1. Introduction

In the field of optical biosensing, fluorescence spectroscopy has become one of the most widespread used tools in a number of specific applications [1]. During the past decade numerous methods based on high-sensitivity fluorescence detection have been introduced, including DNA sequencing and a variety of fluorescence immunoassays [2]. Fluorescence is sometimes preferred to colorimetric analysis due to its larger sensitivity. Advantages of such technique includes high selectivity, rapid response times, reusability and immunity from electrical interferences.

Thanks to the recent progress in micro and nanofabrication, a wide number of metallic, dielectric and metallo-dielectric structures have been made available for improving fluorescence-based detection techniques, mainly in terms of sensitivity and control of the angular and/or frequency spectrum of emission. Those structures include microcavities [3], waveguides and fibers [4, 5], photonic crystals [6], and plasmonic nano-structures exploiting propagating [7] and localized surface plasmons [8].

One of the issues addressed by the use of metallic/dielectric nanostructures in fluorescence sensing consists in the angular separation of different spectral components eventually belonging to different emitters that cannot be individually resolved in space (e.g. a mixture of different dyes in solution). The obtained angular separation generally results from the coupling of the fluorescence emitted from dyes close to the photonic structure having photonic states available to the radiation [9]. This phenomenon has been exploited in a number of configurations and systems, such as plasmonic flat metallic films [10, 11], plasmonic nano-ring antennas [12, 13], metallic nanoparticles dimers [14], and photonic crystals as well [15].

However, in a complementary approach to photon sorting as summarized above, spatial multiplexing can be also considered, as suggested by Cunningham and associates [16]. In this case, the fluorescence emission coming from different regions of a photonic crystal surface is selectively excited according to a “on-resonance/off-resonance” illumination condition, depending on both the illumination wavelength and angle of incidence. In such an enhanced fluorescence configuration [17], the detection system is image-based, aimed at a multi-spot immunoassays detection platform, and no further selection on the propagation direction of emitted fluorescence is performed.

In this work, we exploit a one-dimensional photonic crystal (1DPC) sustaining TE-polarized Bloch Surface Waves (BSW) [18] in a fluorescence detection setup based on a goniometric prism mounting. This type of silicon-based photonic structure has been already demonstrated to be successfully employed in sensitive gas-sensing [19] and specific molecular recognition [20, 21] experiments. Beside label-free detection schemes, 1DPCs sustaining BSW can provide useful means for enhancing and controlling the fluorescence emission of organic dyes in close proximity to the 1DPC surface [22, 23]. It has been shown that the detected BSW-coupled fluorescence is strongly polarized and directional, with a small angular divergence and a corresponding narrow spectral bandwidth. Furthermore, similarly to dielectric-loaded waveguides for surface plasmon polaritons (SPP) on smooth metallic films, BSWs can be spatially confined on ultra-thin relieves [24, 25]. As explained elsewhere [26], this effect is due to a combined role of the remarkably narrow energy/momentum BSW resonance and the redshift experienced by the BSW resonance upon slight surface perturbations, such as the deposition of a small amount of patterned organic add-layers. In a recent work [27], some of the authors demonstrated that such a lateral confinement of BSW-coupled fluorescence can be pushed down to approximately 500 nm in a suitable photonic structure consisting of AlexaFluor 546-labeled polymeric ridges 30nm thick patterned on a silicon nitride/silicon dioxide 1DPC. Here, the spatially-selective BSW coupling conditions on a similar patterned 1DPC are exploited in order to perform an angular/spectral multiplexing of BSW-coupled fluorescence. Since different regions of the patterned 1DPC can sustain BSWs belonging to different and well separated dispersion curves, distributed emitters will provide a fluorescence emission coupled to different BSW dispersion curves, depending on the location where they are grafted on. BSW-coupled fluorescence will be finally detected under different leakage angles out of the prism. In such a way we are able to angularly and spectrally differentiate fluorescence emitted from different regions of the structure characterized by a given relief height. In the present case, a proof of principle of such an effect is presented on a micrometer-sized stripes, but an extension to higher-density patterns [27] or multi-level structures is expected to work as well.

