2218 OPTICS LETTERS / Vol. 30, No. 17 / September 1, 2005

Tapered fiber optic surface plasmon resonancesensor for analyses of vapor and liquid phases

Yoon-Chang Kim, Wei Peng, Soame Banerji, and Karl S. BookshDepartment of Chemistry and Biochemistry, Arizona State University, Tempe, Arizona 85287-1604

Received April 7, 2005; revised manuscript received May 9, 2005; accepted May 9, 2005

We report a new approach to analyze both vapor and liquid phases by utilizing a tapered fiber optic surfaceplasmon resonance (SPR) probe. This technique employs a fiber optic SPR probe with a modified geometry totune the SPR coupling wavelength–angle pair. The observed composite spectrum included two distinct SPRdips associated with surface plasmons excited in the gas and liquid active regions. This sensor is able todetect refractive index changes in both vapor and liquid phases individually by simultaneous monitoringSPR coupling wavelengths from the two sensing surfaces. © 2005 Optical Society of America

OCIS codes: 060.2370, 240.6680, 240.6690.

For the last decade surface plasmon resonance (SPR)sensors have been extensively studied because ofmany applications ranging from analysis of biomate-rials to environmental inspection. Two main types ofSPR sensor have commonly been employed: constant-angle SPR and constant-wavelength SPR. The prism-based SPR sensor system can be used either as aspectral or angular SPR sensor. The prism-basedSPR sensors are not, however, optimal for in situ in-dustrial or environmental process monitoring be-cause of their bulky construction. In comparison witha prism-based SPR sensor, fiber-optic-based SPR sen-sors are fundamentally simpler in construction, areless costly, require a smaller sample volume, andmuch more amenable for remote sensing applica-tions. However, in a traditional fiber-based SPR sen-sor system1,2 there is not a fixed incident angle butinstead is a range of angles that are allowed to propa-gate in the multimode fiber probe. The incident lightconsists of a wide range of wavelengths launched intothe fiber in order to excite the surface plasmon waveacross a range of angles defined by the numerical ap-erture (NA) of the fiber optic waveguide. Recently de-veloped sensors1–4 that employ tapered-end fiber op-tic SPR sensors have displayed performancecharacteristics in line with that of the aforemen-tioned prism-based SPR sensor modality. The SPRcoupling wavelength can be tuned selectively bymodifying the angle of incidence between the photonand the sensing area; with the fiber optic sensors thisis accomplished by modifying the geometry of theprobe tip. We have shown that dual-tapered SPRprobes can be quantitatively employed to monitor ei-ther vapor3 or liquid1,2,4 media.

Multichannel SPR sensors have historically at-tracted attention owing to their potential uses inmultianalyte detection.5,6 In general, multichannelSPR sensors have been constructed from a prism5 ora light pipe6 containing sensing channels with recog-nition elements for detection of specific analytes, andtheir applications were in analyte detection in theliquid phase. In order to detect gas species, single-channel SPR sensors were developed.7 The principleinvolves placing a thin layer of chemically active ma-terial on a highly conductive metal along which the

0146-9592/05/172218-3/$15.00 ©

SPW propagates. Because of the presence of the gas,the optical changes in this layer result in refractiveindex (RI) changes near the metal surface. As far aswe know, this is the first report of a dual-channel fi-ber optic SPR sensor for both liquid and gas phaseanalyses. The method does not necessitate placementof a thick polymer or dielectric overcoat layer.

In this Letter, we present a novel fiber-based SPRmonitoring application in which the RIs of the vaporand liquid media could be monitored by use of asingle probe. This newly developed SPR sensor isbased on the advanced modification of the geometryof the probe tip and the spectral discrimination ofsensing areas.

The basic design of the fiber optic SPR system em-ployed was previously described.1–4 The probe de-signs used in the experimental setup for the SPRmonitoring are shown schematically in Fig. 1. Silica-core optical fibers with 0.12 NA (CeramOptec Indus-tries, Inc.) were employed. The dual-tapered SPRprobe has the sensing surface and the front mirrorpolished at orthogonal angles between the two sur-faces. The tetra-tapered SPR probe has two comple-mentarily positioned mirrors and two sensing sur-faces. The white light from the illuminating fiber–mirror reflects off of the tapered surfaces of the probeand is incident onto the sensing surface at the de-signed angle, �inc. Observed SPR spectral featureswere compared with the theoretical SPR featuresfrom the model proposed by Ishimaru.8 A MATLAB pro-gram was written in house to calculate the theoreti-

Fig. 1. Drawings and photographs of (left) dual-taperedand (right) tetra-tapered SPR probes (diameter 560 �m).

