We show that the tilted-grating-assisted excitation of surface plasmon polaritons on gold coated single-mode optical fibers depends strongly on the state of polarization of the core-guided light, even in fibers with cylindrical symmetry. Rotating the linear polarization of the guided light by 90° relative to the grating tilt plane is sufficient to turn the plasmon resonances on and off with more than 17 dB of extinction ratio. By monitoring the amplitude changes of selected individual cladding mode resonances we identify what we believe to be a new refractive index measurement method that is shown to be accurate to better than .
© 2010 Optical Society of America
Surface plasmon polaritons (SPs) provide an excellent tool for bulk chemical sensing and for monitoring reactions between molecules bound to a sensor’s surface and target molecules in solution . The use of core mode gratings to excite SP on metal-coated single-mode optical fibers has been proposed before [2, 3, 4, 5], but a more recent theoretical calculation has revealed a fundamental problem with this approach : the large optical loss of the SP mode causes a significant broadening and weakening of the grating induced coupling from the core mode to the SP, unless the grating length is kept smaller than the plasmon propagation distance in the metal (a few micrometers in our case, according to Eq. (1) of ). This effect is due to the fact that the coupling strength of contrapropagating grating couplers increases with grating length. However, if the backward reflected mode is lossy the coherent addition of backward waves from each grating plane can only occur over lengths comparable to the mode propagation distance. Therefore the amplitude of the resonant coupling for SP decreases to that of a grating that is only a few micrometers in length: i.e., weak and very wideband. In the work presented here, we use a tilted fiber Bragg grating (TFBG) to demonstrate experimentally for the first time to our knowledge that grating-coupled resonances do in fact disappear when the grating couples the core mode light to a cladding mode that has a strong SP component. However, we also present a technique to determine the peak surface plasmon resonance (SPR) wavelength as well as subnanometer wavelength shifts of the SPR peak by measuring the amplitude changes of cladding mode resonances provided by the TFBG. This level of sensitivity is critical for detecting small changes associated with the binding of biomolecules to the metal surface.
These findings are based on the discovery that the excitation of “SP-active” cladding mode resonances is highly polarization dependent. Using polarization control, we demonstrate that the TFBG can selectively excite SP active and nonactive cladding mode resonances with up to 17 dB of differential amplitude at the same spectral location. This observation differs from the results of Allsop et al.  who used a similar TFBG but in a D-shaped fiber obtained by lapping one side to within 10 of the core. Coating the flat side with metal resulted in the disappearance of the individual cladding mode resonances and broad SP attenuation bands that shifted with changes of the input light polarization azimuth. We use here a gold coated, otherwise unmodified standard single-mode fiber with cylindrical symmetry. We have shown previously that the transmission spectrum of a TFBG in such structures shows anomalous resonances that present strong similarities to SP excitations . However, the use of unpolarized light in these earlier experiments prevented the clear identification of plasmon modes.
In our devices (a diagram of the actual configuration can be found in ), the grating excites a large number of cladding modes . Among these modes, only those whose effective index and polarization state are equal or close to the effective index and polarization of a SP can transfer energy to it by tunneling across the metal layer. At wavelengths for which this situation occurs, the amplitude of the resonance decreases sharply because the cladding mode involved is very lossy, as discussed above . The 10-mm-long gratings used for the work reported here were written using the standard process for FBGs (apart from the tilt angle) in hydrogen-loaded standard single-mode fiber (CORNING SMF-28) using 248 nm wavelength excimer laser light and a phase mask [7, 8]. The internal tilt of the grating planes, is set at 10° to ensure strong coupling to cladding modes with effective indices near 1.3, suitable for the excitation of plasmons in aqueous environments [8, 9]. The sputtered gold coating is deposited in two steps, with the fiber placed horizontally above the sputtering target and rotated by 180° degrees along its axis between each step. So far, we found empirically that the optimal gold thickness for narrow SPR resonances is 50 nm. We used standard instrumentation for the measurement: the unpolarized light from a broadband source ( band pumped amplified spontaneous emission source) goes through a polarization controller before going through the fiber sensor and then to an optical spectrum analyzer (Ando AQ6317B). In the polarization controller, we use a linear polarizer (at an arbitrary angle since the input light is perfectly unpolarized) followed by a half-wave plate (HWP) to rotate the linear state of polarization. In the absence of sharp bends in the fiber between the HWP and the sensor, the state of polarization remains linear but arrives at the sensor at an angle that is unknown but that is rotated by a fixed, constant amount relative to the output of the HWP (this assumption is valid for single-mode fiber provided the fiber does not move and does not have sharp bends). For testing, the portion of the fiber where the metal-coated grating is located is fixed under slight tension using epoxy adhesive.
Figure 1 presents the transmission spectrum of a typical tilted grating SPR sensor for two orthogonal states of linear polarization. In order to find the appropriate polarization states, the HWP is rotated until the measured transmission spectrum shows the unambiguous SPR signature (top spectrum). In the bottom spectrum, the polarization has been rotated to maximize the amplitude of the cladding mode resonances across the spectrum: no evidence of SP effect shows up in this case (the amount of rotation between the top and bottom spectra is 90°). It is now obvious that the excitation of SP depends on the alignment of the linear state of polarization relative to the TFBG tilt direction. Since electromagnetic theory only allows plasmon modes with electric field polarized perpendicular to metal interfaces (radial direction in our case), our results indicate that the TFBG can couple the core mode of the fiber to cladding modes that have either no radial electric field or a strong radial electric field component at the cladding boundary.
