SUPPLEMENTARY INFORMATION

Transcription

1 Silver permittivity used in the simulations Silver permittivity values are obtained from Johnson & Christy s experimental data 31 and are fitted with a spline interpolation in order to estimate the permittivity between the coarse experimental data points for simulations. In Fig. S1, the nominal silver permittivity data with corresponding uncertainties in the real and imaginary part are plotted with upper/lower bounds that are maximized/minimized separately according to the following formulae,, and, 2, where is the complex index of refraction. Fig. S1. Silver permittivity data used in the finite element eigenfrequency analysis. 1

2 To estimate the theoretical Q-factors in the main text, the nominal values of the permittivity (that are denoted by the thick black lines in Fig. S1) are used, if not stated otherwise; and the lower/upper bounds for the theoretical Q-factors are calculated with the maximum/minimum values of the imaginary part (using nominal values of real permittivity). Cavity mode volume V In this section, the cavity mode volume V and the figure of merit Q/V of an SPP microcavity resonator are estimated by using finite element simulations. For the cavity mode volume calculation, the following definition is used with a slight modification noted below 32 : where max corner 1 max 1 corner Re is the electromagnetic energy density for a dispersive media 33 (material dispersions of silver and silica are taken into account in the estimations). The cavity mode volume is calculated for the SPP microcavity of which the bottom/top-radius and thickness of the template silica microdisk resonator are set to 11/7.9 and 2 μm; and the thickness of silver coating is 100 nm. When calculating the cavity mode volume by the definition (1), the integrated energy density is normalized by the maximum energy density excluding the sharp energy density peak at the corner (see Fig. S2). This procedure therefore provides an upper bound V u for the cavity mode volume V. The sharp maximum has a peak energy density that strongly depends on the mesh quality at the corner. Its inclusion would result in uncertainties in the mode volume characterization 34. In addition to this numerical difficulty, any local perturbation at the 2

3 sharp corner of the microdisk resonator, introduced during fabrication, may reduce this strong peak energy density. Hence the cavity mode volume calculated with the ideal sharp corner may not represent the realistic situation. Therefore, as noted, the cavity mode volume defined in equation (1) should be regarded as an upper bound. Fig. S2. Electromagenetic energy density of SPP 1m eigenmode of an SPP microcavity. Figure S3 shows the cavity Q-factor, cavity mode volume V u, and λ 3 Q/V u as a function of resonance wavelength for the same example case. The cavity Q-factor plot in Fig. S3 (a) is adapted from the main text and scaled as a function of wavelength to facilitate the discussions. As shown in Fig. S3 (a), the cavity mode volume tends to decrease as the wavelength becomes smaller. In Fig S3 (b), the figure of merit λ 3 Q/ V u is plotted, where V u is normalized by the cubic free space resonance wavelength λ

4 Fig. S3. (a) Q-factor and mode volume V u, (b) Figure of merit λ 3 Q/ V u of an SPP microcavity. It is also worthwhile to compare the cavity mode volume of the SPP microcavity and the template silica microdisk resonator. Figure S4 shows the cavity mode volume of the fundamental SPP 1m eigenmode for the SPP microcavity and the fundamental eigenmode of the corresponding template silica microcavity. The cavity mode volume of the SPP 1m eigenmode is substantially smaller than the mode volume of the fundamental eigenmode of a wedge-shaped template silica microdisk resonator (note: the fundamental eigenmode of a wedge-shaped microdisk has similar characteristics to the TE eigenmode of a vertical wall microdisk resonator). The calculated reduction factors ( 1,550 nm and ~0.11 at 520 nm, respectively. SPP ) are ~0.36 at 4

5 Fig. S4. Comparison between the fundamental cavity mode volume of the SPP microdisk and its template silica microdisk resonator. Remarks on cavity optimizations The figure of merit, Q/V, can be optimized further with variations in cavity geometry, size, and materials. As an example, in figure S5 the cavity Q-factor, cavity mode volume V u, and Q/V u are plotted as a function of the bottom diameter while all the other geometrical shapes remain unchanged from the previous example in figure S3. Cavity mode volume V u and Q-factor decrease as the diameter becomes smaller, but the degree of reduction in mode volume is larger compared to the decrease in Q-factor (which is primarily due to the increased portion of energy in the silver) and this consequently makes Q/V u increase sharply as the diameter becomes smaller as shown in Fig. S5. (b). The plot is shown for the diameter ranges between 12.5 μm and 40 μm because the radiation caustic comes into the calculation boundary in smaller diameter cavities (the perfectly matched layer was fixed for all the numerical data points for a consistent result). At this specific diameter, the radiation loss becomes comparable to the metallic loss and the SPP microcavity begins to operate in the radiation-limited regime for smaller 5

6 diameters. Although only the effect of cavity diameter variation is considered here, other parameters, such as thickness of silica, thickness of metal coating, or wedge angle, may also be optimized further to achieve higher Q/V values. Fig. S5. Cavity mode volume and Q-factor (upper panel); λ 3 Q/V u and the fraction of energy in the silver coating (lower panel) plotted versus cavity diameter. References 31. Johnson, P. B. & Christy, R. W. Optical constants of noble metals. Phys. Rev. B. 6, (1972). 32. Oulton, R. F., Sorger, V. J., Genov, D. A., Pile, D. F. P. & Zhang, X. A hybrid plasmon waveguide for subwavelength confinement and long-range propagation. Nature Photon. 2, (2008). 6

7 33. Landau, L. D., Lifshitz, E.M. & Pitaevskii, L. P. Electrodynamics of Continuous Media (Butterworth-Heinenann, Oxford, 2002). 34. Miyazaki, H. T. & Kurokawa, Y. Squeezing visible light waves into a 3-nm-thick and 55-nm-long plasmon cavity. Phys. Rev. Lett. 96, (2006). 7