2. Fabrication of patterned 1DPC

The 1DPC structure used in the present work is composed by 8 periods of a pair of high (H) and low (L) refractive index layers made of amorphous silicon-based alloys (Si_xN_1-x, n_H = 1.99 and SiO₂, n_L = 1.48 at λ = 532 nm) with thickness d_H = 60 nm and d_L = 160 nm respectively, the last layer being silicon dioxide.

Fabrication is performed by means of Plasma Enhanced Chemical Vapor Deposition (PECVD) on 130 μm thick glass substrates. PECVD technique allows to deposit amorphous a-Si_1−xN_x:H and a-Si_1-xO_x:H layers. The compositions of the silicon nitride and the silicon oxide can be controlled by varying the ammonia fraction in a SiH₄ + NH₃ plasma, and the carbon dioxide fraction in a SiH₄ + CO₂ + H₂ plasma respectively. The refractive index and the thickness were estimated by means of standard spectroscopic measurements and deposition rate calibration data obtained for homogeneous films. In the PECVD deposition process the substrate temperature and the electrode distance were set to 220° C and 20 mm, respectively. For the a-Si_1-xO_x:H layers, the total pressure was set to 0.60 Torr, the reactive gas flow ratio [CO₂]/([SiH₄] + [CO₂]) to 97,6% the hydrogen dilution [H₂]/([SiH₄] + [CO₂] + [H₂]) to 71% and the RF power density to 104 mW cm⁻², whereas for the a-Si_1-xN_x:H layers the reactive gas flow ratio [NH₃]/([SiH₄] + [NH₃]) was set to 80% without hydrogen dilution, the total pressure to 0,45 Torr and the RF power density to 21 mW cm⁻².

Micrometric ridges (width 50 μm, height 30 nm, length 1 cm) are fabricated on 1DPC by means of standard photo-lithography, followed by a Plasma-Polymerized Acrylic Acid (PPAA) deposition (20 nm thickness) and a liftoff procedure in acetone. Before performing the photo-litographic and the lift-off process, a 5 nm thick Plasma-Polymerized Styrene film (PPST) is deposited on a defined rectangular portion on the silicon oxide surface of the 1DPC.

Plasma polymerization is a well known technique in the domain of biomedical applications mainly concerning with protein immobilization or cell adhesion processes [28]. In this work, a plasma polymerization procedure is used for obtaining both Plasma-Polymerized Acrylic Acid (PPAA) and Plasma-Polymerized Styrene (PPST) films on the top of the 1DPC. The PPAA film exposes carboxylic groups (-COOH) at the surface that can covalently bind to the amino groups (-NH₂) of AlexaFluor 546 Protein A. No extra steps for achieve higher yields of amide bond formation, such as the use of N-hydroxysuccinimide (NHS) and 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide (EDC) are required. Moreover, PPST films exhibit a significant hydrophobicity (contact angle OCA_H2O: 68 ± 2 deg) and a chemical inertness towards biomolecule functional groups (Surface Energy W_sl = 41.55 mN/m, Dispersive component W^d_sl = 41.55 mN/m, Polar Component W^h_sl = 5.71 mN/m). As reported in literature [29, 30], styrene plasma polymer acts as an anti-fouling layer, avoiding non-specific protein adsorption onto the surface.

Plasma treatments are carried out at room temperature in a PECVD reactor made of a stainless-steel vacuum chamber with cylindrical geometry (diameter = 320 mm; height = 200 mm), where two horizontal parallel plates of 15 cm in diameter act as electrodes. These are placed at 4 cm far away from each other. The gas mixture (acrylic acid vapors and argon for PPAA and styrene vapors for PPST) is uniformly distributed in the reactor by the upper showerhead electrode (with pinholes diameter of 2 mm). This electrode is externally connected to a 13.56 MHz RF power supply. For PPAA, argon is used as carrier gas (flow = 20 sccm), which bubbles into liquid Acrylic Acid in order to enhance vapors formation (acrylic acid vapor pressure = 3.1 Torr @ 20C°) that are subsequently driven into the chamber. A function generator is used to properly tune the on/off time of the plasma discharge.

