To achieve the bright and polarized quantum light sources with enhanced emission rates via the Purcell effect, we introduce a metasurface based on Si nanoantennas with Fano resonance, composed of a pair of ellipses and a disk in the middle, as shown in Fig. 1a. The Si ellipses have a length (l) of 176 nm, a width (w) of 64 nm, and is rotated 48° with respect to the x-axis, thus introducing asymmetry for achieving the quasi-BIC resonances.34 The disk with diameter (d) of 120 nm is displaced 75 nm in the positive direction of the y-axis direction. The height (h) of the nanostructure was designed to be fixed at 240 nm, with a pitch size of 327 and 324 nm along x and y axis, respectively. Here, a dip-coating method was carried out to deposit the NDs onto the metasurface.
To elaborate the mechanism of the Fano resonance, the simulated reflectance spectrum under y-polarized incident light and the corresponding electric field intensity distributions were shown in Figs. 1b and 1c, respectively. The quality factor of resonance at 671 nm was evaluated to be ~ 200, while the measured reflectance spectrum is shown in the supplementary information (Fig. S1). The resonance can be further explained by the electric field intensity changes in Fig. 1c and magnetic field intensity distribution (see SI, Fig. S2). At the dip of the reflectance spectra (indicated by the star symbol in Fig. 1b), the strongest electric field intensity occurs at the nanogaps between disk and two ellipses, also called hotspot, where an electric intensity enhancement factor (|E|2/|E0|2) of 230 is obtained. When an in-plane electric dipole (oriented as indicated by the black arrow in the inset of Fig. 1d) is placed in the nanogap, the sharp resonance results in a high simulated Purcell factor of up to 23 (Fig. 1d). The position of ND with NV− center emitter is confirmed by SEM image shown in Fig. 1e, where a single ND with a diameter of 35 nm is clearly visible at the nanogap.
Figure 1f presents the benchmarking summary of our NV− SPEs integrated with dielectric nanoantenna being compared with respect to previous experimental works in terms of lifetime and Purcell factor.6, 27, 35–38 To the best of our knowledge, our work outperforms over the other dielectric nanoantenna enabled NV center-based SPEs in terms of emission lifetime reduction, highlighting the significance of achieving high Purcell factor when using the Fano resonance.
Figure 2 presents the basic principle of mode hybridization for the designed dielectric Fano nanoantenna. Here, we consider the dielectric material to be lossless Si with a refractive index of 3.74, thereby eliminating the influence of dispersion and Ohmic loss. As shown in Fig. 2a, both the disk and ellipse pair exhibit transverse magnetic modes (i.e., Hz is negligible). The disk supports a non-radiative electric dipole oriented along z-axis at the wavelength of 561 nm, denoted as \(\:\left|{\varvec{p}}_{\varvec{z}}⟩\right.\), generating an in-plane magnetic loop with the magnetic field H circulating in the xy-plane (Fig. 2b). The ellipses pair with a radiative quasi-BIC mode induces two in-plane magnetic dipoles \(\:\left|{\varvec{m}}_{\parallel\:}⟩\right.\) at the wavelength of 660 nm (Fig. 2c). The x-components of two magnetic dipoles \(\:\left|{\varvec{m}}_{\parallel\:}⟩\right.\) are aligned in the same direction, thereby reinforcing each other along the x-axis; meanwhile, the ycomponents of magnetic dipoles are oriented in opposite directions, thus cancelling each other out. In other words, this quasi-BIC mode can be readily excited when the incident optical field is y-polarized with a dominating magnetic field being x-polarized.
