Surface Plasmon-Coupled Enhanced Transmission

For distances less 10 nm, a total energy transfer occurs from a quantum emitter to a nearby metallic surface, producing evanescent surface waves that are plasmonic in nature. When investigating a metallic nanohole supported on an optically dense substrate (such as diamond with nitrogen vacancy center), the scattering occurred preferentially from the diamond substrate towards the air for dipole distances less 10 nm from the aperture. In addition, an enhancement to the dipole's radiative decay rate was observed when resonance of the aperture matched the emitters wavelength. The relationship between an emitter and a nearby resonant aperture is shown to be that of the resonance energy transfer where the emitter acts as a donor and the hole as an acceptor. In conjunction with the preferential scattering behavior, this has led to the proposed device that operates in transmission mode, eliminating the need for epi-illumination techniques and optically denser than air superstrates in the collection cycle, hence making the design simpler and more suitable for miniaturization. A design criterion for the surface grating is also proposed to improve the performance, where the period of the grating differs significantly from the wavelength of the surface plasmon polaritons. Response of the proposed device is further studied with respect to changes in nitrogen vacancy's position and its dipolar orientation to identify the crystallographic planes of diamond over which the performance of the device is maximized.


Introduction
The use of optically dense materials to increase the collection efficiency associated with a quantum emitter (QE) is popular [1][2][3][4][5][6][7][8]. The inclusion of diamond superstrates [4] with its low absorption [9], for example, allows the emission from a Nitrogen Vacancy center(NV-) in a nanodiamond to be readily collectable from the superstrate. Most designs, however, are based on the epiillumination techniques where the incident pump power and the luminescent radiation share the same half-space. See for example the study undertaken by Aouani et. al. [10,11] on the fluorescence of an ensemble of molecules with randomly oriented dipole moments, trapped inside a plasmonic aperture, when illuminated and collected through the glass substrate. A more deterministic study showed the dependence of the radiation pattern of plasmonic antennas with respect to the distance and the dipolar orientation of a nearby nano-diamond with NV- [7,12].
Devices based on reflection geometry rely on point-illumination and point-collection using the same objective lens which is an intrinsic feature of the confocal microscopy. Maintaining an optimal distance between the objective lens and the emitter, therefore, is a crucial factor in maximizing the incident power over a single photon emitter. Moving the objective lens away from the emitter into the far-field zone for collection purposes, (hence moving the emitter away from the focal point), reduces the incident pump power over the dipole drastically. Consequently, the excitation of a dipole in reflection mode must either be carried out by side-illumination techniques via a separate objective lens, or by a confocal lens in conjunction with a Bertrand lens allowing for collection from the back focal plane. Such configurations besides being more complex require extra elements on the optical path. For all miniaturization purposes, therefore, it is beneficial for a device to operate in transmission mode where the incident pump utilized the substrate and luminescence is collected from the superstrate. In the context of detecting a quantum emitter, however, this implies that a device must possess multiple qualities, such as enhancement and collimation of the emission.
Recent developments in nano-diamond fabrications have led to designs such as a diamond film on a glass substrate with bullseye (BE) grating etched on the diamond surface surrounding a nitrogen vacancy to collimate its emission [13]. The use of a homogenous diamond film with NV-near the surface aimed to minimize complications (such as total internal reflection) associated with alternative designs where nano-diamonds are integrated with external structures. Although such a multilayer structure may operate in either the reflection or the transmission mode, numerical analysis showed a higher field intensity being trapped inside the diamond film in comparison to those scattered into the air or into the glass substrate [13]. This may be attributed to the complex interaction of the NVemission with the air/diamond and/or diamond/glass boundaries, which includes total internal reflections and the quenching of the dipole's emission by nearby dielectric interfaces. In this report, it will be shown that by introducing a nearby metallic surface, it is possible to prevent the NV-'s emission from being trapped or scattered inside the diamond membrane.