After completion of surface patterning, the sample is incubated with 10ul AlexaFluor 546 Protein A 0.1mg/ml for 30 minutes. After incubation, the sample is rinsed twice in Phosphate Buffered Saline (PBS pH = 7.4) and deionised water for 10 minutes each step. A sketch the obtained 1DPC structure, together with a three-dimensional representation of the AFM topography of the patterned PPAA film (20 nm thick) on a uniform PPST film (5 nm thick) is shown in Fig. 1(a) .

 

Fig. 1 (a) Combined view of a sketch of the 1DPC deposited on glass and an AFM topography of the polymeric pattern on top (vertical dimension not to scale); (b) fluorescence micrograph of a AF546-labelled 1DPC structured with PPAA stripes on regions with and without the underlying 5 nm thick PPST layer. The exemplary profiles are shown to emphasize the fluorescence contrast within the two regions.

Download Full Size | PDF

The presented arrangement allows to enhance selective binding of labeled ptA on PPAA by exploiting the different chemical properties of the two polymers (anti-fouling PPST “trenches” and –COOH-rich PPAA “ridges”). In Fig. 1(b) a fluorescence micrograph of the sample surface is presented, in which it is possible to clearly see the gain in fluorescence contrast (in/out ridge) due to the presence of the underlying 5 nm thick PPST layer. Without PPST (outside the region wherein PPST in deposited), fluorescent ptA is non-specifically adsorbed over the whole surface. The proposed combination of anti-fouling/functional materials such as PPST/PPAA for fabricating ultra-thin photonic structures can be fruitfully exploited in biosensing applications.

3. Optical properties of patterned 1DPC

For the study of the photonic characteristics of the fabricated structure we can start form the calculation of the band diagram (TE polarization) of a bare, planar 1DPC by means of a transfer-matrix formalism [26], as shown in Fig. 2 . The band diagram shows regions in the (ω,β) plane in which the propagation of TE-polarized electromagnetic radiation is permitted (black) or forbidden (white). ω is the temporal frequency and β is the wavevector component parallel to the planar multilayer interfaces. Superposed to the band diagram, two BSW dispersion curves are displayed which are located within the first forbidden band of the crystal. The two curves describe the dispersion of BSWs on a 1DPC homogeneously coated with either a single layer made of PPST (5 nm) or a double layer made of PPST (5 nm) and PPAA (20 nm). The two BSW dispersions can reasonably approximate the corresponding dispersions of BSW coupled either on trenches (PPST) or on ridges (PPST + PPAA) of the patterned sample [26]. Accordingly, we expect to obtain a photonic structure exhibiting spatially-selective BSWs coupling conditions on regions having micrometric lateral size.

 

Fig. 2 Calculated band diagram of the bare 1DPC with dispersion curves of BSWs on the 1DPC homogeneously coated with either a single layer made of PPST (5 nm) or a double layer made of PPST (5 nm) and PPAA (20 nm).

Download Full Size | PDF

The experimental setup for characterizing the sample is shown in Fig. 3(a) . Either a TE-polarized collimated CW Nd:YAG laser beam (λ = 532.0 nm) or a TE-polarized polychromatic light from a pigtailed fibered white led (Doric Lens) illuminates the 1DPC oil-contacted to a glass prism according to the Kretschmann-Raether configuration. The prism is vertically mounted on a motorized rotational stage allowing to accurately adjust the angle of incidence θ_e. The detection arm is mounted on an independent homocentric rotational stage, in such a way that the radiation leaving the sample with an angle θ_d, with respect to the normal to the sample, can be detected on either the prism or the air side of the 1DPC. Light is angularly filtered by two diaphragms (resulting in an angular acceptance of about 0.2 deg.) and focused into a fibered dispersive spectrometer (Ocean Optics USB2000 + ).

 

Fig. 3 (a) Sketch of the experimental setup; (b) Measured normalized reflectance map R(θ_e, λ) of a patterned 1DPC as shown in Fig. 1(a), with stripes parallel to the BSW propagation. TE-polarized white light illumination and theta-2theta detection arrangement.