2005 Optical Society of America

September 1, 2005 / Vol. 30, No. 17 / OPTICS LETTERS 2219

cal SPR response. This is the same model that wasemployed by Johnston et al.9 Note that the theoreti-cal SPR model assumed transverse magnetically(TM) polarized light, whereas the optical fiber probesuse unpolarized light. Thus, where the theoreticalspectral profiles approach complete attenuation atthe spectral feature, the observed spectral featureswould approach 50% attenuation. The fact that SPRspectra are actually an attenuation from unity, notadditive from a zero baseline like absorbance spectra,was accounted for in the assumed linear additivespectral mixing model. Reflectivity references werecollected before each set of samples with the fiber op-tic sensors. References were collected in media wherethe SPR response would be far outside of the530.38–917.38 nm window. For aqueous samples, airspectra were used for the reference; for vaporsamples, water was used for the reference. Normaliz-ing each spectrum by the appropriate referenceyields the familiar SPR dip.

Figure 2 shows contour plots of two theoreticalthree-dimensional SPR reflection spectra for air andwater as the bulk dielectric media. This plot illus-trates that the corresponding reflection is dependenton the resonance of both coupling angle and wave-length. It is clear from Fig. 2 that the surface plas-mon wave is excited at many angles simultaneously,and the SPR spectra for air and water would be ob-served over the entire range of the propagatingangles, 65°–89° and 42°–53°, respectively.

Based on the theoretical spectra predicted by theFresnel equations, nine gold-coated SPR probes wereconstructed with tapered tips of 48° /42°, 49° /41°,50° /40°, 51° /39°, 52° /38°, 70° /30°, 75° /25°,80° /20°, and 85° /15°; these sensors would be re-sponsive to RI changes in vapor or liquid phases. Theprobes were tested with either air or water as appro-priate. The test results are shown in Fig. 3. As theangle is decreased, the SPR dips shift toward longerwavelengths. The observed SPR features for the me-dia are coincident with the theoretical spectra withconsideration of angle profiles. The theoretical SPRmodel assumes a single reflection on the sensing sur-face and a single mode traveling in the waveguide.Thus there are five problems that mutually com-pound to frustrate direct use of multimode optical fi-bers in any practical study. First, for strict control ofincident light, collimation must be implemented toget sharp spectral features such as theoretical SPRspectra. Of course, this is impossible in multimodeoptical fibers. Second, therefore, low-NA fibers must

Fig. 2. Contour plot of the theoretical SPR spectra as afunction of incident angle ��inc� and incident wavelength�� �.

inc

be employed to get the narrow ranges of angles inci-dent on the sensing areas, because higher NAs resultin a broader distribution of angles in the fibers.Therefore the available commercial low-NA (i.e.,0.12) fibers were employed to fabricate the taperedSPR probe in this study. However, third, with low-NAfibers the propagating modes are too acute with re-spect to the fiber optic walls to allow gold-coated fiberwalls to be used directly to support the surface plas-mon wave. Therefore tapering of the fibers is neces-sary to select the desired incident angle to realize aSPR feature at a given RI. Unfortunately, fourth, fi-ber tapering has a certain degree of imprecision at-tached to the mechanical procedures that serves tofurther broaden the SPR features, because the SPRactive surface will not be perfectly flat, smooth, orcomplementary in angle with respect to the reflectingmirror. Finally, fifth, the number of reflections on thesensing surface must be considered when making thesensor probes, and that number depends on the inci-dent angles. If the tapered probes have the same areafor active SPR, the number of reflections on the sens-ing surface will increase as the designed angle de-creases. In the observed SPR features, broader SPRspectra at the smaller angles (i.e., 48°, 70°, and 75°)for each phase may be caused by increasing the dis-tribution of unexpected angles on the surface follow-ing a larger number of reflections.

