Tamm plasmon- and surface plasmon-coupled emission from hybrid plasmonic-photonic structures.

Photonic and plasmon-coupled emissions present new opportunities for control on light emission from fluorophores, and have many applications in the physical and biological sciences. The mechanism of and the influencing factors for the coupling between the fluorescent molecules and plasmon and/or photonic modes are active areas of research. In this paper, we describe a hybrid photonic-plasmonic structure that simultaneously contains two plasmon modes: surface plasmons (SPs) and Tamm plasmons (TPs), both of which can modulate fluorescence emission. Experimental results show that both SP-coupled emission (SPCE) and TP-coupled emission (TPCE) can be observed simultaneously with this hybrid structure. Due to the different resonant angles of the TP and SP modes, the TPCE and SPCE can be beamed in different directions and can be separated easily. Back focal plane images of the fluorescence emission show that the relative intensities of the SPCE and TPCE can be changed if the probes are at different locations inside the hybrid structure, which reveals the probe location-dependent different coupling strengths of the fluorescent molecules with SPs and TPs. The different coupling strengths are ascribed to the electric field distribution of the two modes in the structure. Here, we present an understanding of these factors influencing mode coupling with probes, which is vital for structure design for suitable applications in sensing and diagnostics.


INTRODUCTION
Fluorescence is an established methodology that is used extensively in various fields including biotechnology, flow cytometry, medical diagnostics, DNA sequencing, forensics, genetic analysis, and so on. There has been dramatic growth in the use of fluorescence for cellular and molecular imaging, which reveal the localization and measurements of intracellular molecules, sometimes at the level of single-molecule detection [1]. However, fluorescence technology is reaching some natural limits and only incremental improvements in sensitivity-using classical far-field free-space optics-can be expected [2]. Until recently, almost all fluorescence measurements relied on the unperturbed free-space emission of fluorophores and subsequent manipulation of this emission by external optical components. To gain more opportunities and advances in fluorescence technology, we have been working on near-field coupled fluorescence where we use the interactions of the probes with metal nanoparticles or metal surfaces [3][4][5][6]. These near-field optical effects alter the emissions occurring at the fluorophore states. As a result, the metallic structures can convert the usual omnidirectional emission into directional emission and can modify the polarization of the coupled emission without the use of any lenses or polarizers [7][8][9]. The near-field coupling also results in spectral control on emission [10], enhanced spontaneous emission rates of molecules [11], giant suppression of photobleaching for single-molecule detection [12], and a strong fluorescence enhancement [13][14][15]. Due to these significant advantages of near-field coupled emission, it is of great importance to investigate the factors affecting the near-field coupling interactions between fluorophores and plasmonic fields. In this paper, we present a hybrid photonic-plasmonic structure that simultaneously contains two plasmon modes, surface plasmons (SPs) and Tamm plasmons (TPs). TPs, sometimes called TP polaritons, are a trapped electromagnetic state that exists between a metal and a dielectric Bragg reflector where the electric-magnetic field is highly confined. The electric field confinement in the metal is achieved as a result of its negative dielectric constant. The confinement in the dielectric multilayer structure is due to the photonic stop band of the Bragg reflector. Single quantum dots coupled to TPs were shown to experience acceleration or inhibition of their spontaneous emission depending on their emission spectral shift from the resonant wavelength of the TPs [16]. Based on this finding, new kinds of metal/ semiconductor lasers and a single-photon source using TPs have been realized experimentally [17,18]. TPs and SPs have different resonant angles and can beam the coupled fluorescence in different directions [19][20][21][22]. The observed spatial and intensity distribution of the coupled emission can be directly correlated to the probe locationdependent differences in the coupling efficiencies of the probes with the two