Understanding the interaction between a dipole and nearby metallic surfaces has led to the Surface Plasmon-Coupled Emission (SPCE) [14]. Theoretical and experimental studies have shown that fluorophores within ~10 nm of metallic surfaces are quenched [1]. The quenching effect, however, is understood based on analytical solutions and experimentally reported data being focused on the reflected power or the decayed power through the flat metallic thin films [14][15][16][17][18][19]. The quenching region of less than ~10 nm away from metallic surfaces is in fact related to the total energy transfer between the dipole that acts as a donor and the metallic surface acting as an acceptor [20]. The quenching of quantum emitters by a nearby metallic surface results in surface waves that propagate at the supercritical angle along the metal/dielectric interface [21], not being detectible in the far-field. This mechanism is advantageous if utilized properly. It can be shown that such evanescent surface waves are plasmonic in nature and may be utilized favorably. In other words, by positioning a dipole at distances of ~10 nm (or less) from metals, it is possible to consume most of the power emanating from the dipole to induce Surface Plasmon Polaritons (SPP) that propagate over the bulk metallic surfaces and/or excite Localized Surface Plasmons (LSP) in subwavelength metallic features such as a groove or an aperture. A preliminary study of a resonant aperture in a silver film on a diamond substrate with NV-10 nm away from the aperture, [22], showed the emission from the aperture occurred preferentially towards the air, meaning that evanescent surface waves and freely propagating EM waves were much stronger in the air half-space in comparison to those inside the diamond. This is partially due to the SPPs having longer decay length of ~67 μm at the silver/air interface in comparison to ~3.2 μm at the silver/diamond interface. Such a behavior provided an opportunity for a new design.
The platform propose here consists of a single crystal diamond membrane with a NV-near its surface [9,[23][24][25] that acts as a donor, integrated with a planar plasmonic antenna acting as an acceptor, to achieve a highly unidirectional transmitted luminescence with enhanced Radiative Decay Rate (RDR) and high collection efficiency in the air, hence, eliminating the need for optically dense superstrates and epi-illumination. In the case of a diamond substrate, being a material of choice in this report, it is equally important to identify optimum crystallographic planes of diamond over which the proposed planar structure to be fabricated. With NV-'s symmetry axis being one of the <111> crystallography directions [26,27], it is possible to identify such planes in which the energy transfer efficiency from NV-to the antenna is maximized. Such a platform may also be realized by other substrates with near surface emitters such as a SiC with defects [28] or a silicon with embedded quantum dots [29]. A review on quantum dots and ways to enhance its emission is given by Xu et al. [30]

Resonance Energy Transfer
A rigorous theoretical study on dipolar emission near a flat metallic surface was previously carried out by Novotny [21,31]. Here a numerical study of a diamond membrane with a NVpositioned at zd is undertaken, where z = 0 plane marks the diamond/air interface. The model is developed in 2D due to the limited computational resource available to the author. The analysis is then extended to a multilayer diamond/silver/air structure. Finite Element Method (FEM) simulations are carried out at the target wavelength of λ0 = 700 nm that coincides with the spectral peak of the phonon sideband associated with the emission from nitrogen vacancies in nano-diamonds at room temperature [32,33]. The refractive index of the diamond substrate is set to nsub = 2.4 and the refractive index data for silver was taken from Palik [34]. Figure 1(a) depicts the scattered electric field intensity, |E| 2 , calculated for a diamond membrane with a NV-positioned at (xd, zd) = (0, -10) nm with its dipole moment aligned with the x-axis (i.e. horizontally oriented). The power ratio of Pair/Psub ≈ 0.5 is a clear indication that in the absence of any plasmonic structure most of the emitted power is scattered into the diamond. The power ratio is the free space wavenumber, εm and εd are relative permittivities of the metal and dielectric respectively. It is apparent that some of the emanating power from the dipole is consumed in launching SPPs. Presence of a dipole-like activity associated with a metallic slit has been reported previously [35]. Therefore, by introducing a resonant aperture perforating the film just above the NV-, it is possible establish the dipole-dipole resonance energy transfer between the NV-and the aperture. Figure 1(c) shows the enhancement to the Radiative Decay Rate (RDREair) in the air half-space vs. the film thickness, t, when an aperture having a width wa = 30 nm, perforates the silver film just above the emitter. The RDREair was calculated for the air half-space using where P0 is the total power emitted by a nanodiamond in vacuum. The diameter of the isolated nano-diamond was