demonstrated by calibration curves of tetrachloroethylene in hex- ane recorded with a tapered sensor ber coupled to a Fourier trans- form infrared (FT-IR) ...
Chemically Tapered Silver Halide Fibers: An Approach for Increasing the Sensitivity of Mid-Infrared Evanescent Wave Sensors M . KARLOW ATZ, M . KRAFT, E. EITENBERGER, B. MIZAIKOFF,* and A. KATZIR Institute of Analytical Chemistry, Vienna University of Technology, Getreidemarkt 9-151, A-1060 Vienna, Austria (M. Karlowatz, M. Kraft, E.E., B.M); and Raymond and Beverly Sackler Faculty of Exact Sciences, School of Physics and Astronomy, Tel Aviv University, Tel-Aviv 69978, Israel (A.K.)
In this work an innovative etching technique for tapering silver halide bers is introduced. As silver halides form soluble complexes with thiosulfate in aqueous solution, the ber can be chemically tapered by an etching process, which also warrants a high quality of the ber surface. The evan escent eld sensitivity of thus obtained tapered bers was raised by more than one order of magn itude, demonstrated by calibration curves of tetrachloroethylene in hexane recorded with a tapered sensor ber cou pled to a Fourier transform infrared (FT-IR) spectrom eter. Index Headings: IR ber-optic sensor; Silver halide bers; Chemical tapering; On-line monitoring; Chlorinated hydrocar bons.
INTRODUCTION In recent years, the awareness of the public with respect to the problem of ground water, surface water, and drinking water as well as seawater contamination with volatile organic com pounds (VOCs) has increased considerab ly. V O C s such as chlo rinated h ydro carbo ns (CHCs) and aromatic hydrocarbons are among the most comm on organic pollutants detected in water.1 Preferred standard methods for VOC analysis are static headspace (HS) and purge and trap (PT) gas chrom atography combined with m ass spectroscopy (GC-M S). PT or HS-GC in conjunction with an electron capture detector is the standard m ethod for the analysis of single CHCs such as TCE and TeCE. These methods usually suffer from their restriction to laboratory conditions and, therefore, from error-prone sampling steps. The described sensor system based on ber-optic evanescent wave spectroscopy (FEW S) offers a prom ising alternative, since the possibility of multi-analyte on-line sensing of VOCs in water is provided, as described in the literature. 2–5 In general, the relatively high limit of detection (LOD) for state-of-the-art mid-infrared (m id-IR) ber-optic sensor systems has always been a drawback. Hence, methods for increasing the sensitivity are of substantial interest. Methods of tapering bers creating a thinned section for increasing the num ber of internal total re ections have already been accomplished for optical bers with glasslike properties. 6–11 However, the commonly used drawing techniques for such materials cannot be applied for silver halide bers, due to their polycrystalline structure. In this work an innovative etching technique for taReceived 13 May 2000; accepted 17 July 2000. * Author to whom correspond ence should be sent.
Volume 54, Number 11, 2000
pering silver halide bers is introduced. To enable measurements with bare silver halide bers, we used hexane as a nonabsorbing solvent in the spectral range of interest (around 910 cm 2 1 ) for the exemplary analyte tetrachloroethylene (TeCE). Different mixtures with TeCE concentrations in the percentage range were used to provide pronounced absorption bands and to obtain calibration curves. Untapered and several tapered bers were compared with the use of these analyte mixtures. For a m ore detailed ber quality characterization, the signal-to-noise (S/N) ratios of the tapered bers were examined. Finally, scanning electron m icroscope (SEM ) images of the bers were recorded, to determine the precise ber diameter and to verify the impact of the etching process on the surface quality. EXPERIMENTAL M aterial Properties of Silver H alide Fibers. Common IR ber-optic materials are heavy metal uoride glasses, chalcogenide glasses, tellurium halides, and polycrystalline silver halides. The exact speci cations of these materials are the subject of a wide variety of publications and reviews and shall not be discussed in detail.12–14 Therefore, further information will be given only for silver halide bers, which were used for the presented measurements. Silver halide bers are produced by pressing a rod consisting of a m ixture of pure silver halide crystals (AgCl x Br x 2 1 with 0 , x , 1; temperature: up to 200 8C, pressure: up to 2000 bar) through a diamond form. The composition of the crystal mixture (depending on x) is responsible for the hardness, m elting point, light sensitivity, and refractive index of the formed ber. For all measurements in this work, AgCl 0.3 Br 0.7 bers were used and were provided by the research group of Prof. Abraham Katzir (University of Tel Aviv). The mechanical, physical, and optical properties of AgCl 0.3 Br 0.7 are given in Tables I and II. 15 In contrast to m any other IR ber-optic materials, silver halide bers are exible, nontoxic, and insoluble in water and many organic solvents. Therefore, they are suitable for m any applications in the analytical and the medical area. However, due to the chemical properties of silver halides, certain precautions have to be considered: UV light (l , 400 nm) and contact with m etals less noble than silver or saltwater result in corrosion processes de-