plasmonic modes inside the hybrid structure. In other words, our experimental results show that the probe location determines which plasmonic mode will couple with the fluorophores. In the present structure, the coupling strength of the probe with SPs is stronger than that with TPs, which results in a higher intensity SP-coupled emission (SPCE). In contrast, the TP-coupled emission (TPCE) is more wavelength dispersive than SPCE and presents the advantages of wavelength separation and emission close to the normal angle. We believe that understanding these influential factors of probe interactions with a hybrid plasmonicphotonic structure would present new opportunities in the development of novel fluorescence device formats for real-time applications. Figure 1(a) presents the schematic of the hybrid photonic-plasmonic structure. This multilayer film is a typical one-dimensional photonic crystal (1DPC) [23][24][25][26][27] that can be readily fabricated by using simple vapor deposition techniques on a large scale. Plasmaenhanced chemical vapor deposition (PECVD) was used to fabricate alternate layers of SiO 2 (a low-refractive-index material) and Si 3 N 4 (a high-refractive-index material) on standard microscope cover glasses (0.17 mm thickness). Prior to PECVD, the glasses were cleaned with piranha solution and washed with deionized water. Then the cover glasses were dried in an air stream. Except the top SiO 2 layer, which had a thickness of 152 nm, the thicknesses of the SiO 2 and Si 3 N 4 layers were 126 and 78 nm, respectively. The sequence of layers on the glass substrate is shown in Fig. 1(a). In our experiments, we fabricated three 1DPC samples with the same structural parameters, which were used to obtain three types of hybrid photonic-plasmonic structures used below. A polymethyl methacrylate (PMMA) film (PMMA1) with a thickness of about 49 nm was spin-coated onto the top SiO 2 layer followed by an Ag film with a thickness of 40 nm. Finally, the second PMMA film (PMMA2) with a thickness of 33 nm was spin-coated onto the Ag film. The locations of the probes were determined by the position of the PMMA films. In our experiments, we used the probe Nile Blue (NB) (from Sigma-Aldrich), which was dissolved in PMMA solution (A2, anisole solvent, mass percent 2%, molecule weight 950 K; from ALLRESIST, Germany). The concentration of the NB molecules in the PMMA solution was about 2 mM. The spinning speed for the two PMMA layers was 8000 rpm. Due to the different surface properties of the top SiO 2 layer and the Ag layer, the thickness of the PMMA2 is slightly less than that of the PMMA1 layer [28,29]. Three samples were fabricated: the first one with the NB molecules doped only in the PMMA1 film, which was below the Ag film (this sample was named NB-1); the second one with the NB molecules doped only in the PMMA2 layer on the Ag film (NB-2); and the third one, which contained NB in both PMMA layers (PMMA1 and PMMA2, below and above the Ag film, respectively) (NB-3).

SAMPLES AND EXPERIMENTAL SETUP
The thicknesses of the two PMMA films are much smaller than the fluorescence wavelength, so the probes are all in the near-field region of the Ag film; the excited probes can therefore launch the two plasmon modes and induce corresponding coupled emissions [30,31]. BFP imaging was used to observe the coupled emission. BFP images reveal the angular distribution of the emission from the probes inside the hybrid structures [32]. A schematic of the BFP setup is shown in Fig. 1(b). A 532 nm laser beam was used as the excitation source, which was expanded to fill the rear aperture of the oil-immersed objective [×60; numerical aperture (NA), 1.42]. A polarization converter was used to change the linearly polarized laser beam to a radially polarized beam [33,34]. The samples were placed on the front focal plane of the objective. By using a collection lens, BFP images of the objective were recorded onto a CCD camera (Neo, sCMOS, Andor). A set of bandpass filters with center wavelengths ranging from 600 to 700 nm (10 nm step) was used to reject the excitation laser beam and obtain the fluorescence distribution on the BFP at a narrow wavelength. The bandwidth (full width at half-maximum) of these filters is 10 nm (from Thorlabs). In the transmission spectra measurement, the objective lens and the filters were removed. The collimated white-light source (halide lamps, LS-1-LL, from Ocean Optics) illuminated the sample with normal incidence. The light transmitted was collected by an optical fiber linked to a spectrometer (USB 4000, Ocean Optics).

