1. Introduction

Fluorescent molecules are widely used in numerous research and sensing applications. In order to achieve the most efficient fluorescence detection, we need good understanding of how fluorescent molecules behave in various experimental geometries, especially in close proximity to metal and dielectric interfaces. In many previous papers (see for example [1–8]), the radiation patterns of fluorescent molecules deposited on dielectric/dielectric or dielectric/metal interfaces have been calculated and observed. However, much work remains to be done on development of the most efficient geometries for fluorescence detection. For example, a collection lens may be designed according to the angular distribution of the emitted signal [6]. Alternatively, evanescent wave (EW) structures have been suggested, which considerably enhance fluorescence through enhanced coupling to evanescent waves [9]. Lakowicz et al. demonstrated [10–13] that the fluorescent signal can be enhanced by utilizing surface-plasmon coupled emission (SPCE) (Note however an ongoing discussion of this subject: in ref. [14] it was reported that the fluorescence signal must be attenuated by the presence of a metal film. On the other hand, Kreiter et al. recently observed enhancement of fluorescence near metal surfaces [15]). In both the SPCE and the EW structures, sensitive optical designs must be carefully implemented. The evanescent wave needs to be coupled at a precise angle; while the collection angle for SPCE covers a cone with a large semi-vertical angle, and can be inconvenient to collect. In this paper we propose and study another type of fluorescence detection geometry in which the fluorescence signal is enhanced by at least a factor of 10 compared to more usual geometries, yet does not require any complicated optical arrangement, so that a regular commercial fluorescence optical microscope (FOM) may be used. In our geometry the fluorescence enhancement is achieved through enhancement of pumping light at the substrate because of surface plasmon excitation facilitated by dielectric surface gratings. The studied geometry may be potentially useful in sensing applications.

2. Experiment

2.1 Comparison between gratings deposited onto a metal and a dielectric substrate

Our device geometry is shown in Fig. 1. A layer of fluorescent material (R6G) dissolved in ethanol has been spin-coated onto a PMMA grating. 40 nm thick PMMA nano-stripe gratings have been formed by E-beam lithography on top of two kinds of substrates: an ITO/Glass and an Au/Glass substrates, as shown in Fig. 1(a) and 1(b), respectively. The thickness of the Au layer was about 50nm. A typical periodicity of the PMMA stripe gratings shown in Fig.2 was 500nm. Figure 2 shows the AFM image of the PMMA grating and how the gratings were arranged on the surface of our samples. All samples have been examined under a fluorescence optical microscope. The wavelength of the excitation filter is centered at 560nm with 40nm bandwidth. The emission barrier filter is located at 610nm. The emission peak of R6G (in ethanol) is 590nm. The results are shown in Fig. 3. Figure 3(a) corresponds to R6G on ITO substrates. Figure 3(b) was taken with R6G on Au film substrates. The sample surfaces were prepared using the same procedure, and the images were taken with identical exposure times and gains of the CCD camera used in the fluorescence microscope. In order to analyze our numerically, we have extracted the digital values of the signal for each pixel of the JPEG image file produced by the CCD camera, and compared the ratios of the digital values (DV) at different pixels. Note that these ratios have to be analyzed by taking into account the gain factor Γ of the CCD camera. Typical values of Γ range from 0.45 to 2.5, and the light intensity is related to the digital value (DV) according to Intensity=const(DV)^Γ. The CCD of the fluorescence microscope used in our experiments was set at Γ~2.2 in order to emulate the response of the human eye. Figure 3 shows that the fluorescence of R6G deposited on top of the unperturbed PMMA layer (without a grating) is barely detectable. Contrary to the conclusions of ref.[14], the fluorescence coming from the PMMA gratings formed on top of the gold film is the brightest. Compared to the grating on the ITO substrate the fluorescence is enhanced by at least a factor of 10 (the ITO patterns have some defects so there are some dark dots on the grids). In this example, we have not yet optimized the grating pitch and the excitation polarization. Since fluorescence microscopes (FOM) are commonly used in bio-detection, we conclude that these results already indicate the competitive potential of our geometry in biosensing applications.

 

Fig. 1. Device structures

Download Full Size | PDF

 

Fig. 2. (a) AFM image of the nano-stripe gratings; (b) Dimensions of the pattern

Download Full Size | PDF

 

Fig. 3. The intensity comparison of R6G/PMMA gratings on (a) ITO/glass substrate; (b) Au/glass substrate.

Download Full Size | PDF

2.2 Polarization and periodicity dependence

In order to understand how the fluorescence signal taken with the FOM is affected by the periodicity of the gratings, we made a sample with regions of different periodicity varying from 400nm to 1μm. The sample geometry is illustrated in Fig. 4(a), in which the grating periodicity is given in nanometers. Figures 4(b,c) indicate that the fluorescence enhancement depends strongly on the grating periodicity. We also studied the polarization dependence of the observed effect. In this experiment a mercury lamp filtered by a film polarizer was used as the excitation source at normal incidence. The sample was rotated so that the polarization direction was changed with respect to the grating trenches. The results of these experiments are shown in Figs.4 and 5. The exposure time for (b) and (c) was 250s and 700s respectively. Figure 5 shows the normalized digital value taken from the images. Every value is normalized to the background and the exposure time. The fluorescent efficiency is 10 times higher when the E field is parallel to the grating trenches.

