Efficient transmission of light through a metal layer has become a key issue for a variety of applications including light-emitting diodes and solar cells. We report here on a novel strategy where localized and extended surface plasmons are combined to maximize the fluorescence transmission through a metallic film. We show that the dispersion of an artificial material formed by an array of metal nanoparticles coupled to a flat metal layer can be engineered to make the metal film, in a specific direction, 100% transmissive.
© 2007 Optical Society of America
The emergence of “plasmonics” has given rise to several important breakthroughs on control, enhancement and confinement of surface optical fields. In particular, the control of surface plasmons has become increasingly attractive for the miniaturization of interconnect signal carriers , surface enhanced spectroscopy  and sensor technology . Extraordinary transmission through subwavelength apertures in metal films is expected to be a key process which may serve a wide range of applications . More recently, it has been first predicted  and lately experimentally demonstrated that surface plasmons can mediate high spatial harmonics through a homogeneous thin metal film which acts as a superlens able to image with a subwavelength resolution .
Depending on the geometry of the metal, two distinct types of surface plasmons can be identified. Surface Plasmon Polaritons (SPP) sustained at a flat metal/dielectric interface are propagating electromagnetic surface waves associated to a collective oscillation of the free electrons of the metal with the incident electromagnetic field. SPP have been shown to significantly affect the dynamics of a nearby emitter [7–9] providing an alternative deexcitation channel to radiative decay. Since SPP fields are bound to the metal interface, light coupled to the SPP mode remains trapped so that it cannot be measured by a far-field detector. In order to recover this trapped energy, several works have investigated the use of a periodically corrugated metal surface [10–12]. In this case, an appropriate periodicity of the corrugation allows the otherwise non-radiative SPP mode to be coupled out as light into the far field along a direction determined by the general grating diffraction condition. Unlike SPP on flat and extended metal interfaces, Localized Surface Plasmons (LSP) are associated with bound electron plasmas in nano-voids or particles with dimensions much smaller than the incident wavelength. Whilst SPP have a continuous dispersion relation and therefore exist over a wide range of frequencies, LSP resonances only exist over a narrow frequency range owing to additional constraints imposed by their finite dimensions. The spectral position of this resonance is governed by the particles size and shape and by the dielectric functions of both the metal and the surrounding media. As with SPP, the enhanced local field around resonant nanoparticles can significantly increase the fluorescence rate of a nearby molecule provided that it sits at a suitable distance where quenching is negligible [13, 14]. In contrast to SPP, LSP can be directly coupled with propagating light; indeed their enhanced scattering cross-section at resonance makes them very efficient antennas.
In this study, we report on a novel strategy where the attributes of LSP and SPP are combined to maximize the fluorescence extraction through a metallic film. We show that the dispersion of an artificial material formed by an array of metal nanoparticles coupled to a metal layer can be engineered to achieve a strong emission along the normal direction so as to recover that energy which would otherwise be lost to the metal.
In the configuration under study, sketched in Fig. 1(A). a thin emissive polymer film lies between a glass substrate and a thin metallic film. The metal film is coupled to a periodic array of metallic nanoparticles through a thin dielectric spacer. In a previous study, we identified that such LSP/SPP systems feature multiple resonances which can be respectively attributed to the collective LSP resonance of the particles and to the SPP modes of the underlying continuous metal film, excited by the grating nature of the particle array . In reference , electromagnetic interaction between LSP and SPP was shown to contribute to the enhanced transmission through a metal film coupled to a hole array. Since in our configuration, a significant part of the light emitted by the polymer couples directly to the SPP at the metal layer interfaces , a maximum out-coupling to the far-field is expected to be achieved by making the resonances of the array of plasmon particles overlap with those from the film within the emission band of the emissive polymer.