RESULTS AND DISCUSSION
Before the BFP measurement, the transmission spectrum of the bare 1DPC without the Ag film and the PMMA films was measured as shown in Fig. 2. A broad dip appears at wavelength ranging from 600 to 775 nm, which is due to the photonic bandgap of the 1DPC. After coating with the Ag film and the two PMMA films, the transmission spectrum changes dramatically (Fig. 2). Here, a transmission peak appears at a wavelength of 682 nm. According to our previous work [22], the appearance of the transmission peak is due to the excitation of TPs at the 682 nm wavelength. To record the TPCE, a bandpass filter with center wavelength at 680 nm (which is close to the 682 nm transmission peak) was used. Figure 3(a) presents the corresponding fluorescence BFP image from the NB-1 sample where the NB molecules were doped in the PMMA1 film. A bright spot appears in the BFP image, which is due to TPCE. It is known that TPs have an in-plane wave vector less than the wave vector of light in vacuum, and sometimes the in-plane wave vector can even reach zero, so the emitting angle of TPCE is much smaller than the critical angle [21,35]. As a result, a bright spot associated with TPCE appears in the middle of the image [ Fig. 3(a)]. We also observed a weaker emission as a ring near 46.36°, which is due to SPCE [36]. We found this ring to be P-polarized, which is consistent with SPCE (image not shown). The angle, which was determined from the diameter of the ring and the NA of the objective (1.42), was larger than the critical angle of glass-air interface. The BFP image changed dramatically when NB was located above the Ag film (NB-2). In this case, only the SPCE ring was observed [ Fig. 3 Figures 3(a) and 3(b) clearly demonstrate that the plasmoncoupled emission is sensitive to the probe locations inside the hybrid photonic-plasmonic structures. There are two plasmon modes here, namely, the SP mode on the upper surface of the Ag film and the TP mode on the subsurface of the Ag film. These modes can be seen in numerical simulations of the angle-dependent reflectivity of the structure for illumination through the bottom (Fig. 4). The model and structural parameters used in the simulations are the same as those shown in Fig. 1(a) [23]. The transfer matrix method (TMM) is used in the simulation. The wavelength of the incident light is 680 nm, which is the same as the fluorescence wavelength shown in Fig. 3. There is a pair of narrow dips with resonant angles at +46.62° and −46.62°. A broad dip with center position at 0° also appears. To reveal the nature of these dips, electric field (E-field) intensity (|E| 2 ) distributions with the incident angle fixed at these resonant angles are plotted in Fig. 4(b). Here z = 0 represents the interface between the Ag film and the PMMA1 layer below as illustrated in Fig. 1(a). When the incident illumination is fixed at +46.62° or −46.62°, the strongest E-field localizes above the Ag film and inside the PMMA2 film, which decays along the +z direction. This is the typical feature of an SP field, so this pair of narrow dips corresponds to the SP mode. Similarly, E-field intensity distribution with the incident angle fixed at 0° is also plotted, in Fig. 4(b), which shows the strongest field localizing inside the PMMA1 layer. We can note that the SP mode is not visible with S-polarized illumination, but the TP mode remains the same (not shown). These are the typical features of a TP mode. The calculated SP resonant angle (SPRA) and TP resonant angle (TPRA) are consistent with the experimental results shown in Fig. 3. The dip associated with the TPs is much broader than that related to the SPs, so the emission divergence of the TPCE is much wider than that of the SPCE, which is consistent with the BFP image shown in Fig. 3(a).
We investigated the influence of probe location on the coupling strength between NB molecules and the two plasmon modes. When the NB molecules were doped in the PMMA1 layer (below the Ag film), their locations overlapped with the E-field of the TP mode [ Fig.  4(b)]. Under irradiation with a laser beam at 532 nm wavelength, the excited NB molecules acted as near-field point sources, which could generate TPs, thus resulting in TPCE. As shown in Fig. 3(a), the TPCE is much obvious. SPCE is also found in this case, but its intensity is weak. This observation indicates that in the NB-1 sample the coupling efficiency of the NB molecules with the TP mode is larger than that with the SP mode. By contrast, when the NB molecules are doped in the PMMA2 layer (above the Ag film), they are located in the E-field of the SP mode, and the BFP image [ Fig. 3(b)] shows a distinct SPCE ring. In this case, no significant TPCE is observed. This indicates that the coupling strength of the NB molecules with the SP mode is much larger than that with the TP mode. The two control experiments demonstrate that the optical coupling is dependent on the distance between the probes and the E-field of the plasmon modes: the shorter the distance, the stronger the coupling. But if the fluorescent molecules are too close to the surface of the metallic film, either the upsurface or the subsurface, fluorescence quenching by the metal surface can occur [37,38].