0003-7028 / 00 / 5411-1629$2.00 / 0 q 2000 Society for Applied Spectroscopy
APPLIED SPECTROSCOPY
1629
TABLE I. Mechanical and physical properties of silver halide bers (AgCl 0.3 Br 0.7 ). Speci c weight Melting point Tensile strength Knoop hardness Minimal bending radius
6.39 g cm 2 3 412 8C 100 M Pa 15 kg mm 2 2 5 * diameter
grading to these bers. Nevertheless, for many applications a proper polymer coating can be applied to solve these problem s. Chemical Tapering Procedure. The sensing part of the ber with a diameter of 900 mm was dipped into Na 2S 2 O 3 solution contained in a Petri dish with a diameter of 10 cm corresponding to the length of the subsequently used glass ow cell. The dish was placed on an eccentric plate m ixer (Minishaker IKA MS 1). The rotary speed was set to approximately 60 turns per minute after experimental optimization. Due to the eccentric rotation of the dish, a wavelike movement of the thiosulfate solution was created. This movem ent ensured that only the m iddle section of the ber was in permanent contact with the etching solution, while the tapering zone to the left and to the right was only temporarily in contact with the etching agent (Fig. 1). This approach led to tapered bers with a cone-shaped decreasing ber diameter without indentations. All tapering experiments were performed with a 0.8 M thiosulfate solution, which led to tapering times from 5 to 17 m in, depending on the desired tapering ratio. A certain limitation of tapering silver halide bers is given by the resulting fragility of the material: bers tapered for a longer than 17 m in broke while being mounted into the glass ow cell. Sensor Setup. All measurements were performed with a Bruker Vector 22 Fourier transform infrared (FT-IR) spectrometer. The interface between the spectrometer and the ber optics was realized with the use of off-axis parabolic mirrors which focus the light onto the polished ber tip and at the distal ber end onto the detector element of an appropriate liquid nitrogen-cooled, mercurycadmium-telluride (MCT) broadband detector. X,Y,Z -
TABLE II.
O ptical properties of silver halide bers (AgCl 0.3 Br 0.7 ).
Transmission range Refractive index Practical numerical aperture
4 –20 mm 2.21 0.5
Attenuation losses at 5.25 mm (CO laser) at 10.6 mm (CO 2 laser)
1–2 dB m 2 1 0.5–1 dB m 2 1
ber positioners on both ber ends enabled accurate alignment in order to achieve maximum light throughput. Untapered and etched bers with a length of 28 cm and a diameter of 900 mm were used. The length of the tapered section (taper length) was 6 cm and was located in the middle section of the ber. Subsequently, the bers were mounted into a glass ow cell of 10 cm length with solution in- and outlet close to the cell ends. Both cell ends were sealed with silicone septa. The TeCE/hexane m ixtures were injected into the cell with a glass syringe. Spectra were acquired by averaging 100 scans at a spectral resolution of 4 cm 2 1. For each ber, spectra with 1, 2, 3, 5, and 10 vol % TeCE were recorded. RESULTS AND DISCUSSIO N The tapering experiments showed a linear dependence of the diameter on the etching time. The smallest diameter (tapering time 17 min) was about 240 mm, which corresponds to a taper ratio of 1:3.8. All bers that were tapered for a longer period broke when m ounted into the glass ow cell, although this problem could be improved in the future by a new cell design. Exemplary spectra of each TeCE dilution are shown in Fig. 2 for a selected silver halide ber, which was tapered for 10 min (taper ratio 1:1.4). The m ost signi cant and selective absorption band of TeCE in solution, is the n(C–Cl) vibration at 910 cm 2 1 . Therefore, this band was chosen to create calibration curves, evaluating the peak area vs. the TeCE concentration in solution. Figure 3 displays the obtained calibration curves for one untapered ber as well as for three tapered bers. Each concentration was measured ve times, and the er-
F IG . 1. Tapering procedure. The wavelike movement of the thiosulfate solution is created by an eccentric rotation of the plate. Fiber sections 1 and 3 are only temporarily in contact with the etching agent, while ber section 2 is etched perm anently. This leads to a cone-shape d decreasing diameter of the ber, with the thinnest part in section 2.