For the emissive layer, we chose a copolymer (MDMO-PPV) emitting at around 600 nm (see Fig. 1(B).). Extinction calculations using the Fourier Modal Method (FMM)  allowed us to determine the structural parameters needed to make the two types of plasmon resonances overlap within the emission band of the MDMO-PPV for normal emission. FMM is a modal method, i.e. where the field is developed on a modal basis in each region of space. The Smatrix is used for satisfying the boundary conditions at the limits between the different media and ensures the numerical stability of the code when the thickness of the grating increases . Finally, one should impose to the diffracted field a radiation condition that is equivalent to the Sommerfeld condition but for the grating geometries. For a silica spacer thickness set to 40 nm, the optimized configuration is found for particles made of gold (125 nm-diameter and 40 nm-high) and arranged in a square array of period, D, 300 nm while the homogeneous film is made of silver (20 nm thick). Figure 1. B shows the experimental extinction curve of the LSP/SPP system for these structural parameters. For reference, the double peak extinction curve for a non-optimum period D=200 nm is displayed in the inset. On the basis of these calculations, the range of relevant nominal parameters for the structures was defined, and different samples were fabricated by e-beam lithography. The samples’ emission was characterized by angle-resolved spectroscopy while MDMO-PPV was pumped under normal incidence through the glass substrate using a 410 nm laser diode. For each angle, a normalization by the emission spectrum of the system without the gold particles was performed allowing the calculation of the enhancement emission factor γ attributed to the contribution of the coupled LSP/SPP system. Figure 2 shows the enhanced emission dispersion diagrams γ=f(ω, k‖) for different sample parameters, when carefully aligning one axis of the arrays within the detection plane. A dramatic influence of the metallic system parameters on the emission diagram is observed.
Maps of Fig. 2(A). and (D). correspond to the experimental parameters, D=300 nm, d=125 nm, for which the LSP and SPP overlap in the normal direction (k//=0) for λ0=610 nm. For p-polarization, the diagram presents the two oblique branches associated with the folding of the SPP dispersion curve within the first Brillouin zone along the ΓX direction. For normal emission these two branches converge to overlap with the LSP band (the weak band independent on k‖seen at the intersection) leading to the maximum enhancement emission. For s-polarization, some emission enhancement (γ> 1) also occurs, almost independently of the angle, within a band centered around λ0=610 nm. We associate this enhancement to the coupling of the LSP mode with the SPP mode through the direction (0, 1) of the grating, which allows SPP emission for this polarization.
To confirm that the maximum fluorescence extraction is achieved through the energy channel engineered between the SPP at the silver film interface and the LSP associated to the particles array and not just due to grating effects, we have also measured the dispersion diagram for arrays of D=275 nm, d=125 nm (Fig. 2(B).). Since the gold particles size has been kept constant, the LSP band remains in the same spectral position; however the folding of the SPP is displaced and now the LSP/SPP(1,0) overlap takes place at k///2π=0.4 µm-1, together with the maximum in the enhancement emission.
This leads to an optimized emission at around 15° with respect to the surface normal. If the array period in considerably reduced (D=200 nm), see Fig. 2(C)., there is no folding of the SPP dispersion curve within the (ω, k‖) window examined. The LSP resonance position is almost unchanged with respect to the previous cases, but surprisingly the particles do not lead to fluorescence extraction enhancement here but instead there is a reduction of the emission below the reference level (γ<1). This effect is attributed to a stronger absorption with respect to scattering associated to a different distribution of the LSP and SPP modes when uncoupled. The contrast between the arrays of D=300 nm (emission enhancement) and D=200 nm (absorption) is clear from the fluorescence images shown in Fig. 2(E). Finally, map of Fig. 2(F). shows the calculated emission dispersion diagram computed for a matrix of D=300 nm, d=125 nm. Due to slight deviations on the considered refractive indexes of the materials and the particles shapes compared to the actual ones, this case is closer to the experimental results obtained for D=275 nm. There is a fair agreement between the calculations and the obtained experimental results: the particle grating couples the SPP mode through the (1,0), (0,1) and (1,1) scattering, but the maximum emission extraction comes from the overlap between LSP and SPP modes.