Comparisons between Figs. 3(a) and 3(b) also reveal another interesting phenomenon. The probes located above or below the Ag film are within the near-field region of the Ag film, so they can potentially excite the two plasmon modes, thus enabling the corresponding plasmon-coupled emission. But when NB was placed on the Ag film, we did not observe significant TPCE. When NB was below the Ag film, we observed both TPCE and SPCE. This difference may indicate that the energy conversion efficiency of the near-field NB molecules with the SP mode is larger than that with the TP mode. This result may be explained by the numerically simulated E-field distributions. As shown in Fig. 4(b), the strongest E-field intensity of the SP mode is about five times stronger than that of the TP mode. The stronger E-field may mean that there is a high fluorescence density of states for SPCE than for TPCE. An alternative possibility is a conversion from TPs to SPs. Additional research is needed to clarify whether either speculation is correct, which will be carried out in later experiments.
Another sample (NB-3) was investigated where the NB molecules were simultaneously doped in both PMMA layers (PMMA1 and PMMA2). The excitation laser beam at 532 nm is polarized radially. The expanded laser beam with radial polarization can excite the SPs in the Ag-PMMA2-air interface, which virtually forms an optical tip on the upper surface of the Ag film and results in the E-field enhancement of the excitation beam [39,40]. Then the NB molecules in the PMMA2 layer can be excited strongly. On the other hand, the excitation field in the PMMA1 layer is weaker than that inside the PMMA2 layer, so the NB molecules inside the PMMA1 will not be excited fully. Figure 5(a) presents the BFP images of the fluorescence from the NB-3 sample at 680 nm wavelength, which is similar to that shown in Fig. 3(b). In this case, SPCE is much brighter than TPCE.
It is well known that SPs can be excited only with P-polarized light at the SPRA, so when the polarization of the expanded laser beam is tuned to be azimuthal, the SPs will not be excited by the expanded 532 nm laser beam. The E-field of the excitation beam on the Ag film will decrease, although the intensity of excitation laser beam will remain unchanged. The corresponding fluorescence BFP image is presented in Fig. 5(b), where the intensity of the SPCE becomes about 2.5 times weaker than that in Fig. 5(a). TPCE becomes a little more obvious, which means an enhanced intensity ratio between TPCE and SPCE. Next, an aperture is placed before the objective, upon which the excitation laser beam cannot fill the rear aperture of the objective. The diameter of the aperture is about 3 mm, so the effective NA of the objective used for focusing the laser beam is about 0.39. In this case, although the laser beam is polarized radially, it cannot excite the SPs because all the incident angles of the laser beam are smaller than the SPRA. The depth of focus of the focused laser beam will be elongated due to the reduced effective NA [41]. By using this kind of an excitation condition, the NB molecules in PMMA1 and PMMA2 layers can be excited uniformly. In our experiment, we used the same concentration of the NB-PMMA aliquot to obtain both PMMA layers. There are more NB molecules in the PMMA1 layer than in the PMM2 layer because the PMMA1 layer is thicker. What is more important is that the SPCE has to penetrate the Ag film, which will be attenuated due to losses in the metal, whereas the TPCE does not need to pass through the Ag film. Figure 5(c) presents the corresponding fluorescence BFP image, where the TPCE and SPCE are nearly of the same intensity. This result shows that the optical conditions can be chosen to obtain nearly equal intensities for TPCE and SPCE, which may be valuable in sensing applications.
To demonstrate the coupling between the incident wavelength and the two plasmon modes, the laser is replaced with a white-light source (nonpolarized). A bandpass filter with center wavelength at 680 nm is used to filter the white light. Figure 5(d) presents the BFP image of the 680 nm light reflected from the hybrid structure. A distinct dark ring and a dark disk appear in the reflection BFP image, which are due to the respective dips in the attenuated total reflection (ATR) curves [ Fig. 4(a)]. The BFP image demonstrates experimentally the energy conversion from the far-field light to the near-field optical modes, such as the SPs and TPs. The dark ring corresponds to the SPs by the far-field light. From the diameter of the dark ring and the known NA of the objective (1.42), the SPRA can be derived 46.36°, which is consistent with the angle of SPCE. The dark disk in the center of the BFP image corresponds to the excitation of TPs with normal-incident light. Due to the excitation of these plasmon modes, the incident light has not been reflected at the corresponding resonant angles, which results in the dark areas (ring or disk) in the reflected BFP image. The darker the ring or disk, the more light has been coupled to the plasmon modes. Figure 5(d) shows that the ring is darker than the disk, and Fig. 3(a) also shows that the SPR dip is deeper than the TPR dip, which indicates that the energy conversion from the far-field light to the SP mode is higher than that to the TP mode for the proposed hybrid structure. Similarly, the energy conversion from the near-field excited NB molecules to the SPCE was found to be higher than that to the TPCE (Fig. 3).
