- Email: [email protected]
Spatially resolved analyte mapping with time-of-flight optical sensors
593
trends in analytical chemistry, vol. 17, no. 10, 1998
[ 36 ] C.S. Creaser and B.L. Williamson, J. Chem. Soc. Chem. Commun. ( 1994 ) 1677. [ 37 ] C.S. Creaser, B.L. Williamson, J. Chem. Soc. Perkin Trans. 2 ( 1996 ) 427. [ 38 ] A. Colorado, J. Brodbelt, Anal. Chem. 66 ( 1994 ) 2330.
[ 39 ] R.E. March and R.J. Hughes, Quadrupole Storage Mass Spectrometry, Wiley Interscience, New York, 1989. [ 40 ] J.L. Stephenson, S.A. McLuckey, Anal. Chem. 68 ( 1996 ) 4026.
Spatially resolved analyte mapping with time-of-£ight optical sensors Radislav A. Potyrailo**, Gary M. Hieftje*
Department of Chemistry, Indiana University, Bloomington, IN 47405, USA Simultaneous real-time acquisition of analytical data from several locations is attractive in a variety of applications. This brief review traces the evolution of approaches to such measurements. Greatest emphasis is placed on optical time-of-£ight chemical detection when signals are multiplexed from several point sensors or when the measurements are taken along the length of a single continuous extended-length `distributed' sensing element. The use of sensors featuring extended-length continuous chemically sensitive optical ¢bers offers detection arrangements for which there is no counterpart in conventional chemical sensor technologies. z1998 Elsevier Science B.V. All rights reserved. Keywords: Optical time-of-£ight; Chemical sensor; Optical ¢ber; Multiplexing; Distributed; Signal-to-noise; Discrete location; Continuous sensing element
1. Introduction Spatially resolved mapping of chemical constituents is an important need in a variety of practical applications. For example, spatially resolved analyte monitoring can simultaneously indicate and locate when an accepted level of exposure to toxic or explosive spe*Corresponding author. **Present address : Characterization and Environmental Technology Laboratory, Corporate Research and Development, General Electric Company, P.O. Box 8, Building K-1, Schenectady, NY 12301, USA.
cies has been exceeded, can track a source of contamination in an industrial or technological process, can follow the formation and movement of environmental pollutants, and can perform early nondestructive detection and location of corrosion ( such as in advanced structural components of space shuttles, aircraft, and civil engineering structures ). For these and other applications, information about the spatial distribution of species of interest can be obtained when signals are multiplexed from different discrete point sensors or when the measurements are taken along the entire length of a single continuous extended-length sensing element. This brief review traces the evolution of approaches to such measurements. It is demonstrated that the use of ¢ber optics provides a unique solution to this monitoring problem. The combination of a `distributed' sensor ( featuring an extended-length continuous chemically sensitive optical ¢ber ) and optical timeof-£ight (OTOF ) techniques offers detection schemes for which there is no counterpart in conventional chemical sensor technologies.
2. Optical time-of-£ight approaches for spatially resolved analyte mapping For spatially resolved chemical measurements, a variety of techniques can be adapted from mature approaches for diagnosing ¢ber-optic communication networks and physical sensors [ 1 ]. In these areas, OTOF techniques are the most widely used because they offer strong advantages over other methods. These advantages include improved quantitation, operation with multimode ¢bers, simple signal-gener-
0165-9936/98/$ ^ see front matter PII: S 0 1 6 5 - 9 9 3 6 ( 9 8 ) 0 0 0 6 4 - 8
ß 1998 Elsevier Science B.V. All rights reserved.
TRAC 2460 30-10-98
594
trends in analytical chemistry, vol. 17, no. 10, 1998
Fig. 1. Con¢gurations of optical time-of-£ight chemical sensors for spatially resolved measurements. ( A ) Star con¢guration for multiplexing point sensors in transmission mode, ( B ) star con¢guration for multiplexing point sensors in backpropagation mode, ( C ) serial con¢guration for multiplexing point sensors in backpropagation mode, ( D ) distributed sensor con¢guration. FOC, ¢beroptic coupler; DL, ¢ber-optic delay line; PD, photodetector; MS, monitoring site.
ation and processing methods, and relatively inexpensive opto-electronic components. These advantages of OTOF methods are transferable to the ¢eld of chemical sensing. With OTOF-based sensing, it is possible to locate zones along a ¢ber where analyte-induced changes in absorption or £uorescence take place, to determine the magnitude of these variations, and to relate the magnitude of the variations to the concentration of an analyte. The different OTOF approaches for spatially resolved analyte mapping shown schematically in Fig. 1 have a common concept. In OTOF methods, a pulse of light is periodically launched into a ¢ber or ¢bers and the analytical information is derived from the amplitude and temporal characteristics of the light which interacts with the sensing element and then
reaches a detector. The amplitude of the pulsed signal is affected by the analyte concentration at each of the point sensors ( Fig. 1A^C ) or along the continuous sensing ¢ber ( Fig. 1D ). Information about the location where sensing occurs is obtained from the measured time delay between the launched and detected light pulses. In multiplexing strategies, used with discrete point sensors, this time delay is controlled by connecting each individual sensor to a detector through a different length of a ¢ber. The signal from each of the multiplexed sensors can be obtained in either a transmission ( Fig. 1A ) or backpropagation ( Fig. 1B,C ) mode. When the OTOF technique is based on the detection of backpropagated radiation, it is known as optical time-domain re£ectometry (OTDR ) [ 2 ]. Very similar mathematical expressions describe the signals for backscatter and backpropagated £uorescence in multiplexed point and distributed sensors [ 3 ]. Thus, although backpropagated £uorescence is not strictly a backscatter signal, the term OTDR can also be applied to £uorescence detection. Point sensors are commonly multiplexed by means of star ( Fig. 1A,B ) or serial ( Fig. 1C ) arrangements; features of these multiplexed point sensors are summarized in Tables 1 and 2. The number of point sensors used in a star OTDR-based multiplexing topology can be limited by the power budget of the system; at most, 25^30 sensors can be supported by a single laser source with acceptable detection sensitivities [ 19 ]. This limitation arises from losses at each multiplexing point [ 20 ]. These losses can be reduced signi¢cantly in a serially multiplexed topology if a single extendedlength ¢ber is reclad over short regions with a chemically sensitive reagent [ 12,15^18 ]. Unfortunately, replacement of the original cladding with a chemically sensitive ¢lm in such a `quasi-distributed' sensor creates a number of problems, including reclad lengths of only a few centimeters long, non-uniformity of the reclad sections, scattering losses, and loss of mechanical £exibility. In addition, multiplexing of short sections is still limited to spatially discrete measurements ( i.e. the continuity parameter D in Table 2 is much less than unity ). All these problems can be eliminated if a single continuous extended-length sensing ¢ber is used ( Fig. 1D ). In addition, a distributed sensor can provide a more complete and cost-effective spatial pro¢le of chemical constituents. Several such distributed sensors have been reported ( see Table 3 ).