To gain further insight into the underlying physics, we investigated the influence on the fluorescence emission of the grating period D and the particles diameter d when they are varied around the optimum values predicted by the theory. In Fig. 3(A)., we plot the evolution of the fluorescence emission spectrum in the normal direction with the grating period D (ranging from 250 to 375 nm) for a fixed particle diameter (d=125 nm). As discussed above, by changing the grating period the folding of the SPP dispersion curve is modified so that the SPP resonance at k‖=0 moves across the polymer emissive band.
As a reference, the emission spectrum through the silver layer without the particles is also shown. The normal emission efficiency is extremely sensitive to the period D so that changes as small as 25 nm can have a significant influence. We would like to notice here that the effect of the presence of the grating on the pumping conditions of the polymer was explored through modeling and found to be insignificant. For the optimum period (D=300 nm), the maximum emission intensity is found to be twice that of the reference level. Additionally, we observe that in this case the emission bandwidth is reduced by a factor of two (from 100 nm to 50 nm (FWHM)) compared to the intrinsic emission of the copolymer. Measurements were repeated for a fixed period (D=300 nm) as a function of the particle diameter d (ranging from 115 to 130 nm). While the resonance associated to the SPP at the silver interface is now maintained fixed, it is the LSP resonance of the particles which is spectrally tuned. The results plotted in Fig. 3(B). show the dramatic influence of d, confirming the crucial role played by the intrinsic resonant properties of the metal nanoparticles in assisting the fluorescence emission towards the far-field.
Since the maximum out-coupling of the fluorescence signal is given by the overlap of the LSP and SPP modes, we have also analyzed the influence of the spacing distance between the particles and the metal film. In Fig. 3(C)., we plot the dependence of the maximum enhancement emission with the silica spacer thickness. The change in the SiO2 thickness implies a modification in the effective refractive index seen by the top interface of the silver thin film, and consequently the set of parameters (d, D) for which maximum emission has been obtained varied. In particular we observe that the optimum array period D increases when decreasing the silica spacer thickness. This behavior indicates that in our experiment the copolymer mainly couples to the SPP mode at the top silver/SiO2 interface. Figure 3(C). reveals an optimum value for the spacer thickness around 10 nm. Above this value, the efficiency of the electromagnetic overlap between LSP and SPP decays fast until the two subsystems become independent. Conversely, for shorter spacing distances (below 10 nm) the LSP resonance of the particles gets strongly affected by the metal film [19–21]. This is first indicated by an increasing red-shift of the LSP central band arising from the interaction of the particle with its image dipole (that was experimentally verified by the evolution of d to overlap the emission spectrum of the polymer). Additionally, the particle scattering crosssection decreases due to direct recombination into the metal layer. At contact, the particles behave as non-resonant corrugations isolating therefore the actual contribution of the SPP from that of the LSP. A key requirement in optimizing the extraction of the trapped fluorescence is achieved for an appropriate tradeoff between maintaining a strong LSP resonance and maximizing the coupling with the SPP.
At this stage, we are interested in evaluating the efficiency of the proposed method in recovering the fluorescence power lost when emission takes place through the thin silver layer. For this purpose, we compare in Fig. 4 the normal emission spectrum for different configurations while the SiO2 spacing layer is set close to its optimum value (h=10 nm): (i) bare MDMO-PPV layer, (ii) MDMO-PPV + silver film and (iii) MDMO-PPV + silver film + gold or silica particles (d=125 nm and D=275 nm). For this comparison, the four configurations have been implemented on a single sample. The data show that the optimized LSP/SPP configuration enables one to entirely recover, in the normal direction and for a specified range of wavelengths, the 55 % of lost power through the flat silver film so that the emission level is comparable with the emission level from the bare polymer. For reference, we also show on the same graph the emission from a configuration where SiO2 particles instead of gold ones are used. No significant fluorescence extraction enhancement is achieved in that case. The full compensation of losses suggests that in addition to efficiently extracting part of the fluorescence coupled to the SPP modes, the proposed configuration also contributes in redirecting the emission. In order to verify this hypothesis, we plot in Fig. 5 the directivity of the fluorescence emission (at 610 nm), with and without particles, for both polarizations. For the s-polarization, the emission intensity is weakly increased, independently of the emission direction. Conversely, for the p -polarization a significantly stronger enhancement is observed within a solid angle of approximately 20 degrees centered on the normal axis. In any real device due consideration will have to be given to the emission integrated over the required emission solid angle and any effect of the viewing angle on the perceived color.