1630
Volume 54, Number 11, 2000
F IG . 2. IR spectra recorded with a silver halide ber tapered with aqueous 0.8 M thiosulfate solution for 10 m in. Related to the primar y (left) ordinate, spectra of various dilutions of tetrachloroethylene in hexane: (a) 2% TeCE, (b) 3% TeCE, (c) 5% TeCE, and (d ) 10% TeCE. An arbitrary off-set was added to the spectra. Spectrum e is related to the secondary (right) ordinate and shows the noise of the 100% line that was recorded with this ber.
ror bars are— though hard to see—included in the gure, demonstrating the repeatability of the measurements. The untapered ber provides almost perfect linearity over the whole concentration range, with a correlation of R 2 5 0.9991, which has also been demonstrated in several previous works. 2– 4 The graph shows that a tapering ratio of 1:1.6 already has a positive effect on the sensor response: the obtained peak areas are m ore than three times higher than the ones of the untapered ber. In addition, the slope of the calibration graph is signi cantly higher. As already demonstrated, the linear correlation decreases with increased tapering of the ber.16 The reason for this effect is still under discussion but may be attributed to variable outcoupling effects in the tapered section of the ber. The ber with a taper ratio of 1:3.1 continues this trend, resulting in a 53 higher sensor response than the untapered ber. However, the thinnest ber with a taper ratio of 1:3.8 shows by far the highest sensor response, with more than one order of m agnitude sensitivity gain compared to the untapered ber. The nonlinear behavior of the calibration curve is most pronoun ced at this taper
F IG . 4. Spectra of 5% TeCE in hexane, recorded once with an untapered silver halide ber (fat line) and once with a silver halide ber tapered for 15 min (thin line).
ratio as well. For an approximation of the achievable limit of detection, a linear regression t has been perform ed, resulting in r 2 5 0.9832 for the taper ratio of 1:1.6, r 2 5 0.9896 for the taper ratio of 1:3.1, and r 2 5 0.9399 for the taper ratio of 1:3.8. To outline the signi cant changes in spectra of untapered and tapered bers, Fig. 4 shows spectra of a 5% TeCE solution recorded with an untapered ber and a ber with taper a ratio of 1:3.1, respectively. For each ber the noise level [root mean squared (rm s)] in the range of the evaluated band at 910 cm 2 1 was calculated, and according to the 3*s criterion the estimated LODs were obtained (Table III). A prem ise for this evaluation method is a suf cient linearity of the calibration curve, which is not given in the case of the 17 min tapered ber. For this reason the calculated LOD for this ber is put in parentheses. As can be seen, the LOD for the untapered ber is about one order of magnitude higher than the ones obtained with tapered bers. On the basis of these ndings, an LOD for TeCE in comparison with the literature, where m easurements were performed with polymer-coated bers in aqueous solutions 5,17 [LOD som e hundreds of parts per billion (ppb)], the LOD for polymer-coated tapered bers can be expected in the range of some tens of ppb. If one evaluates the singlebeam spectra of the bers (Fig. 5), it is obvious that the ber with a taper ratio of 1:3.8 produces the highest noise, as the total light throughput is reduced by about 50% compared to an untapered ber. In conclusion, there are two competitive effects occurring for tapered bers: higher taper ratios lead to a higher sensor response (increasing numbers of internal re ections), but also to a higher noise level, because of a lower total light throughput due to outcoupling effects. For the demonstrated tapering method it seems that the optimum between these TABLE III. Estimated LODs for TeCE in hexan e for differently tapered bers, obtained with the 3*s criterion. Taper time 0 min
F IG . 3. Peak area of the n(C–Cl) absorption band of TeCE at 910 cm 2 1 vs. concentration of tetrachloroethylene for one untapered and three differently tapered silver halide bers. Each measurem ent was repeated ve times; the resulting error bars are included in the graph.