TRAC 2467 30-10-98
595
trends in analytical chemistry, vol. 17, no. 10, 1998
Table 1 Features of reported OTOF star multiplexed point sensors Analyte species or measured parameter
Detection method
Design of a sensing element
Number of point sensors ( ¢ber length, m )
Ref.
Acetylene
Absorption OTOF in transmission mode ( 1.528 Wm ) Two-photon £uorescence OTDR ( exc. at 0.59 Wm ) Absorption OTDR ( 0.68 or 0.78 Wm )
A 7 cm long absorption cell coupled to a single mode ¢ber
4 ( 1^25 )
[4]
Reagent-free ¢ber tip in analyte solution A ¢ber section reclad with a sol-gel ¢lm doped with phosphomolybdic acid for evanescent-wave sensing or the reagent is deposited onto the distal end of the ¢ber Reagent-free ¢ber tip immersed into monitored sample An etched cavity at the distal end of a graded index multimode ¢ber packed with sol-gel-derived silica doped with trimethinecyanine
3^10 ( 40-140 )
[5]
NRb
[ 6^8 ]
8 (NR )
[9]
3 ( 100^300 )
[ 10 ]
K-NPOa Hydrazine, nitrogen dioxide, nitrogen tetroxide
Degree-of-cure of epoxy resin pH
a b
Fresnel re£ection OTDR ( 0.904 Wm ) Absorption OTDR ( 0.85 Wm )
K-NPO = 2-1 (-naphthyl )-5-phenyloxazole. NR = not reported.
3. Transduction mechanisms and fabrication strategies for distributed chemical sensors Most distributed chemical sensors reported to date have been based on evanescent-wave detection. In this sort of device, the detected signal is dictated by analyte-induced changes in the refractive index, absorption, or £uorescence of an analyte-permeable cladding. In order to be detected, of course, the analyte must lie within the penetration depth of the evanescent ¢eld produced by a guided mode( s ) in the ¢ber ( cf. Table 3 ), requiring the ¢ber cladding to be permeable to the analyte. The use of analyte-induced variations in refractive index of the cladding [ 21,22,27 ] is limited in scope by the small number of species that can be determined using this approach. As a result, indirect [ 23 ], including reagent-based [ 3,24,25 ], sensing approaches have received more attention. The ¢rst extended-length chemically sensitive ¢bers useful for spatially resolved analyte mapping were produced by Blyler et al. [ 28,29 ] in the late 1980s. They doped special formulations of liquid silicones with analyte-sensitive reagents, applied these solutions onto the silica core of a multimode optical ¢ber to form the ¢ber cladding, and cured the cladding in-line during the ¢ber-drawing operation. Because the analyte-sensitive reagent was distributed along the entire length of the ¢ber, analyte sensing could
occur over extended regions. Unfortunately, as admitted by the inventors [ 28^30 ], the technology suffered from several serious drawbacks. First, the photocuring chemistry involved the dissolved reagent and resulted in incomplete reversibility of the sensor. Second, the technology was complex, very expensive, and limited to those who have a ¢ber-optic drawing facility. Also, an organic reagent doped in the cladding was exposed to strong UV light or high temperature required for the crosslinking of silicone during the ¢ber manufacture [ 28 ]; these conditions were likely to cause partial degradation of the reagent. To overcome these limitations, an attempt was made also to chemically immobilize reagents onto the surface of a commercially available plastic optical ¢ber [ 31,32 ]. Although the technique was useful for short ¢ber sections, it was not attractive for distributed sensing over extended regions because of scattering losses in the modi¢ed ¢bers. In addition, the indicator had to be synthesized to contain the appropriate functional groups for immobilization. Recently, a more attractive strategy for the fabrication of extended-length sensors was introduced [ 33 ]. The technique involves immobilization of a reagent directly into the original ( crosslinked ) silicone cladding of a conventional plastic-clad silica (PCS ) ¢ber. For indicator immobilization, the ¢ber is soaked in a solution containing the required indicator dissolved in a non-polar solvent. After removal from the indicator
TRAC 2467 30-10-98
596
trends in analytical chemistry, vol. 17, no. 10, 1998
Table 2 Features of reported OTOF serially multiplexed point sensors Analyte species or measured parameter
Detection method
Design of a sensing element
Continuity parameter Da
Ref.