It seems clear that some kind of surface plasmon polariton excitation is involved in the phenomena observed in our experiments. If we compute the SP dispersion and try to match the k-momentum, we find that for 640nm emission, the plasmon mode and radiation mode are strongly coupled because of k-vector momentum matching provided by the grating and the emission angle is zero degree when the grating periodicity is 736 nm. In order to study the enhancement mechanism in more detail we have studied how the excitation angle affects the fluorescence excited at various grating periodicities.

2.3 Rotation of incident angle

A one-dimensional (1D) PMMA grating on an Au film surface acts like a 1D plasmonic crystal [16]. In order to relate the fluorescence enhancement with plasmonic crystal properties of our substrates we have performed more detailed measurements of fluorescence at different angles of the excitation light. In these experiments the incident laser light was tilted at an angle θ with respect to the z-axis and rotated by an angle α with respect to the y-axis in the x-y plane as shown in Fig. 6. The emission intensity of each pattern is recorded with the α rotation of every 10 degrees. Figure 7 shows two patterns illuminated at different rotation angles a. In Fig. 7(a), 411nm and 840nm patterns emit efficiently, while in Fig. 7(b), the most efficient fluorescence comes from the 693nm pattern. Fig. 7(c) indicates the position and periodicity of different gratings. The fluorescence signal measured as a function of angle is shown in Fig. 8 for different periodicities of the PMMA gratings. The background signal was subtracted from every data point and normalized to the CCD exposure time. The angle α is scanned from -10 to 90 degrees.

 

Fig. 4. Fluorescence under normal excitation - (a) the arrangement of grating periodicity (in nm). (b) FOM pictures taken under the polarized Hg Lamp. E field is parallel to the grating trenches (exposure time: 250s). (c) The sample was rotated 90 degrees clockwise. E field is perpendicular to the grating trenches (exposure time: 700s).

Download Full Size | PDF

 

Fig. 5. Polarization effect on gratings with normal incidence to the sample surface.

Download Full Size | PDF

3. Discussion

The angle α, which corresponds to the maximum of the fluorescence signal, can be determined from Fig. 8. The error of the measured angle is in a range of ± 2.5°. The reason for the unsymmetrical intensity at -10° and +10° is that α is not accurately tuned to a symmetrical position. To explain the angle effect, the incident wave vector k₀=2π/532nm is decomposed as the projected wave vector k₀sinθ in the x-y plane and k₀cosθ along the z-axis. The component k₀sinθ can be decomposed into x and y components as k₀sinθsinα and k₀sinθcosα, respectively. The grating k vector 2πn/a can provide momentum matching along the x direction as shown in Eq. (1), while the y component remains unchanged as shown in Eq. (2). If a surface plasmon is excited, the k-vector of the incoming photons mediated by the grating periodicity should match the k vector of surface plasmons as shown in Eq. (3):

 

Fig. 6. The geometry of the incident laser beam and angle definitions.

Download Full Size | PDF

 

Fig. 7. (a) α=30° Patterns with 841nm and 411nm periodicity are excited most strongly. (b) α=30° The pattern with 693nm periodicity fluoresce most strongly. (c) The pattern arrangement of (a) and (b).

Download Full Size | PDF

 

Fig. 8. Fluorescence emission vs. angle α for different gratings.

Download Full Size | PDF

where n is an integer, and k^* is the composite k value in the x-y plane. In our experiments the incident angle θ was about 44°. The angle α is determined from the peak intensity shown in Fig. 8. The theoretical k_sp for long-range SPPs on the vacuum/Au interface is 2π ∙10⁶ 2.087 m^-1 at 532nm. Figure 9 shows k^* for n=-2, -1, 0, and 1. At least one good integer order n can be fitted to the theoretical k_sp for every periodicity. Table 1 shows the coupling order n, the maximum excitation angle α and the maximum digital value of the image intensity for each periodicity. The digital values for each period are comparable to each other because they are normalized to the exposure time. For 604nm and 693nm, the coupling order is ± 1. These gratings exhibit a higher fluorescence intensity compared to other gratings for which only one diffraction order is coupled efficiently.

 

Fig. 9. k^* for different order n

Download Full Size | PDF

Table 1:. The relation between the coupling order n and the digital value of the image intensity.

View Table

4. Conclusion

We have observed strongly enhanced fluorescence from a layer of Rhodamine 6G (R6G) on a dielectric grating deposited on top of a thin gold film. The fluorescence enhancement has been studied as a function of the grating periodicity and the angle of incidence of the excitation light. The enhancement mechanism is consistent with excitation of surface plasmon polaritons on the metal film surface. The observed phenomenon may be promising in sensing applications.