Large spontaneous emission enhancement in plasmonic nanocavities

Cavity–emitter coupling can enable a host of potential applications in quantum optics, from low-threshold lasers to brighter single-photon sources for quantum cryptography1. Although some of the first demonstrations of spontaneous emission modification occurred in metallic structures2,3, it was only after the recent demonstration of cavity quantum electrodynamics effects in dielectric optical cavities4 that metal-based optical cavities were considered for quantum optics applications5–13. Advantages of metal–optical cavities include their compatibility with a large variety of emitters and their broadband cavity spectra, which enable enhancement of spectrally broad emitters. Here, we demonstrate radiative emission rate enhancements approaching 1,000 for emitters coupled to the nanoscale gap between a silver nanowire and a silver substrate. A quantitative comparison of our results with analytical theory shows that the enhanced emission rate of gap-mode plasmons in our structures can yield high internal quantum efficiency despite the close proximity of metal surfaces. In dielectric optical cavities, the signature of emission enhancement is unambiguous—changes in the radiative emission rate nR are correlated with proportional changes in the emission intensity. However, metal structures are inherently lossy and can suffer from high rates of non-radiative recombination (nNR). In addition, metallic structures can act as antennas to alter the efficiency of excitation and collection14–16. An ambiguous situation can therefore arise in which antenna effects increase the emission intensity while loss increases the total decay rate nT1⁄4 nRþ nNR. The experiments described here are designed to demonstrate unambiguously the large spontaneous emission enhancement capabilities of gap-mode plasmonic nanostructures. The enhancementwe observe is significantly larger than that seen in dielectric cavities17, but is comparable to the field enhancements in surface-enhanced Raman scattering from structures similar to the one reported here18. Our cavity design (Fig. 1) was based on surface plasmon coupling between a silver substrate and a silver nanowire lying parallel to the substrate19. This cavity geometry was chosen to enable a high degree of control over the spacing dG between the nanowire and substrate, which was established by coating the substrate in thin uniform layers of Al2O3 and the fluorescent organic dye tris-(8-hydroxyquinoline) aluminium (Alq3). This design ensures that the dye will be located at the high-field regions of the cavity modes (Fig. 1, inset). We fabricated six cavity structures with different gap spacings dG ranging from 5 to 25 nm. These samples contained identical layers of Alq3 (thickness, 2.5 nm) but different thicknesses of Al2O3 spacer layer. Fluorescence spectra from individual cavities show a strong modification from the uncoupled Alq3 spectrum (Fig. 2a). The wavelengths of the peaks in the cavity spectra have been shown to correspond with the cavity resonances19, establishing that cavity– emitter coupling can modify the spectral properties of the emission. The dynamics of the fluorescence are also greatly modified by coupling to a cavity, with an enhancement of both decay rate and emission intensity (Fig. 2b). Fits indicate that the fluorescence decay is governed by a distribution of decay rates, which we interpret as different degrees of coupling to the cavity (Supplementary Fig. S1 and Section S1). Our analysis will focus on the fastest fitted decay rate (termed the total decay rate, nT) and its dependence on the thickness of the Al2O3 spacer layer. This ultimately yields a quantitative characterization of the emission enhancement characteristics of our structures. Ideally, the incorporated dye absorbs energy from an incident laser pulse and then relaxes back to its ground state by coupling to the states of the cavity (referred to as ‘gap modes’), leading to an enhanced radiative emission rate nR. However, there will also be coupling to the metal layers, which provides a non-radiative (lossy) pathway characterized by a non-radiative decay rate nNR1. Finally, intrinsic defects in the Alq3 film itself lead to non-radiative decay with a characteristic rate nNR2 (Supplementary Section S3). Both nNR1 and nNR2 contribute to nNR1⁄4 nNR1þ nNR2, but in our case nNR1≫ nNR2 (Fig. 3), and we can therefore use the approximation nNR≈ nNR1 in our analysis. The important outcome of our experiment is that the relative influence of nR and nNR changes substantially with gap thickness. Analytical calculations indicate Nanowire