RHb
Rayleigh OTDR ( 0.85 Wm ) Absorption OTDR
A PCS ¢ber segment reclad with microporous silica for evanescent-wave sensing A PCS ¢ber section reclad with a reagentdoped cellulose acetate ¢lm A U-shaped unclad segment of silica ¢ber with a V0.5 mm bend radius for evanescentwave sensing Supported catalytic point sensor converted analyte concentration change into temperature change A multimode ¢ber section reclad with a cobalt chloride-doped gelatin ¢lm for evanescentwave sensing A PCS ¢ber section reclad with a £uoresceindoped sol-gel ¢lm for evanescent-wave sensing A PCS ¢ber section reclad with a £uorophoredoped sol-gel ¢lm for evanescent-wave sensing
4 cmU3 / 130 m
[ 11 ]
25 cmU3 / 25 m
[ 12 ]
1 mmU2 / 500 m
[ 13 ]
NR
[ 14 ]
5 cmU4 / 100 m
[ 15 ]
NRU9 / 100 m
[ 16 ]
4 cmU4 / 9 m; 4 cmU4 / 7 m
[ 17 ]
2 mmU4 / 8 km
[ 18 ]
Alkaline and acid vapors RH ( via RIc )
Rayleigh OTDR ( 0.85 Wm )
Methane ( via temperature )
Rayleigh OTDR
RH
Absorption OTDR ( 0.67 Wm, reference at 0.85 Wm ) Fluorescence OTDR ( exc. at 0.44 Wm ) Fluorescence OTDR ( exc. at 0.424, 0.5, or 0.56 Wm ) Rayleigh OTDR ( 1.31 Wm )
pH pH, quinone RH ( via RI )
A polished single-mode half-coupler with a polyethylene oxide overlay for evanescentwave sensing
a
Continuity parameter D is given by D = lN /L, where l is the length of a sensing region, N is the number of sensors, and L is the total length of the ¢ber which includes the sensing regions and the ¢ber link. b RH = relative humidity. c RI = refractive index.
solution, the ¢ber is washed with pure solvent and dried [ 33,34 ]. In contrast to earlier reported methods for fabricating extended-length indicator-based sensing elements [ 28,29,31,32 ] the new strategy does not suffer from photo- or temperature-induced degradation of the indicator, offers reversible sensor response, simpli¢es the fabrication process, lowers the construction cost of the sensing ¢ber, does not induce scattering losses upon indicator immobilization, and permits immobilization of several reagents in adjacent sections of a single optical ¢ber for multianalyte determinations. The silicone cladding of a conventional PCS optical ¢ber, a copolymer of vinyl-terminated poly( dimethylsiloxane ) and poly( dimethyl-methylhydrosiloxane ), is an attractive immobilization matrix for a wide variety of reagents and opens up new avenues of sensor design [ 34 ]. The silicone cladding material offers a number of advantages over other common silicones used as host matrices for immobilization of indicators. These advantages accrue from both the composition and crosslink density of the silicone cladding. First, the cladding makes dynamic £uorescence quenching of an immobilized £uorophore more
ef¢cient than is provided by common room-temperature-vulcanized (RTV ) silicones. For example, the Ru( phen )3 cation exhibits the highest oxygenquenching constant reported when immobilized in a silicone elastomer [ 35 ] ( Fig. 2 ). Also, the high network density in the silicone cladding material leads to reduced aggregation of immobilized indicator molecules. In a chemical sensor that employs dynamically quenched £uorescence, reduced reagent aggregation results in more effective quenching. Second, the high crosslinking density in the cured elastomer affords the cladding a high degree of resistance to aggressive media ( such as alkaline solutions ) and therefore protects immobilized reagents from leaching or decomposition [ 38 ]. The resistance of the reagent-doped silicone cladding to highly alkaline solutions is important in sensors intended for high ammonia concentrations ( cf. Fig. 3 ). Previous attempts to develop a sensor useful for such determinations, without sample dilution, suffered from leaching of the immobilized indicator [ 39 ]. Longterm stability tests of a PCS ¢ber modi¢ed with a pH indicator in aqueous solutions of 14 M NH4 OH and 10 M NaOH revealed no observable washout
TRAC 2467 30-10-98
597
trends in analytical chemistry, vol. 17, no. 10, 1998
Table 3 Features of reported OTOF distributed sensors Analyte species or measured parameter
Detection method
Design of a sensing element
Length of the sensing region ( m )
Ref.
Any substance via RI
Rayleigh OTDR
NR
[ 21 ]
Oil ( via RI )
Rayleigh OTDR ( 1.31 Wm )
2000
[ 22 ]
Water
Rayleigh OTDR
60
[ 23 ]
pH, moisture
Absorption OTDR ( 0.85 Wm, reference at 1.30 Wm ) Fluorescence OTDR ( exc. 0.532 Wm )
A multimode ¢ber with analyte permeable cladding for evanescent-wave sensing Specially drawn single-mode eccentric core ¢ber with silicone coating on top of the glass ¢ber cladding for evanescent-wave sensing A multimode optical ¢ber in intimate contact with a hydrogelcoated rod by a helically wound thread. Hydrogel swelling induces the microbending loss of light within the ¢ber through a change in strain caused by the con¢ning thread Specially drawn multimode ¢ber with dye-doped polymer cladding for evanescent-wave sensing Conventional PCS ¢ber with chemically modi¢ed original silicone cladding for evanescent-wave sensing Conventional PCS ¢ber with chemically modi¢ed original silicone cladding for evanescent-wave sensing Conventional PCS ¢ber with chemically modi¢ed original silicone cladding for evanescent-wave sensing
20
[ 24 ]
48
[ 3,25 ]
40
[ 26 ]
40
[ 26 ]
Oxygen, atmospheric pressure Ammonia Ammonia
Absorption-modulated £uorescence OTDR ( exc. 0.532 Wm ) Bidirectional OTDR ( exc. 0.532 Wm )
of the immobilized dye over a period of several months [ 38 ]. Dye-doped silicone-cladding sensors for continuous humidity monitoring, possibly useful for early detection of moisture-induced corrosion in structural components in civil, aerospace, and electrical engineering, have been examined [ 35 ]. While liquid water has a very low solubility in silicone resins, such resins are highly permeable to water vapor [ 40 ]. The dynamic response curve for a Nile-redmodi¢ed PCS ¢ber is compared with that of a commercial humidity monitor in Fig. 4. The sorption and desorption of water vapor in the silicone ¢ber cladding was completely reversible, as veri¢ed by monitoring the hypochromic effect at 522 nm. More than 30 indicators of different sorts ( polycyclic aromatic hydrocarbons, rhodamines, coumarins, cyanides, sulfonaphthaleins, azo-dyes, porphyrins, and others ) have been successfully immobilized into the silicone cladding of PCS ¢bers, by utilizing a variety of non-polar solvents [ 34 ]. The molecular weight of the immobilized indicators spanned almost an order of magnitude from 140 ( nitrophenol ) to 1200 ( a Ru
complex ). It is possible that this range could be broader; however, it encompasses most £uorescence and absorption indicators.