Finally, as shown in the above BFP images, the emitting angles of SPCE and TPCE are different. As we know, the SPRA changes with wavelength, and we expected that the TPRA would be similar. Bandpass filters with center wavelengths ranging from 600 to 700 nm are used to investigate the wavelength-dependent emitting angles expected for TPCE and SPCE. These bandpass filters are placed between the objective lens and the white-light source to select different wavelengths. From the reflected BFP images at the selected wavelengths [similar to Fig. 5(d)], the angles for TPR and SPR can be estimated. The curves of SPRA and TPRA versus wavelengths are shown in Fig. 6. The SPRA changes gradually from 49.30° to 46.36° with increase in wavelength from 600 to 680 nm. The TPRA, however, changes more drastically from 31.41° to 3.96° for the same wavelength change. It should be noted that we also used bandpass filters with center wavelengths at 690 and 700 nm; the corresponding BFP images showed no TP mode, but the SP mode still existed (Fig. 6). The two curves show that the TP mode is more sensitive to wavelength than the SP mode. For comparisons, the image plot of the reflectivity spectrum for TM polarization is calculated by the TMM as shown in Fig. 6. In this image, the position of minimum reflectivity represents the resonant angle and the wavelength of the SP and TP modes. The consistency of the numerical and experimental results clearly shows the different wavelength sensitivity of the TPs and SPs.

CONCLUSIONS
In conclusion, our work provides novel insights into the near-field optical coupling between fluorescent molecules and plasmon modes. A hybrid photonic-plasmonic structure was used to investigate directional fluorescence emissions enabled by near-field optical effects. This hybrid structure displays two plasmon modes (SPs and TPs), which can induce both the directional fluorescence emissions. By tuning the locations of the probes and by recording the intensity changes of SPCE and TPCE, the effect of probe location on plasmon-coupled emission was revealed. The experimental results show that the intensity of SPCE or TPCE can be enhanced if the probes are located inside the E-field of the corresponding plasmon modes. The mechanism of this effect can be explained from the ATR curve and the E 2 -field distribution of the two modes. Our experimental results also demonstrate that the energy conversion from the near-field probes to the SP mode is larger than that to the TP mode, which can also be analyzed from the calculated E 2 -field intensity of the two modes. TPCE displays more sensitivity to wavelength than SPCE [42]. The hybrid plasmonic-photonic structure presents many opportunities for new applications of fluorescence. The multilayer structures described in this report are easy to fabricate using only vapor deposition and/or spin coating methods, and they can be prepared with large areas. The angle of TPCE can be easily adjusted by changing the thickness of the film. The simultaneous presence of two plasmon modes presents opportunities for self-reference sensing or diagnostic devices. We expect these structures to become widely used in analytical instruments for the biosciences [43,44] and for the development of a new kind of laser [45]. Transmission spectra of the hybrid photonic-plasmonic structure (1DPC + PMMA1 + Ag +PMMA2, red line) and the bare 1DPC (black line).  (a) Angle-dependent reflectivity curves calculated from the photonic-plasmonic hybrid structure for P-polarized incident light. The incident wavelength is 680 nm. (b) Electric field intensity distributions in the hybrid structure with incident angles fixed at the resonant angle of SPs (46.62°) and TPs (0°). The position z = 0 corresponds to the subsurface of the Ag film. Fluorescence BFP images of the NB-3 sample where the NB molecules are doped in the two PMMA layers, above and below the Ag film. The excitation laser beam is polarized radially in (a) and (c) and azimuthally in (b). (c) BFP image when an aperture is placed before the objective to decrease the diameter of the expanded laser beam. (d) Reflected white-light BFP image from the NB-3 samples. A bandpass filter with center wavelength at 680 nm is used to filter the white light and fluorescence. Experimental curves (dotted line) of SPRA and TPRA versus wavelength. The color image shows the plot of the calculated reflectivity spectrum for TM polarization. The color scale bar on the right represents the normalized reflectivity.