Summarizing, we have demonstrated a novel method to maximize light transmission through thin metal layers by coupling of localized and extended surface plasmons. In addition to improve light extraction compared to passive scatterers, the use of resonant metal nanoparticles opens new opportunities to control the emission bandwidth of the system with respect of the intrinsic emission bandwidth of the emissive polymer. This feature provides a novel degree of adjustment of OLED design which could extend their range of applicability.
This research was carried out with the financial support of the European Commission through the NoE “Plasmon Nano-devices” (PND)-FP6-507879. M.U.G. and S.C. acknowledge funding from the Spanish Ministry of Education and Science through the “Ramón y Cajal” program. We thank G. Winter for assistance with the emission measurements.
References and links
1. S. I. Bozhevolnyi, V. S. Volkov, E. Devaux, J. Y. Laluet, and T. W. Ebbesen, “Channel plasmon subwavelength waveguide components including interferometers and ring resonators,” Nature 440, 508–511 (2006). [CrossRef] [PubMed]
3. A. D. McFarland and R. P. V. Duyne, “Single silver nanoparticles as real-time optical sensors with zeptomole sensitivity,” Nano Lett. 3, 1057–1062 (2003). [CrossRef]
7. K. H. Drexhage, Progress in Optics, (Elsevier, 1974) Vol. 12, pp. 63.
8. R. P. Chance, A. Prock, and R. Silbey, Advances in Chemical Physics (Wiley, 1978) Vol. 37, pp 1.
10. R. W. Gruhlke, W. R. Holland, and D. G. Hall, “Surface plasmon cross coupling in molecular fluorescence near a corrugated thin metal film,” Phys. Rev. Lett. 56, 2838–2841 (1986). [CrossRef] [PubMed]
12. J. Feng, T. Okamoto, and S. Kawata, “Highly directional emission via coupled surface-plasmon tunneling from electroluminescence in organic light-emitting devices,” Appl. Phys. Lett. 87, 241109 (2005). [CrossRef]
14. S. Kühn, U. Hakanson, L. Rogobete, and V. Sandoghdar, “Enhancement of Single-Molecule Fluorescence Using a Gold Nanoparticle as an Optical Nanoantenna,” Phys. Rev. Lett. 97, 017402 (2006). [CrossRef] [PubMed]
15. J. Cesario, R. Quidant, G. Badenes, and S. Enoch, “Electromagnetic coupling between a metal nanoparticle grating and a metallic surface,” Opt. Lett. 30, 3404–3406 (2005). [CrossRef]
16. L. Lin, R. J. Reeves, and R. J. Blaikie, “Surface-plasmon-enhanced light transmission through planar metallic films,” Phys. Rev. B. 74, 155407 (2006). [CrossRef]
17. A. L. Li, “New formulation of the Fourier modal method for crossed surface-relief gratings,” J. Opt. Soc. Am. A 14, 2758–2767 (1997). [CrossRef]
18. L. Li, “Formulation and comparison of two recursive matrix algorithms for modeling layered diffraction gratings,” J. Opt. Soc. Am. A 13, 1024 (1996). [CrossRef]
19. A. Christ, T. Zentgraf, S. G. Tikhodeev, N. A. Gippius, J. Kuhl, and H. Giessen, “Controlling the interaction between localized and delocalized surface plasmon modes: Experiment and numerical calculations,” Phys. Rev. B. 74, 155435 (2006). [CrossRef]
21. H. R. Stuart and D. G. Hall, “Enhanced Dipole-Dipole Interaction between Elementary Radiators near a surface,” Phys. Rev. Lett. 80, 5663–5666 (1998). [CrossRef]