Download Full Size | PDF

In the case of white light illumination, an angularly resolved reflectance map R(θ_e, λ) can be obtained by operating the rotational stage according to a θ-2θ scheme. When angularly-resolved fluorescence measurements are concerned, and laser excitation takes place, the leaking radiation through the prism is spectrally filtered by means of an edge filter (RazorEdge from Semrock) for λ = 532.0 nm radiation before the spectroscopic detection. However, when white-light illumination is used, an angularly resolved reflectance map R(θ_e, λ) can be obtained by operating the rotational stage according to a theta-2theta scheme. In Fig. 3(b) a normalized angularly resolved reflectance map R(θ_e, λ) is shown as a function of the illumination angle of incidence (at the prism base) on a patterned 1DPC, as shown in Fig. 1(a). The two BSW dispersion curves are identified as low-reflectance regions beyond the critical angle. The appearance of these two modes is not due to a collective grating-like effect [15], but rather to the simultaneous and independent coupling of BSWs (at different wavelengths) over micrometer-spaced regions included in the illumination area of the impinging beam. Rigorous calculations performed with a well-known modal method (C-method [31]) show that, for a given fixed wavelength in the interval λ$\in$[530nm,620nm], the patterned 1DPC sustains two BSWs, i.e. a low-momentum and a high-momentum mode, coupled at θ_e$\in$[42deg,48deg], wherein the electric field is localized either on trenches or on ridges, respectively.

In Fig. 4 , the exemplary case of BSWs coupled at λ = 532 nm is presented. The high- and the low-momentum BSW for λ = 532 nm are found at θ_e = 45.3 deg and θ_e = 47.2 deg respectively, in accordance to the position of reflectance dips found experimentally (Fig. 3(b)). It should be appreciated that both BSWs are inherently surface modes and therefore well suited for surface sensing and enhanced fluorescence [32] applications. Since the two BSWs are associated to spatial regions having different chemical properties (i.e. anti-fouling PPST and functional PPAA in this case), it is be possible to exploit a “control” BSW as an indicator of refractive index variations due to non-specific binding and/or temperature or volume refractive index fluctuations occurring in a label-free bio-affinity measurement with self-referencing feature [33].

 

Fig. 4 Cross-sectional view of the electric field intensity distribution |E(x,z)|² within the patterned 1DPC: (a) low-momentum BSW (θ_e = 45.3 deg) localized on trenches; (b) high-momentum BSW (θ_e = 47.2 deg) localized on ridges. Wavelength λ = 532 nm, electric-field polarization perpendicular to the plane of view (TE).

Download Full Size | PDF

4. BSW-controlled fluorescence emission

In a previous paper [23], a suitable 1DPC uniformly coated with fluorescent organic emitters was demonstrated to produce enhanced TE-polarized, and directional fluorescence emission by means of BSW-coupling of both the excitation (laser) and the emitted radiation. Such an enhancement is produced by a corresponding intensity enhancement of the BSW-coupled laser radiation. Here, the scenario is somewhat more complicated. In fact, the fluorescent AF546-labelled ptA is inhomogeneously distributed on a patterned 1DPC sustaining two different BSWs that propagating on well separated spatial regions of the 1DPC surface. In biosensing, this feature can be exploited for selectively collecting fluorescence only from distributed regions (e.g. ridges) wherein a specific binding event (e.g. a specific molecular recognition process) is expected to take place. In our exemplary case, we are interested in discriminating between the fluorescence coming from emitters covalently bound to PPAA because of a NH₂-COOH reaction (resulting in CO-NH₂ amidic bond) and the fluorescence coming from PPST regions, where the adsorption is non-specific and thus less efficient. More complex patterned structures using multi-step functional materials can be considered as well, provided that the fluorescence coupled to BSW possesses enough angular/spectral separation after leaking out of the prism.