When integrating an ellipse pair and a disk in a single unit cell, two hybrid modes are generated at 534 and 671 nm, respectively, due to the coupling between \(\:\left|{\varvec{p}}_{\varvec{z}}⟩\right.\) and \(\:\left|{\varvec{m}}_{\parallel\:}⟩\right.\). To be more precise, the upper hybridized mode (\(\:\left|{\varvec{m}}_{\parallel\:},-{\varvec{p}}_{\varvec{z}}⟩\right.\)) arises from the coupling between \(\:\left|{\varvec{m}}_{\parallel\:}⟩\right.\) along + x axis and \(\:\left|{\varvec{p}}_{\varvec{z}}⟩\right.\), where its corresponding electric and magnetic field distributions are shown in Fig. S3. On the other hand, the lower one (\(\:\left|{\varvec{m}}_{\parallel\:},{\varvec{p}}_{\varvec{z}}⟩\right.\)) is due to the interaction of \(\:\left|{\varvec{m}}_{\parallel\:}⟩\right.\) and \(\:-\left|{\varvec{p}}_{\varvec{z}}⟩\right.\) (see Fig. 2d). Consequently, a narrow dip emerges in the reflection spectrum, as shown in Fig. 1b, where this mode is able to achieve a strongly enhanced optical field at the nanogap region between disk and ellipse, especially at the plane of z = 25 nm above substrate for strong Purcell enhancement (see the electric and magnetic field distributions at Fig. S4). Moreover, we have also investigated the dependence of this hybridized mode on the nanogap size, which is a key parameter to engineer the ultimate field enhancement factor and Purcell effect (see Fig. S5). In addition, the resonances depend on the disk diameters (see Fig. S6), and we chose the design where the resonance matches well with the NV spectrum (a measured PL spectrum is shown in Fig. S7)
Figure 3a presents the home-built confocal microscope setup, which was able to measure the lifetime and secondorder correlation function, g(2)(τ), for the characterization of SPEs. The excitation was performed by a 515-nm laser, which is capable to operate either in continuous wave (CW) or pulsed modes (110 ps). A photoluminescence (PL) intensity map was first measured to determine the emitter locations on the metasurface (see Fig. S8 for details). To verify the emission of single photons, we performed the g(2)(τ) measurement for a representative single NV− center coupled to Fano resonance using a Hanbury-Brown and Twiss (HBT) system. The collected photon was directed to a fiber-based beam splitter and avalanche photodiodes (APDs) via single mode fiber. The correlation function shown in Fig. 3b exhibits a distinctive dip at τ = 0 with g(2) (0) = 0.24, proving the antibunching behaviour of the emitted single photons.
Figure 3c depicts a comparison of the lifetime between the SPEs on Si metasurface and on flat substrate region. Notably, the emission lifetime of SPE on metasurface is significantly reduced to 0.93 ns (blue curve); in comparison, the emitter at the flat substrate region has a lifetime of 9.56 ns, revealing a Purcell factor of ~ 10. The significant reduction of lifetime is a direct consequence of the higher densityof-opticalstates in the nanogap between disks and the ellipses, which leads to enhanced radiative emission (Purcell effect). This experimental result verifies the simulation investigation (Fig. 1d) on the peak Purcell factor. The difference in simulated and observed Purcell enhancement is explained by the fact that emission linewidth of NV− centers is larger rather than width of the resonance, which makes enhancement less efficient and thus reduces the effective Purcell factor.40 Similar decay dynamics were observed from other 5 coupled and 5 uncoupled single photon and quantum emitters, and their extracted lifetime are summarized in Fig. 3d and detailed in Table S1. These results confirm that the metasurface with Fano resonance could reduce the lifetime of the NV− center emitter effectively. Additionally, we measured the emission rate I as a function of the pump power shown in Fig. 3e. The data fits well to a saturation behaviour given by \(\:I\left(P\right)={I}_{sat}\times\:P/(P+{P}_{sat})\),41 where P is the pump power, \(\:{I}_{sat}\) and \(\:{P}_{sat}\) represent fitting parameters that denote the saturation intensity and saturation power, respectively. From the measured pump dependence, we found that the enhancement of the emission rate constituted 7.5 times in the linear regime. The measured emission rate enhancement agrees well with the lifetime reduction, proving the absence of Ohmic dissipation loss in dielectric nanostructures.
Next, we explored the polarization properties of single photon emission. We simulated the Purcell factor under various dipole excitations (Fig. 4a). It is observed that positioning a dipole at a 45-degree angle results in a Purcell factor as high as 23, indicating significant radiation enhancement. In contrast, setting the dipole at a 135-degree angle leads to a Purcell factor of only 0.95, demonstrating radiation suppression. We thus obtain a simulated degree of polarization (DOP) of 0.92. To verify the emission polarization experimentally, an analyser composed of a half-wave plate followed by a linear polarizer was put before collection single mode fiber and then sent to spectrometer. The collected photons exhibit the largest contrast with a ratio of 9 at the wavelength of 680 nm, which matches well with the simulation result (Fig. 4b). These observations clearly demonstrate strongly polarized single photon emission.