Fig. 2. Stern^Volmer calibration plots for quenchingbased oxygen sensors based on the Ru( phen )3 complex immobilized in different matrices: ( A ) silicone cladding of a PCS optical ¢ber, ( B ) GE RTV 118 silicone ¢lm ( experimental points taken from Ref. [ 36 ]), ( C ) single-component RTV acetic acid releasing silicone ¢lm ( experimental points taken from Ref. [ 37 ]).
TRAC 2467 30-10-98
598
trends in analytical chemistry, vol. 17, no. 10, 1998
These requirements include high signal levels, a fairly uniform detection limit over the length of the sensing ¢ber, and high spatial resolution. Recently, however, several signal generation and processing methods have been devised to address these requirements. 4.1. Raising signal levels in OTOF distributed sensors
Fig. 3. Calibration curve for a solid-state sensor with the phenol-red-modi¢ed optical ¢ber, a green light-emitting diode, and a silicon photodiode. The 95% con¢dence intervals are given for three replicate measurements during a month of experiments. The total exposure time to 0.006^0.6 M ammonium ion solutions in these measurements was 4 h. Concentration ranges of ( I ) environmental and ( II ) industrial interest.
4. Recent progress in OTOF methods in spatially resolved analyte mapping The analytical signal in an OTOF measurement system with a distributed sensing region was ¢rst formulated for Rayleigh backscatter and then for Raman backscatter. If the relationship is generalized also for £uorescence detection, the backpropagated impulse response from a distance l along the ¢ber excited by a N-pulse is given by [ 3 ] Z l 1 P l rPo S lexpf3 Kf z Kb zdzg 0
Scattering-based OTDR suffers from extremely low signal levels. The Rayleigh backscattered signal is typically 104 ^105 times weaker than the forward traveling pulse [ 2 ]; Raman backscatter is about 103 times weaker still [ 41 ]. As a result, recording the backscattered signal is often a problem since it requires sophisticated detection schemes, high-power lasers, and time-consuming signal averaging techniques. The S /N-enhancement methods devised for distributed physical sensors ( e.g. based on pseudonoise, polarimetric techniques, and nonlinear optical effects [ 42,43 ]) are limited to applications involving singlemode optical ¢bers and, thus, are not useful for most distributed chemical sensors. A new approach to signal generation and processing was recently introduced to enhance signal levels in distributed chemical sensors [ 26 ]. In this method, an analyte-sensitive absorbing dye and an analyte-insensitive £uorophore were immobilized together into the ¢ber cladding. These reagents were chosen so their absorption bands overlap and the excitation wavelength was selected to be within this overlap region. The £uorescence intensity of an analyte-insensitive £uorophore was then monitored as a function of the evanescent-wave absorption of an analyte-sensitive
where Po is the excitation power in the input pulse coupled into the ¢ber, r is the ratio of the transmitted to the re£ected optical power in the ¢ber-optic coupler ( cf. Fig. 1D ), S is a constant that depends on the local numerical aperture (NA ) of the ¢ber, the £uorescence quantum yield of the immobilized £uorophore ( in a £uorescence-based sensor ), or on the Rayleigh and Raman scattering parameters of the ¢ber core and cladding ( in a scattering-based sensor ), and Kf ( z ) and Kb ( z ) are the attenuation coef¢cients of the forward and backward traveling pulses, respectively. In £uorescence-based sensors, both Kf and Kb have two components, one each for the excitation and emission wavelengths. Traditional signal-processing techniques used with OTOF-based chemical sensors do not meet the requirements for spatially resolved chemical sensing.
Fig. 4. Dynamic response to change in relative humidity: ( b ) PCS optical ¢ber modi¢ed with Nile red and ( a ) a commercial humidity monitor.
TRAC 2467 30-10-98
599
trends in analytical chemistry, vol. 17, no. 10, 1998
Fig. 5. Spatially resolved distributed measurements with OTOF absorption-modulated £uorescence. Regions 2, 3, 4, and 9 of a continuous 40 m long distributed ammonia sensor were exposed to 213 ppm of ammonia while the rest of the ¢ber was in air. ( A ) Collected intensity waveform I is normalized by the intensity waveform I o of the whole ¢ber in absence of analyte. ( B ) Analyte-induced increase in the local absorption coef¢cient in decibels as a function of ¢ber length. The waveform in B is smoothed by a running-average function.
indicator. In essence, the analyte-sensitive absorber `steals' excitation light from the £uorophore, so the emission of the £uorophore is inversely dependent on the analyte concentration. The low £uorophore concentration does not contribute appreciably to the ¢ber attenuation at the excitation and emission wavelengths and therefore does not lead to a loss in dynamic range of the sensor. The £uorophore can be chosen with a £uorescence lifetime shorter than the pulse width of the source in order not to degrade spatial resolution ( see Section 4.4 ). This concept was experimentally veri¢ed with a distributed sensor constructed from a 40 m long continuous chemically sensitive optical ¢ber. This sensing element was produced by immobilization of an ammonia-sensitive absorbing reagent ( phenol red )
and an analyte-insensitive £uorophore ( rhodamine 640 ) into the silicone cladding of a PCS ¢ber. Typical results of these experiments are plotted in Fig. 5A as the ratio of the measured £uorescence intensities in the presence and in the absence of the analyte. By plotting the analyte-induced increase in the local absorption coef¢cient ( in decibels ) as a function of time ( distance along the ¢ber ) it is possible to spatially resolve analyte concentrations ( Fig. 5B ). When this data representation is employed, ¢ber regions that cause an increased attenuation ( because of higher ammonia concentration ) exhibit noticeable dips below the constant background. The vertical scale on a plot such as that in Fig. 5B could be easily converted by calibration from analyte-induced ¢ber attenuation to analyte concentration.