also set to 30 nm in all calculations. This is a typical size for a nano-diamond with luminescence properties [36][37][38]. The factor of 0.5 in the denominator is due to P0 being calculated along the arc length of a full circle, whereas Pair was calculated over a semicircle in the air half-space. With zd = -10 nm, a maximum RDREair ≈ 140 was achieved for t = 110 nm that corresponds to the first Fabry-Pérot resonance of the aperture [39]. Figure 1(d) depicts the electric field intensity, |E| 2 , calculated for the diamond/silver/air multilayer with a 30 nm wide aperture perforating the 110 nm silver film just above the NV-. The power ratio Pair-Ag110nm+hole/Psub-Ag110nm+hole = 10 is a clear indication that scattered power is preferential towards the air. Calculated energy transfer efficiency between the donor and the acceptor [21], ET = PD→A/(PD + PD→A) = Pair-Ag110nm+hole/(Pair-noFilm + Pair-Ag110nm+hole) ≈ 1 was obtained for zd ≤ -10 nm. Power ratios Psub-Ag110nm+hole/Psub-noFilm = 1.5 and Pair-Ag110nm+hole/Pair-noFilm = 33, however, suggest that the presence of the silver film with a resonant aperture enhances the reflected power inside the diamond as well as the transmitted power into the air, hinting at a change in the density of states. It is therefore intuitive to infer that RDREair ∝ PD→A. Radiation pattern of transmitted field, however, is dispersive with 1/e of its maximum intensity not reaching beyond z ≈ 1.7 μm along the optical axis, i.e. z-axis. Figure 1(d) also shows the presence of high intensity evanescent field along the silver/air interface. Electric field components, Figure 1(f), and their corresponding FFTs, Figure 1(h), calculated over the silver/air interface, confirmed the presence of surface waves with λSPPa = 667 nm which is in agreement with λSPPa = 682 nm obtained analytically. Although evanescent waves do not contribute to the far-field [21], it is possible to intercept and scatter them into freely propagating EM waves by introducing periodic surface gratings surrounding the aperture [40]. Note however, that for a horizontally oriented dipole moment, induced Ex is an even function of x whereas Ez is an odd function. Consequently, any contribution made by Ez to the scattered fields leads to a destructive interference along the optical axis. Therefore, considering a 2D model, Ex (with even parity) is the only contributing factor to the far-field intensity along the optical axis.

Design of Surface Gratings
Detecting the transmitted luminescence of a QE through the aperture, requires reshaping its dispersed radiation pattern into a highly directional light with a maximum possible intensity along the optical axis of a BE structure. Tailoring the exit surface surrounding the aperture with properly designed corrugations is crucial since poor choices may lead not only to the suppression of an already enhanced RDR, but also the formation of side-lobes in the radiation pattern, hence a reduction of transmitted power along the optical axis. Analytical models based on forward propagating SPPs has led to discrepancies between the model and the experimentally observed shortened propagation length attributed to excess losses that required experimentally obtained fitting parameter to correct the theory [41]. It can be shown, however, the shortening of the propagation length is due to the superposition of surface waves. Relation governing the EM waves along the optical axis of a BE structure consisting of a hole surrounded by N periodic corrugations with periodicity P, may be described by:  Figure 2(a). The role of a corrugation (or a groove) positioned at xn = nP, is to utilize some of the power carried by the SPPs to excite LSPs inside the groove, Figure 2(b), whose far-field radiation pattern is similar to that of a dipole antenna. The strength of the LSP inside the n th groove is denoted by An. To simplify the argument consider equation (1) with N = 1 in a symmetric configuration consisting of an aperture at x = 0 and a pair of grooves at x1 and x-1 with x-1 = -x1, where the superposition of SPPs at x = x1 may be defined as: Here, the first term on the right-hand side is the forward wave launched by the aperture arriving at x1, the second term is the wave launched by the aperture towards x-1 being reflected back towards x1, and the third term is the reflected wave from x1 towards the aperture, being reflected back towards itself. The phase difference between the wave emanating from the aperture and the wave scattered by the groove calculated along the z-axis, Figure 2(c), but z = ∞ is not a quantitative measure for the farfield. Although matching the period of the corrugations to the SPP wavelength leads to a collimated beam at z = ∞, simulations confirmed that it did not produce the highest possible intensity along the z-axis (results not shown here). Therefore, matching the period of the corrugations to λSPPa is an oversight. An alternative criterion for maximizing the field intensity along the optical axis is proposed here. Ψ xx = , vs. x1 obtained at the silver/air interface using equation (2). Note that the square of the amplitude is proportional to the total power available to excite LSPs inside the groove, hence  