Experimental radiation patterns of fluorescence coupled to the two BSW are presented in Fig. 5(a) and Fig. 5(b) for regions of the same sample with and without PPST underlayer respectively. Measurements are performed collecting the fluoresence at several detection angles θ_d, under a direct and orthogonal illumination of the patterned 1DPC surface from the air side (θ_e = 180 deg). With such kind of illumination, no resonance effects are expected to influence the excitation efficiency and both regions of the sample are illuminated with the same intensity. Since the presence of the underlying PPST layer makes the relative amount of fluorescent ptA different from trenches to ridges, fluorescence intensities associated to the two BSW are different (Fig. 5(a)). This observation confirms the fact that the two BSW modes are associated to separated spatial regions. When a PPST-free region is illuminated, the two BSW-coupled fluorescent curves show almost equal fluorescence intensity levels (Fig. 5(b)). The observed BSW-driven fluorescence outcoupling is not likely to be subjected to a real intensity enhancement effect, but rather to an angular energy redistribution, as outlined in [23]. The comparison between dispersion curves within a single fluorescence map is made easier by the insets, where a plot showing the normalized angle-resolved integrated-fluorescence intensity in a 10 nm wide spectral region (as indicated in the figure) is reported. It can be clearly seen that the contrast between the normalized fluorescence intensities is larger in the case the PPST layer is present.

 

Fig. 5 Angle-resolved fluorescence maps collected upon illumination from the air side, normally to the patterned 1DPC surface. (a) Sample with 5-nm thick PPST layer; (b) sample with no PPST layer. The vertical band indicates the spectral interval used for calculating the integrated fluorescence as a function of the detection angle (inset).

Download Full Size | PDF

Figure 6 shows the angle-resolved fluorescence maps obtained by prism-illuminating the sample at the coupling angle of the high-momentum BSW for the two cases in which the PPST layer is present (Fig. 6(a)) or not present (Fig. 6(b)). In this case, fluorescence is actually enhanced upon occurrence of the field enhancement effect experienced by the laser radiation, either on trenches (Fig. 4(a)) or on ridges (Fig. 4(b)). It can be seen that when the PPST layer is used, the fluorescence coming from the non-specific binding of the ptA is strongly suppressed (on/off resonant illumination [16]). The benefits of a scheme using both the selective binding of a fluorescent dye through a PPST layer and the selective BSW-coupled illumination can be summarized by comparing Fig. 5(b) with Fig. 6(a). In the first case the SiO2 bare surface is binding aspecifically ptA, the fluorescent ptA is grafted almost homogeneously all over the surface and no resonant illumination occurs. Under these conditions, the fluorescence collected from inside and outside of ridges is comparable. In the second case, the binding of ptA is strongly conditioned by the role played by the PPST layer, and mostly occurs on the top of the ridges. In addition, the resonant excitation is such that only the fluorescence from the ridges experiences an intensity enhancement. The combination of these two effects results in a drastic suppression of the detected fluorescence from non-specific sites. The relative ratio of the fluorescence intensities ridge/trench obtained with air-side illumination was 100/20 (Fig. 5(a)) is pushed to 100/6 (Fig. 6(a)). We explain the absence of a full suppression of the off-resonance fluorescence (in this case, the low-momentum BSW-coupled fluorescence) because of cross-coupling effects of fluorescence emitted from ptA close to the ridge edges. A non-zero near-field laser intensity is also responsible for fluorescence excitation also in off-resonance conditions.

 

Fig. 6 Angle-resolved fluorescence maps collected upon illumination from the prism side, at the coupling angle of the high-momentum BSW (localized on ridges). (a) Sample with 5-nm thick PPST layer; (b) sample with no PPST layer. The vertical band indicates the area used for calculating the integrated fluorescence as a function of the detection angle (inset).

Download Full Size | PDF

5. Conclusion

In this work we presented a silicon-based 1DPC with an ultra-thin functional polymeric pattern on its surface, aimed at photon angular sorting. The photonic structure is designed in a way that a set of two BSWs are sustained and can propagate on different regions of the surface, each region having different chemical properties with respect to the covalent binding of AF-546-labelled ptA. In particular, we fabricated a structure in which a high-momentum BSW can propagate on relieves and a low-momentum BSW can propagate on trenches. It was shown that fluorescence coming from ridges can be clearly separated from fluorescence coming from trenches due to the preferential coupling of the emitted radiation to the available BSW modes. The combination of a resonant excitation together with a proper chemical functionalization results in a net suppression of the detected fluorescence from non-specific sites. The proposed method may open up new possibilities for the design of a multiplexing-fluorescence integrated-diagnostic device, in combination with angle-sensitive optical photodetectors [34]. In addition, photonic structures exhibiting different surface modes associated to spatially separated nanometric relieves could be used in self-referenced biosensing detection schemes.