TRAC 2467 30-10-98
600
trends in analytical chemistry, vol. 17, no. 10, 1998
The resulting signal levels in absorption-modulated £uorescence OTOF detection were well above those found in OTDR methods that rely upon backscattering for spatially resolved detection [ 2,24 ]. The degree of enhancement was between 10- and 120-fold and was dependent on the concentration of immobilized £uorophore and on its quantum yield. With this technique, the range of species detectable with distributed £uorescence sensors might be expanded to include analytes previously detectable only with attenuationbased methods, and therefore dif¢cult to monitor with an OTOF distributed sensor. By using £uorescence rather than backscatter detection, it is possible also to increase the speed of data collection by reducing the number of collected waveforms that must be averaged, and to use more compact and less powerful light sources. 4.2. Improvements in S /N
Methods have recently been introduced that help overcome the loss in S /N that usually accompanies the use of a long ¢ber [ 26 ]. The loss in S /N with ¢ber length [ 2 ], caused by the ¢nite attenuation coef¢cient of the ¢ber material, usually limits the useful length of an OTDR sensor. To overcome this limitation, bidirectional detection was used. The light was launched sequentially into the two ends of a 40 m long continuous chemically sensitive optical ¢ber and the backpropagated portion was collected from the same end into which the respective pulse was launched [ 26 ]. Because the two pulses provided complementary information, only the ¢rst half of each of the collected waveforms was used for analyte quantitation. As shown in Fig. 6A, the S /N degraded rapidly with distance from the respective launch end of the ¢ber. The relative standard deviation (RSD ) dropped from its initial level of 0.5% to between 12 and 15% at the distant ends of the ¢ber. When bidirectional detection was used, the RSD at the far end of the chemically sensitive ¢ber was reduced by 3^4.5-fold. The result of combining the ¢rst halves of the waveforms collected from the different ¢ber ends is illustrated in Fig. 6B. The middle section of the ¢ber had an unchanged 3% RSD whereas the two ends yielded the highest S /N. The bidirectional pulse-launching method to improve S /N for measurements over the length of a sensing ¢ber might be especially useful for sensors based on analyte-induced changes in ¢ber attenuation; in this type of sensor the S /N decays rapidly with distance from the launch end of the ¢ber.
4.3. OTOF measurements with dynamically quenched £uorophores
A limitation of absorption-based OTOF distributed sensors is the accumulative analyte-induced light attenuation that they suffer. A local increase in evanescent-wave absorbance by the immobilized reagent leads to a drop in light intensity available for propagation farther down the ¢ber. The dynamic range can therefore be limited by high analyte concentrations along the ¢ber. Sensors based on immobilized dynamically quenched £uorophores are free from this limitation. Such a sensor does not exhibit a change in attenuation coef¢cient upon exposure to a quencher. Rather, the £uorescence lifetime is a function of the quencher concentration, as described by the Stern^Volmer equation. The variable £uorescence lifetime then changes the emission intensity of the immobilized £uorophore detected with an OTOF system [ 3 ]. The ¢rst distributed OTOF sensor based on dynamically quenched £uorescence was developed by immobilizing a far-red emitting £uorophore ( tetraphenylporphyrin ) onto a 48 m long PCS ¢ber and was used for spatially resolved oxygen mapping [ 25 ] ( Fig. 7A ). The £uorophore was selected to have a short £uorescence lifetime ( 10 ns ) to provide adequate spatial resolution. Numerical simulations were performed using known parameters of the immobilized indicator and of the detection system to predict the observed intensity waveform in spatially resolved measurements ( Fig. 7B ). 4.4. Spatial resolution of distributed OTOF sensors
Spatial resolution in a distributed sensor should be as great as possible to provide the maximum number of resolved local variations of analyte concentration along the ¢ber length and to make the distributed sensor as cost-effective as possible compared to multiplexed point sensors. Spatial resolution of an absorption-based sensor is limited by the pulse width of the primary light source. For example, a laser having a 1-ns pulse width can provide roughly 10-cm spatial resolution in an absorption-based sensor [ 24 ]. In £uorescence-based sensors, spatial resolution is limited by the pulse width of the source, the impulse response ( excited-state lifetime ) of the immobilized £uorophore, and the bandwidth of the £uorescence that is detected [ 44 ]. Thus, for the best spatial resolution, it is desirable to use a narrow-bandpass inter-
TRAC 2467 30-10-98
601
trends in analytical chemistry, vol. 17, no. 10, 1998
Fig. 6. Spatially resolved analyte monitoring with OTOF by means of a bidirectional distributed sensor. Regions 2, 3, 6, and 7 of a continuous 40 m long distributed ammonia sensor ( similar to that in Fig. 5 ) were exposed to 100 ppm of ammonia while the rest of the ¢ber was in air. The returned waveforms were collected sequentially from end 1 and end 2 of the ¢ber for three ammonia-air cycles. The waveforms collected from end 2 were reversed and plotted as if they had been measured from end 1. ( A ) Collected intensity waveforms I are normalized by the waveforms I o of the whole ¢ber in absence of analyte. ( B ) Combined ¢rst halves of each of the normalized waveforms. Every 10th data point in ( A ) and ( B ) is shown with an error bar, which represents one standard deviation from the mean.