Given that the aperture and the grooves form an antenna array, as long as their equidistance is less than λ0, high intensity radiation along the optical axis is guaranteed [42] (p. 399). But one must also maximize the radiating power available to both the aperture and the grooves. It is intuitive, therefore, to set x1 to the average of the two peaks mentioned above, i.e. x1 = 0.5 × (475 + 505). It is obvious that equation (2) becomes convoluted for N > 1. Considering such complex interactions, the only viable approach in designing a BE structure in the presence of a quantum emitter, is to solve the Maxwell equations numerically. With t = 110 nm and wa = 30 nm being the optimum values for the film thickness and the aperture width, r1 was set to x1 + w1/2 = 555 nm and a series of parametric sweeps were carried out over each of the remaining parameters w1, wp, h and P for zd = -10 nm. In each case, the spectrum for the scattered power into the air vs. the parameter was calculated to identify the peak position that corresponds to the optimum dimension. Note that this is the same approach used in determining the optimum film thickness, t, see Figure 1(c). One possible configuration, C1, capable of collimating the emission as well as improving the RDRair was found to be wa = 30 nm, w1 = 130 nm, wp = 210 nm, h = 70 nm, r1 = x1 + w1/2 = 555 nm, P = 590 nm and t = 110 nm. Enhancement to the transmitted RDRair peaks for N = 5 with no significant changes to the radiation pattern for N > 5, see Figure 3. This implies that most (if not all) of the power carried by the SPPs are intercepted and scattered by the first five grooves. Therefore, for the rest of the calculations only 5 grooves are considered. Note that in this report, all 2D field plots are normalized to the same scale and comparable to one another. With the smallest spot size achievable by an Electron Beam Lithography (EBL) or a Focused Ion Beam (FIB) being as low as 5 nm [43,44], fabrication of the aperture and the surrounding surface features poses no challenge. The resonance energy transfer between the donor and the acceptor, on the other hand, is sensitive to the changes in dipole's position and orientation. The rest of the report, therefore, is dedicated to study such variations.

Dipole's Position and Orientation
The transmitted collection efficiency CE = Pair-C1/(Pair-C1 + Psub-C1) and RDREair vs. zd is depicted in Figure 4(a). Note that preliminary investigations with 1 nm steps, revealed the presence of a single peak in RDREair at zd = -3 nm. Therefor the resolution for the final parametric sweep over zd was set to 1 nm in the range of -5 nm ≤ zd ≤ 0 and 5nm for zd < -5 nm. An enhancement of RDREair ≥ 216 was achieved for zd ≤ -5 nm with a maximum of 219 at zd = -3 nm and a drop to 1/e at zd = 17 nm. Rate of change in CE vs. zd was found to be much slower in comparison to that of RDREair. A minimum of 80% efficiency in the range of zd ≤ 25 nm with a maximum of 86% at zd = -5 nm was obtained. For of a horizontally oriented dipole, transmitted RDREair showed a reciprocal relation to those corresponding to dipole emissions near flat metallic surface, compare Figure 4(a) in this report to Figure 10.5(b) in "Principles of nano-optics" [21], for example. The slow drop of ~1/zd 0.8 in RDREair is attributed to the energy transfer between the NV-and the resonant aperture that involves a mutlichannel mechanism. While the direct excitation of LSPs inside the resonant aperture, with its finite cross section, is dominated by short range interactions, the surface surrounding the aperture is able to capture NV-'s emission for longer distances, producing SPPs which consequently couple to the aperture's LSPs. This slow rate of change with respect to the dipole's distance from the surface is extremely beneficial. Beaming profiles for few arbitrary zd values are depicted in Figure 4  For dipole distances zd ≤ -10 nm where the near total energy transfer occurs, the donor-acceptor relation between the NV-and the aperture implies that the performance of such a device is sensitive to the donor's dipolar orientation with respect to that of the acceptor. Retaining only the near-field term, radiation pattern of the donor (i.e. NV-) may be written as [21]: where 22 r x z =+ with ϑ being the angle between r and the donor's dipole moment, μD, see inset of Figure 5(a). C1's response vs. the direction of µD in the x-z plane was calculated in the range of 0° ≤ Θxz ≤ 90°. Figure 5(a) shows that the power transfer between the donor (NV-) and the acceptor (aperture) does in fact comply with PD→A ∝ |µA·ED| 2 [21], where µA is the dipole moment of the acceptor. Consequently, for a vertically oriented µD, PD→A and RDREair drop to zero. Unlike the simple donor-acceptor interaction observed in florescent molecules, the interaction between the NVand C1 is influenced by plasmonic effects, therefore surface wave launched by the aperture must also be examined. Numerically obtained phase difference between Ex oscillations on the corners of the aperture at the silver/diamond interface, L-R 0 aa ( / 2, 0) ( / 2, 0) respectively. For pure 1 state of Ex, however, one must not assume that NV's emission is quenched.