RMS 3.976E-05 Peak height (TeCE band at 910 cm 2 1 ) 0.01405 LOD (ppm) 85
10 m in
15 min
2.199E-05
4.722E-05
0.0719474 9
0.151023 9
17 min 1.104E-04 0.316887 (10)
APPLIED SPECTROSCOPY
1631
Fig. 6 sections of one untapered and three tapered bers with different taper ratios are shown. The white scaling line on the bottom of the images corresponds to a length of 100 mm. The SEM images show that the etching process maintains and even improves the surface quality since the initial roughness appears smoothed in the tapered sections. CO NCLUSION
F IG . 5. Single-beam spectra of an untapered and two differently tapered silver halide bers.
two effects resulting in the lowest LOD can be found at a tapering time between 15 and 17 m in, resulting in a diameter of the tapered ber between 240 and 290 mm. For obtaining information on the surface quality of the bers, scanning electron m icrographs were recorded. In
F IG . 6.
1632
Etching of silver halide bers with an aqueous thiosulfate solution was introduced as an innovative tapering technique. Calibration curves of TeCE in hexane were recorded with a set of differently tapered ber sensors coupled to an FT-IR spectrometer. The obtained results demonstrate that chemically tapering a ber with an original diameter of 900 mm by a ratio of 1:1.6 to 1:3.8 results in a sensitivity increase of one order of m agnitude. SEM images point out that the etching process maintains and even improves the surface quality in the tapered section compared to the untapered ber. The summ ary of these rst measurements indicates the potential of this m ethod: by combining this tapering technique with proper polymer enrichment layers applied to the ber surface, it is m ost probable that selective m id-
SEM images of an untapered and three differently tapered silver halide bers.
Volume 54, Number 11, 2000
IR ber-optic m easurements of VOCs in the sub-ppb concentration level will be possible in the near future. ACK NOWLEDGMENTS This work was nancially supported by the European Union under Project MAS3-CT97-0157 and by the Austrian Federal M inistry of Education, Science and Culture (BMWF). 1. S. Ramamoorthy, Chlorinated Organic Compounds in the Environment (Lewis Publishers, Boca Raton, Florida, 1997). 2. R. Krska, R. Kellner, U. Schiessl, M . Tacke, and A. Katzir, Appl. Phys. Lett. 63, 1 (1993). 3. B. Mizaikoff, R. Go¨ bel, R. Krska, K. Taga, R. Kellner, M. Tacke, and A. Katzir, Sens. Actuators B 29, 58 (1995). 4. J. E. Walsh, B. D. MacCraith, M. Meaney, J. G. Vos, F. Regan, A. Lancia, and S. Artjushenko, Proc. SPIE-Int. Soc. Opt. Eng. 2508, 233 (1995). 5. M . Jakusch, B. Mizaikoff, and R. Kellner, Proc. SPIE-Int. Soc. Opt. Eng. 3105, 283 (1997). 6. S. E. Plunkett, S. Propst, and M. S. Braiman, Appl. Opt. 36, 4055 (1997).
7. S. E. Plunkett, R. E. Jonas, and M. S. Braiman, Biophys. J. 73, 2235 (1997). 8. H. Tai, H. Tanaka, and T. Yoshino, Opt. Lett. 12, 437 (1987). 9. A. Bornstein, M . Katz, A. Baram, and D. Wolfman, Proc. SPIEInt. Soc. Opt. Eng. 1591, 256 (1991) 10. R. J. Burger, P. J. M elling, W. A. Moser, and J.-R. Berard, Proc. SPIE-Int. Soc. Opt. Eng. 1591, 246 (1991). 11. L. C. Shriver-Lake, G. P. Anderson, J. P. Golden, and F. S. Ligler, Anal. Lett. 25, 1183 (1992). 12. J. Harrington, Proc. SPIE-Int. Soc. Opt. Eng. 1591, 2 (1991). 13. M . G. Drexhage and C. T. Moynihan, Sci. Am . 11, 110 (1988). 14. (a) Infrared Fiber Optics I, SPIE Vol. 1048, J. A. Harrington and A. Katzir, Eds. (SPIE, Bellingham, Washington , 1989); (b) Infrared Fiber Optics II, SPIE Vol. 1228, J. A. Harrington and A. Katzir, Eds. (SPIE, Bellingham, Washington, 1990); (c) Infrared Fiber Optics III, SPIE Vol. 1591, J. A. Harrington and A. Katzir, Eds. (SPIE, Bellingham , Washington, 1991). 15. S. Shalem, A. German, N. Barkay, F. Moser, and A. Katzir, Fiber Integrated Opt. 16, 27 (1997). 16. A. Bornstein, M . Katz, A. Baram, and D. Wolfman, Proc. SPIEInt. Soc. Opt. Eng. 1591, 256 (1991). 17. B. Mizaikoff, Proc. SPIE-Int. Soc. Opt. Eng. 3849, 7 (1999).
APPLIED SPECTROSCOPY
1633