So when the SP-active polarization state is chosen, the effective index of the SPR is found by locating the most attenuated cladding mode resonances. When the external refractive index changes or when the surface of the gold is modified (by the addition of a biolayer for instance), the complex effective index of the plasmon changes and different cladding modes become attenuated. This is shown in Fig. 2 for immersion in water and a calibrated refractive index liquid (supplied by Cargille Corp.). In Fig. 2, the locations of the SP maxima are fully supported by simulations carried out using a commercial finite-difference complex mode solver in cylindrical coordinates (FIMMWAVE 5.2, from Photon Design Inc.). The simulations provide the complex effective indices (and mode fields) of the structure used in the experiment (fiber , , core and cladding and respectively, metal and complex ). For each mode found, we use the grating phase matching condition (Eq. (3) of ) to calculate the resonance wavelength of the mode and we plot the imaginary part of the mode index (the mode loss) as bar graphs on Fig. 2. The simulations indicate that for both values of the external refractive index used in Fig. 2 (corrected for dispersion from the value, which corresponds to a wavelength of 589 nm), the loss of a small subset of the cladding modes resonances increases sharply almost exactly at the location of the experimentally observed SP (the small discrepancy is likely caused by uncertainty in the exact value of the permittivity of our sputtered gold layer.
The results of Fig. 2 confirm that we can actually measure refractive index changes using SP resonances excited by a grating in a fiber. For “large” refractive index changes such as those in Fig. 2, the SPR location can be determined by fitting a suitably shaped envelope function (a second order polynomial over the four central resonances for instance). The results of such fits, for experiments carried out in calibrated refractive index liquids, indicate a constant sensitivity of 555 nm/RIU ( index unit) with an rms deviation of in refractive index between the linear fit and actual measurements. The refractive index value of each liquid (i.e., ) was measured separately with an Abbe refractometer (with a nominal accuracy of ). The rms value provides a good estimate of the uncertainty in the measurement of the refractive indices provided by the device. More importantly, however, in most applications of SPR the purpose is not absolute refractometry over large refractive index ranges but rather the detection of very small changes that are occurring in a highly localized area near the device surface. For these applications a measurement uncertainty of the order of in refractive index is not acceptable, regardless of the nm/RIU sensitivity . In order to circumvent this problem, we use the fine spectral comb of cladding mode resonances provided by the TFBG in a slightly different manner.
Because the slopes of the SPR envelope are relatively large when polarization control is used, typically near 2 dB/nm, a SPR wavelength shift as small as 50 pm will cause a change of 0.1 dB in the amplitude of a cladding mode resonance located on the side of the SPR maximum. This is easily measured since only relative measurements are needed: an internal power level and wavelength reference is provided for each grating by the resonance of the core mode reflection (the resonance with the largest wavelength on Fig. 1). We demonstrate this high-resolution measurement mode by using mixtures of ethanol and water to generate small calibrated index changes in several steps over a total range of . The measurement error in the mixing ratio was estimated to be 0.001, yielding a nominal refractive index uncertainty of . Figure 3 shows both the amplitude change of the resonance and the equivalent SPR shift plotted against the refractive index change of the water–ethanol mixture. The results indicate a sensitivity of 1370 nm/RIU, comparable to the best SPR sensors available, and an rms refractive index uncertainty of .
In conclusion, we have shown that the excitation of SP by TFBGs on cylindrical fibers depends strongly on the orientation of the linearly polarized incident core guided light relative to the tilt direction. We also confirmed that a FBG-coupled optical mode that is perfectly phase matched to a SP does not produce a measurable resonance in transmission. However, we proposed an accurate method to measure small SP resonance shifts from the changes in the amplitudes of individual cladding mode resonances. This method provides a demonstrated refractive index measurement uncertainty lower than over a full range of 0.001. The device is easier to fabricate, less costly, and more reliable than other fiber-based SPR sensors because the fiber cladding does not have to be etched, polished, or tapered [1, 4, 5].
This work is supported by the Natural Sciences and Engineering Research Council of Canada and by LxData Inc.
2. Y. -J. He, Y. -L. Lo, and J. -F. Huang, J. Opt. Soc. Am. B 23, 801 (2006). [CrossRef]
4. T. Allsop, R. Neal, S. Rehman, D. J. Webb, D. Mapps, and I. Bennion, J. Opt. Soc. Am. B 25, 481 (2008). [CrossRef]
5. S. M. Tripathi, A. Kumar, E. Marin, and J. -P. Meunier, J. Lightwave Technol. 26, 1980 (2008). [CrossRef]
6. Y. -C. Lu, W. -P. Huang, and S. -S. Jian, J. Lightwave Technol. 27, 4804 (2009). [CrossRef]
9. Y. Y. Shevchenko, D. A. D. Blair, M. C. DeRosa, and J. Albert, in Conference on Lasers and Electro-Optics, (Optical Society of America, 2008).