ference ¢lter and a £uorophore with a short £uorescence lifetime. This approach can be useful with a sensor that utilizes static quenching [ 26 ]. Unfortunately, this strategy does not work with a sensor utilizing a dynamically quenched £uorophore; the sensitivity of such a sensor usually drops when a £uorophore of very short lifetime is used. However, spatial resolution in a sensor that utilizes a reagent with a long £uorescence lifetime can be improved to its theoretical limit by using appropriate signal processing [ 44 ]. This processing involves deconvolution of the returned waveform with the laser pulse width and with the analyte-dependent £uorescence lifetime of the immobilized £uorophore. The method retrieves information about the analyte-modulated £uorescence
lifetime of the immobilized £uorophore at any location along the ¢ber. Several practical limitations often make deconvolution dif¢cult, noise usually being the most serious one [ 45 ]. If deconvolution is not used, the spatial resolution vlmin of a detection system is the smallest resolved distance between two ¢ber locations given by [ 1] vlmin cdb = nf nb
2
where c is the velocity of light in vacuum, nf and nb are the refractive indices of the ¢ber core at the wavelengths of the forward ( excitation wavelength ) and backward ( analytical wavelength ) propagated pulses,
TRAC 2467 30-10-98
602
trends in analytical chemistry, vol. 17, no. 10, 1998
Fig. 7. Spatially resolved oxygen determinations with the distributed oxygen sensor. In experiments I^IV several regions of a continuous 48 m long distributed oxygen sensor were exposed to nitrogen or oxygen, while the rest of the ¢ber was in air: I, regions 2 and 4 in nitrogen; II, regions 2 and 4 in oxygen; III, region 11 in nitrogen; IV, regions 11 and 13 in oxygen. ( A ) Collected intensity waveforms I are normalized by the intensity waveforms I air of the whole ¢ber in air. ( B ) Comparison of experimental ( data points ) and calculated ( solid lines ) results for spatially resolved oxygen monitoring.
and db is the width of the backpropagated pulse. The number of sensing regions N resolved with such a distributed sensor of length L is given by [ 3 ] N nf nb L=cdb
3
Several OTDR systems designed speci¢cally for diagnostic applications in ¢ber-optic communications
( spectral region from 0.85 to 1.55 Wm ) have been adapted for spatially resolved analyte mapping [ 11,13,18,23,24 ]. Typical pulse widths of conventional OTDR systems used for chemical detection range from 1 to 100 ns [ 13,18,24 ], which corresponds to a spatial resolution from 0.1 to 10 m. Such systems require near-infrared reagents or refractive-index detection.
TRAC 2467 30-10-98
603
trends in analytical chemistry, vol. 17, no. 10, 1998
5. Conclusions and future prospects Chemical detection represents a strong potential market for distributed sensors, after those for temperature, pressure, strain, displacement, and intruder detection [ 46 ]. However, activity in this direction has been limited by dif¢culties in fabricating sensing ¢bers of extended length, by low signal levels in backscatter-detection schemes, and by poor spatial resolution in £uorescence-based sensors. As new fabrication technologies for extended-length ¢bers are invented, as methods for advanced signal processing are designed, and as techniques for improving the spatial resolution of £uorescence-based sensors are devised, a speedier implementation of this technology can be expected. The basic and improved OTOF methods can be combined to advantage with other techniques. One of the possibilities is to couple spatially resolved OTOF detection with determination of the modal power distribution in a multimode waveguide [ 47 ]. Schemes based not only on OTOF techniques but also on other strategies also could be developed on the basis of amplitude-modulation and phase-modulation approaches. In the future, it should be possible to use optical ¢bers with alternative cladding materials, to allow different reagents to be immobilized, so a wide range of species in gaseous and aqueous phase can be determined over extended, remote areas. Spinoffs of this technology might be useful also in other areas of chemical sensing. For example, extended-length sensing elements could be used not only for distributed sensing, but also in point sensing; sensing segments of a desired length could be cut from an already prepared and calibrated ¢ber spool and used without further calibration.
Acknowledgements This work was supported in part by the National Institutes of Health through Grant GM 53560. R.A.P. acknowledges support from a McCormick Science Grant provided by the College of Arts and Sciences, Indiana University.