In order to analyze both vapor and liquid phases, atetra-tapered SPR probe was designed based on inci-dent angle discrimination for vapor and liquid phasesamples. The wavelength separation between vaporand liquid phase samples is shown in the theoretical(Fig. 2) and experimental (Fig. 3) spectra. Collimatedpolychromatic light launched into the tetra-taperedSPR probe is incident on a gold layer in the four dis-tinct regions; each region has a different incidentangle that is complimentary to the opposite region(e.g., 50° /40° and 75° /15°). The light reflects off thetapered sides at the designed angles (i.e., 50° for va-por, 75° for liquid) and is directed back in parallel tothe incident light through the mirrors polished tocomplimentary angles (i.e., 15° and 40°). The light isincident upon the mirrors as well, but SPR on themirrors is not activated because of their thickness��200 nm�. The light bounced off mirrors is reflectedto the tapered SPR active areas with the same inci-dent angles as the directly incident light and then re-turns to the detector. The optical wave in the tetra-tapered SPR probe generates two dips in thespectrum that are centered at two different wave-

Fig. 3. Air and water SPR spectra from dual-taperedprobes in relation to incidence angle.

lengths. Consequently, as shown in Fig. 4, the ob-

2220 OPTICS LETTERS / Vol. 30, No. 17 / September 1, 2005

served wavelength spectrum of the light exhibits twodips when the tetra-tapered SPR probe was testedwith air and water. As far as the authors are aware,this is the first reported SPR feature demonstratingdetection of both gas and liquid species by use of anSPR probe.

To test the quantitative capabilities of the liquidphase sensing, a sensor with the tetra-tapered probewas tested with six glucose solutions (0%–5%). Theliquid SPR coupling wavelength around 680 nm in-creased linearly with glucose solution concentration,while the SPR feature around 570 nm for air exhib-ited no shift. The corresponding linear regressionequation can be represented as follows:

��SPR = 3.7�C − 0.048, �1�

where ��SPR is the SPR coupling wavelength in-crease in nanometers and �C is the glucose concen-tration increase in percent. The root mean squarederror of calibration across the range 0.00%–5.00% is0.431% glucose solution.

The tetra-tapered SPR probe was also employed tomonitor ammonia gases with a range of 0.08%–4%concentrations. An equation for the fitting curve is

��SPR = − 0.012�C2 + 0.095�C + 0.090, �2�

where ��SPR is the SPR coupling wavelength in-crease in nanometers and �C is the ammonia gasconcentration increase in percent. The root meansquare error of the calibration across the range

Fig. 4. Air and water SPR spectra from a tetra-taperedprobe (�inc 50° and 75°).

0.080%–4.0% is 0.099% ammonia gas. Nonlinearityat high concentrations ��3% � may result from poormixing of ammonia gas in the glass vessel or excess(nonlinear) dielectric properties of the air–ammoniabinary mixture at higher ammonia concentrations.

In conclusion, a simple and rapid method for theevaluation both of gas and liquid media was devel-oped. This method is based on the advanced modifi-cation of the geometry of the SPR probe with wave-length division multiplexing of sensing surfaces. Thedevelopment of these fiber optic sensors makes it pos-sible to employ vapor- and liquid-sensitive thin filmsthat change in refractive index following exposure togas and liquid phase analytes. This investigation in-dicates that the advanced probe geometry is promis-ing for the development of fiber optic SPR sensors forboth gas and liquid detection or simultaneous mul-tiple liquid or gas species detection.

The authors thank the National Science Founda-tion (OCE 0119999) for financial support. Y.-C. Kim’semail address is Yoon-Chang.Kim@asu.edu.

References

1. L. A. Obando and K. S. Booksh, Anal. Chem. 71,5116–5122 (1999).

2. L. A. Obando, D. J. Gentleman, J. R. Holloway, and K.S. Booksh, Sens. Actuators B 100, 439–449 (2004).

3. Y.-C. Kim, S. Banerji, J.-F. Masson, W. Peng, and K. S.Booksh, Analyst (Cambridge, U.K.) 130, 838 (2005).

4. Y.-C. Kim, J.-F. Masson, and K. S. Booksh, “Single-crystal sapphire-fiber optic sensors based on surfaceplasmon resonance spectroscopy for in situmonitoring,” Talanta (to be published).

5. C. E. H. Berger, T. A. M. Beumer, R. P. H. Kooyman,and J. Greve, Anal. Chem. 70, 703–706 (1998).

6. G. G. Nenninger, J. B. Clendenning, C. E. Furlong, andS. S. Yee, Sens. Actuators B 51, 38–45 (1998).

7. X. Bévenot, A. Trouillet, C. Veillas, H. Gagnaire, andM. Clément, Meas. Sci. Technol. 13, 118–124 (2002).

8. A. Ishimaru, Electromagnetic Wave Propagation,Radiation, and Scattering (Prentice-Hall, 1991).

9. K. S. Johnston, S. R. Karlsen, C. C. Jung, and S. S. Yee,Mater. Chem. Phys. 42, 242–246 (1995).