattering-suppression benefit persists relative to the stated uncoated-device baselines, given that only 632 nm was simulated. Answers must be quantitative — transmission or insertion-loss figures with stated conventions — so that the outcome decides the practical viability of the design class.

References

  1. Hong X, Wang Y, Yang G, Sang T (2017), 'Scattering-Suppressed Plasmonic Bends and Adapters with Gradient Refractive Index Medium', Plasmonics 13(3), 811–814 (online 14 April 2017; print issue June 2018). DOI: 10.1007/s11468-017-0576-3, https://doi.org/10.1007/s11468-017-0576-3
  2. Zentgraf T, Liu Y, Mikkelsen M, Valentine J, Zhang X (2011), 'Plasmonic Luneburg and Eaton lenses', Nature Nanotechnology 6(3), 151–155. DOI: 10.1038/nnano.2010.282, https://doi.org/10.1038/nnano.2010.282
  3. Grajower M, Lerman G M, Goykhman I, Desiatov B, Yanai A, Smith D R, Levy U (2013), 'Subwavelength plasmonics for graded-index optics on a chip', Optics Letters 38(18), 3492–3495. DOI: 10.1364/OL.38.003492, https://doi.org/10.1364/OL.38.003492
  4. Steinberger B, Hohenau A, Ditlbacher H, Aussenegg F R, Leitner A, Krenn J R (2007), 'Dielectric stripes on gold as surface plasmon waveguides: Bends and directional couplers', Applied Physics Letters 91(8), 081111. DOI: 10.1063/1.2772774, https://doi.org/10.1063/1.2772774
  5. Bozhevolnyi S I, Volkov V S, Devaux E, Laluet J-Y, Ebbesen T W (2006), 'Channel plasmon subwavelength waveguide components including interferometers and ring resonators', Nature 440(7083), 508–511. DOI: 10.1038/nature04594, https://doi.org/10.1038/nature04594
  6. Shi Z, Kochergin V (2009), 'Loss analysis of bends and splitters in subwavelength plasmonic waveguides', Proceedings of SPIE 7221 (Photonics Packaging, Integration, and Interconnects IX), 722114. DOI: 10.1117/12.809490, https://doi.org/10.1117/12.809490
  7. Johnson P B, Christy R W (1972), 'Optical Constants of the Noble Metals', Physical Review B 6(12), 4370–4379. DOI: 10.1103/PhysRevB.6.4370, https://doi.org/10.1103/PhysRevB.6.4370