References [ 1 ] A.D. Kersey, in E. Udd ( Editor ), Fiber Optic Sensors: An
Introduction for Engineers and Scientists, Wiley, New York, 1991, p. 325. [ 2 ] M. Tateda, T. Horiguchi, J. Lightwave Technol. 7 ( 1989 ) 1217. [ 3 ] R.A. Potyrailo, G.M. Hieftje, Anal. Chem. 70 ( 1998 ) 1453. [ 4 ] D.R. Hjelme, S. Neegaard, J.S. Rambech, N. Ryen, Proc. SPIE ^ Int. Soc. Opt. Eng. 2507 ( 1995 ) 113. [ 5 ] R.L. Steffen, F.E. Lytle, Anal. Chim. Acta 215 ( 1988 ) 203. [ 6 ] C. Klimcak, G. Radhakrishnan, B. Jaduszliwer, Proc. SPIE ^ Int. Soc. Opt. Eng. 2367 ( 1995 ) 80. [ 7 ] B. Jaduszliwer and C.M. Klimcak, Sensor for detection of nitrogen dioxide and nitrogen tetroxide, U.S. Patent 5,567,622, 1996. [ 8 ] C.M. Klimcak, G.L. Loper and B. Jaduszliwer, Diode laser interrogated ¢ber optic reversible hydazine-fuel sensor system and method, U.S. Patent 5,610,393, 1997. [ 9 ] Y.M. Liu, C. Ganesh, J.P.H. Steele, J.E. Jones, J. Composite Mat. 31 ( 1997 ) 87. [ 10 ] V. Murphy, B.D. MacCraith, T. Butler, C. Mcdonagh, B. Lawless, Electron. Lett. 33 ( 1997 ) 618. [ 11 ] K. Ogawa, S. Tsuchiya, H. Kawakami, T. Tsutsui, Electron. Lett. 24 ( 1988 ) 42. [ 12 ] F. Kvasnik, A.D. McGrath, Proc. SPIE ^ Int. Soc. Opt. Eng. 1172 ( 1989 ) 75. [ 13 ] T. Takeo, H. Hattori, Proc. SPIE ^ Int. Soc. Opt. Eng. 1544 ( 1991 ) 282. [ 14 ] Y.-W. Shi, Y.-C. Wang, H.-T. Jiang, C.-S. Yao, Z.-C. Wu, Proc. SPIE ^ Int. Soc. Opt. Eng. 1572 ( 1991 ) 308. [ 15 ] A. Kharaz, B.E. Jones, Sensors Actuators A 47 ( 1995 ) 491. [ 16 ] P.A. Wallace, Y. Yang, M. Campbell, A.S. HolmesSmith, Proc. SPIE ^ Int. Soc. Opt. Eng. 2838 ( 1996 ) 153. [ 17 ] C.A. Browne, D.H. Tarrant, M.S. Olteanu, J.W. Mullens, E.L. Chronister, Anal. Chem. 68 ( 1996 ) 2289. [ 18 ] D.C. Bownass, J.S. Barton, J.D.C. Jones, Opt. Lett. 22 ( 1997 ) 346. [ 19 ] A.D. Kersey, Proc. SPIE ^ Int. Soc. Opt. Eng. 1797 ( 1992 ) 161. [ 20 ] J.C. Walker, R. Holmes and G.R. Jones, in K.T.V. Grattan and A.T. Augousti ( Editors ), Sensors VI. Technology, Systems and Applications, Institute of Physics, Bristol, 1993, p. 251. [ 21 ] J.S. Gergely and F.M. Evens, Distributed ¢ber-optic sensor with substance selective permeable coating, US Patent 5,138,153, 1992. [ 22 ] K. Nakamura, N. Uchino, Y. Matsuda and T. Yoshino, IEICE Trans. Commun. E80-B ( 1997 ) 528. [ 23 ] W.C. Michie, B. Culshaw, M. Konstantaki, I. McKenzie, S. Kelly, N.B. Graham, C. Moran, J. Lightwave Technol. 13 ( 1995 ) 1415. [ 24 ] E.A. Mendoza, J. Sorenson, A. Iossi, Z. Sun, D. Robinson, R.A. Lieberman, Proc. SPIE ^ Int. Soc. Opt. Eng. 2836 ( 1996 ) 242^249. [ 25 ] R.A. Potyrailo and G.M. Hieftje, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Atlanta, GA, 16^21 March 1997, Paper 921. [ 26 ] R.A. Potyrailo and G.M. Hieftje, M., Anal. Chem. ( 1998 ) in press. [ 27 ] J. Buërck, E. Sensfelder, H.-J. Ache, Proc. SPIE ^ Int. Soc. Opt. Eng. 2836 ( 1996 ) 250.
TRAC 2467 30-10-98
604
trends in analytical chemistry, vol. 17, no. 10, 1998
[ 28 ] L.L. Blyler, Jr., L.G. Cohen, R.A. Lieberman and J.B. MacChesney, Optical ¢ber sensors for chemical detection, US Patent 4,834,496, 1989. [ 29 ] L.L. Blyler Jr., R.A. Lieberman, L.G. Cohen, J.A. Ferrara, J.B. MacChesney, Polymer Eng. Sci. 29 ( 1989 ) 1215. [ 30 ] R.A. Lieberman, Sensors Actuators B 11 ( 1993 ) 43. [ 31 ] J.B. Puschett and K. Matyjaszewski, Surface modi¢cation of plastic optical ¢bers, U.S. Patent 5,077,078, 1991. [ 32 ] B.S. Rao, J.B. Puschett, K. Matyjaszewski, J. Appl. Polymer Sci. 43 ( 1991 ) 925. [ 33 ] R.A. Potyrailo and G.M. Hieftje, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, New Orleans, LA, 6^9 March 1995, Paper 1022. [ 34 ] R.A. Potyrailo and G.M. Hieftje, Anal. Chem. ( 1998 ) submitted. [ 35 ] R.A. Potyrailo, S.J. Ray, D.L. Burden and G.M. Hieftje, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Chicago, IL, 3^8 March 1996, Paper 1080. [ 36 ] E.R. Carraway, J.N. Demas, B.A. DeGraff, J.R. Bacon, Anal. Chem. 63 ( 1991 ) 337. [ 37 ] I. Klimant, O.S. Wolfbeis, Anal. Chem. 67 ( 1995 ) 3160. [ 38 ] R.A. Potyrailo and G.M. Hieftje, FACSS XXII Annual Meeting, Cincinnati, OH, 15^20 October 1995, Paper 335. [ 39 ] T. Werner, I. Klimant, O.S. Wolfbeis, Analyst 120 ( 1995 ) 1627. [ 40 ] W. Noll, Chemistry and Technology of Silicones, Academic Press, New York, 1968. [ 41 ] R. Feced, M. Farhadiroushan, H. Rodriguez, A.J. Rogers, Proc. SPIE ^ Int. Soc. Opt. Eng. 2838 ( 1996 ) 105. [ 42 ] A.J. Rogers, V.A. Handerek, S.E. Kanellopoulos, J. Zhang, Proc. SPIE ^ Int. Soc. Opt. Eng. 2507 ( 1995 ) 162. [ 43 ] J.K.A. Everard, J. Opt. Sensors 2 ( 1987 ) 363. [ 44 ] R.A. Potyrailo and G.M. Hieftje, FACSS XXIV Annual Meeting, Providence, RI, 26^30 October 1997, Paper 371. [ 45 ] J.N. Demas, Excited State Lifetime Measurements, Academic Press, New York, 1983. [ 46 ] S.D. Crossley, Proc. SPIE ^ Int. Soc. Opt. Eng. 2294 ( 1994 ) 14. [ 47 ] V. Ruddy, J. Maguire, Sensors Actuators A 42 ( 1994 ) 525. Radislav A. Potyrailo received his Ph.D. in Analytical Chemistry at Indiana University in 1998. Earlier, he received a degree in Engineering in the ¢eld of Optoelectronics from Kiev Polytechnic Institute. In 1992^1993 he served as a visiting scientist at the University of Toronto. His doctoral work at IU focused on the development of innovative instrumentation, materials,