It can be demonstrated that for a vertically oriented µD a small displacement in its horizontal position with respect to the center of the aperture, not only overcomes the suppressed PD→A, but also causes the transition from the odd to even state, 10 → , hence satisfying the two mandatory conditions for a high intensity constructive interference along the z-axis. Numerical results depicted in Figure  5(b) confirm that for 16 ≤ xd ≤ 20 nm both conditions are satisfied. The trend observed in the RDREair spectrum is due to the donor's projected field intensity over the aperture, |ED| 2 (x=0,z=0), varying with xd. Analytical values for |ED| 2 (x=0,z=0) vs. xd were calculated using equation (3), with sin ϑ = |xd|/r, see Figure 5(b)-inset. Although, equation (3) does not take into account the aperture's dimension, reflection terms and interactions with surface waves, the trend in |ED| 2 spectrum agrees with that obtained numerically for RDREair. Directional gain of the BE antenna, |E(θ)| 2 /|E0(θ)| 2 , vs. Θxz is depicted in Figure 5(c). The field intensity along the optical axis can be 800 times that of a free standing nano-diamond when Θxz = 0°. Here |E(θ)| 2 and |E0(θ)| 2 are the electric field intensities as a function of angle, θ, from the x-axis calculated for C1 and a nano-diamond particle in vacuum respectively. FWHM of the beam remains approximately 14° for all Θxz values. Corresponding directional gains vs. xd = {10, 16, 21, 33 nm when Θxz = 90° are shown in Figure 5(d). The angle θ, may be expressed in terms of Θxz, xd and zd as:  Previously, it was shown that the xz (or yz) plane of experimentally obtained volumetric radiation patterns of a bullseye structure, corresponds to the radiation pattern of a simulated 2D grating [22](see chapters 8, 9 and Appendix A). To evaluate C1's response vs. the change in µD in the xy plane, C1 was first modelled in 3D as a BE structure with a 250 nm long slit and later with a symmetric cross-shaped aperture having the same arm-lengths. The spherical simulation domain was divided by the planar structure into two hemispheres, with the bottom being the diamond substrate and the upper being the air. The first Fabry-Pérot resonance of the aperture in both cases occurs at t = 130 nm. To reduce the required computational resources, number of corrugations were limited to N = 3. All other dimensions were kept in accordance to the 2D model. A parametric sweep over the orientation of µD in the xy plane, Θxy, showed that RDREair ∝ |µA·ED| 2 for the slit, as anticipated. RDREair in 3D was calculated using being carried over the surface s enclosing the hemisphere. The maximum RDREair obtained from the 3D model was calculated to be ~100 (results not shown here). Far-field intensities, |Efar| 2 vs. Θxy also followed a similar trend with collimated beams being achieved for Θxy < 90°, Figure 6(a)-(c). Far-field intensity emanating from the symmetric cross-shaped aperture, on the other hand, was impervious to the change in Θxy but the maximum far-field intensity in this case was almost half of that achieved by the slit, i.e. |Efar-cross| 2 / |Efar-slit| 2 = 0.45 at their respective maxima. Presence of an additional arm in the cross-shape aperture reduces the nearby metallic surface area over which |ED| 2 is projected on to, hence the reduction in absorption cross section of the acceptor. For a horizontally oriented µD, field amplitude of the donor projected over the aperture vs. xd was calculated analytically using equation (3) with sin ϑ = |zd|/r, see Figure 7(a). The trend observed in |ED| 2 (x=0,z=0) spectrum agrees with that of the RDREair obtained numerically, Figure 7(b). Numerical results revealed that the device tolerates a lateral displacement of Δxd ≈ ±15 nm beyond which the RDREair drops below 1/e of its maximum. The phase difference ΔΦL-R in this case rises at xd ≥ 7 nm, reaching a maximum of ~105° at xd = 23 nm but never attaining a 180° difference to qualify Ex as a pure odd function. RDREair, on the other hand, drops to 1/e of it maximum at xd ≈ 15 nm where ΔΦL-R is 70°. Defining the state of even function-ness as cos 2 (ΔΦL-R/2), a 70° phase difference must drop the RDREair only to 0.67 (not 1/e) of its maximum. The 55% excess reduction, therefore, is partially due to the Ex not being a pure even function and partially due the reduction in |ED| 2 (x=0,z=0) with increase in xd.

Crystallographic Directions
From all possible dipole orientations considered in this report, it can be inferred that a horizontally oriented µD with a plasmonic slit offers the best outcome with the highest RDREair and tolerance. Descriptions on triplet notations and Miler indices used in crystallography, may be found in undergraduate solid state texts [45,46]. With NV-symmetry axis being one of the <111> crystallography directions [26,27], one may narrow the search for a horizontally oriented µD. Assuming that the planar design lies on an arbitrary plane, z, defined by its normal vectorẑ , possible scenarios are (a) z plane parallel to diamond's ( ) 1 10 plane which implies that µD is parallel to the 1 12   direction, hence the slit must be fabricated such that µA ∥ µD. To maximize the emission from the NV-in this case, the excitation field, Eex, must also be polarized in the 1 12 (c) z plane∥111) (orẑ being aligned with the symmetry axis of the NV-) with no preferential direction for neither Eex nor µD since µD = α 2 µD1 + β 2 µD2, where α 2 + β 2 = 1, see Figure 8. Depending on the application, one can identify other diamond planes over which the planar structure to be set. For examples: if the emphasis is on the extraction of single photons, then the diamond plane must be chosen such that (µA ∥ µD1 AND µA ⊥ µD2) OR (µA ∥ µD2 AND µA ⊥ µD1).
On the other hand, if one is to investigate the two-entangled photon arising from the mutual excitations of µD1 and µD2, then it is best to implement the planar structure using a symmetric cross aperture on the (111) plane with the arms of the cross being aligned with 1 12   and 1 10   . There is no preferential direction for either Eex or µD since µD = α 2 µD1 + β 2 µD2, where α 2 + β 2 = 1.

Conclusion
Calculations demonstrated the donor-acceptor relation between a quantum emitter and a resonant aperture in Surface Plasmon-Coupled Enhanced Transmission (SPCET). The emission of a dipole captured by and transmitted through a metallic resonant aperture is reciprocally related to the quenching of dipole's emission near flat metallic surface. Meaning that, for distances where the dipole emission is quenched near a flat metallic surface, transmission through a metallic resonant aperture is maximized. Unidirectional scattering of SPCET through the aperture into the air is particularly advantageous in moving away from designs based on the reflection geometry that utilize epi-illumination/collection from optically dense substrates or superstrates. Considering that SPP propagation over the plasmonic bullseye structure involves Fabry-Pérot-like resonances between the grooves and the aperture, a new approach for designing the corrugations was proposed with periodicity that differs significantly from the SPP wavelength. Combining the SPCET effect with the newly proposed BE structure, it was shown that it is possible to collimate the radiation of a quantum emitter (positioned inside the substrate) with a high yield and collection efficiency in the air. The platform was studied with diamond substrate as the material of choice with NV-near the surface. The dipole moment of a NV-in a single crystal diamond retains its direction at all time. By characterizing potential NV-centers and their dipolar orientations in a single crystal diamond substrate, it is then possible to fabricate the structure over an already identified NV-to maximize the overall performance of the device. Depending on the application, the planar plasmonic structure may be fabricated over a specific diamond plane. The challenge, however, is to grow the diamond film so that the target plane is exposed to the proposed plasmonic structure.