and signal-processing techniques for chemical and biochemical diagnostics. These studies encompass the areas of evanescent-wave chemical and biochemical sensors, signal generation and processing for spatially resolved analyte mapping, and radioluminescent light sources. His research interests cut across the boundaries of analytical chemistry, environmental science, biochemistry, optoelectronics, image processing, and materials science. He has co-authored over 20 technical papers and over 30 presentations, ¢ve patents, and two books. He was recently awarded the McCormick Science Grant from IU and the Tomas Hirschfeld Award from the Federation of Analytical Chemistry and Spectroscopy Societies. Professor Gary M. Hieftje is Distinguished Professor and Chairman of the Department of Chemistry at Indiana University in Bloomington, Indiana. He received the A.B. degree from Hope College, Holland, MI in 1964 and a Ph.D. in 1969 from the University of Illinois. His research interests include the investigation of basic mechanisms in atomic emission, absorption, and £uorescence spectrometric analysis, and the development of atomic methods of analysis. He is interested also in ¢ber-optic sensors, on-line computer control of chemical instrumentation and experiments, the use of timeresolved luminescence processes for analysis, the application of information theory to analytical chemistry, near-infrared re£ectance analysis, and the use of stochastic processes to extract basic and kinetic chemical information. He currently serves on the editorial boards of Analytica Chimica Acta, Journal of Analytical Atomic Spectroscopy, Laboratory Microcomputer, Spectrochimica Acta, Part B, Advances in Inorganic Mass Spectrometry, the Analytical Chemistry Bench Top Series from Springer Verlag, Talanta, and Spectroscopy and Spectral Analysis. In 1991, he served as President of the Society for Applied Spectroscopy, and received the gold medal of the Quality Control Academy of the Upjohn Company. In 1992, he received the Eastern Analytical Symposium Award for Outstanding Achievements in the Fields of Analytical Chemistry and was awarded a second Lester Strock Medal. In 1993, he was elected as an Honorary Member of the Golden Key National Honor Society and was the recipient of the 1993 Distinguished Faculty Award from the College of Arts and Sciences alumni of Indiana University. In 1996, he received the Humboldt Research Award for Senior U.S. Scientists and the Meggers Award for the year 1995 from the Society for Applied Spectroscopy. Dr. Hieftje is the author of more than 380 scienti¢c publications, has authored or edited 10 books, and holds 11 patents.
Commercial reprints We can supply companies with reprints of any article in TrAC that directly or indirectly support their products at very reasonable rates ( minimum order 100 ). Please apply for details to: TrAC Editorial Of¢ce, P.O. Box 330, 1000 AH Amsterdam, The Netherlands
TRAC 2467 30-10-98
trends in analytical chemistry, vol. 17, no. 10, 1998
[ 36 ] C.S. Creaser and B.L. Williamson, J. Chem. Soc. Chem. Commun. ( 1994 ) 1677. [ 37 ] C.S. Creaser, B.L. Williamson, J. Chem. Soc. Perkin Trans. 2 ( 1996 ) 427. [ 38 ] A. Colorado, J. Brodbelt, Anal. Chem. 66 ( 1994 ) 2330.
[ 39 ] R.E. March and R.J. Hughes, Quadrupole Storage Mass Spectrometry, Wiley Interscience, New York, 1989. [ 40 ] J.L. Stephenson, S.A. McLuckey, Anal. Chem. 68 ( 1996 ) 4026.
Spatially resolved analyte mapping with time-of-£ight optical sensors Radislav A. Potyrailo**, Gary M. Hieftje*
Department of Chemistry, Indiana University, Bloomington, IN 47405, USA Simultaneous real-time acquisition of analytical data from several locations is attractive in a variety of applications. This brief review traces the evolution of approaches to such measurements. Greatest emphasis is placed on optical time-of-£ight chemical detection when signals are multiplexed from several point sensors or when the measurements are taken along the length of a single continuous extended-length `distributed' sensing element. The use of sensors featuring extended-length continuous chemically sensitive optical ¢bers offers detection arrangements for which there is no counterpart in conventional chemical sensor technologies. z1998 Elsevier Science B.V. All rights reserved. Keywords: Optical time-of-£ight; Chemical sensor; Optical ¢ber; Multiplexing; Distributed; Signal-to-noise; Discrete location; Continuous sensing element
1. Introduction Spatially resolved mapping of chemical constituents is an important need in a variety of practical applications. For example, spatially resolved analyte monitoring can simultaneously indicate and locate when an accepted level of exposure to toxic or explosive spe*Corresponding author. **Present address : Characterization and Environmental Technology Laboratory, Corporate Research and Development, General Electric Company, P.O. Box 8, Building K-1, Schenectady, NY 12301, USA.
cies has been exceeded, can track a source of contamination in an industrial or technological process, can follow the formation and movement of environmental pollutants, and can perform early nondestructive detection and location of corrosion ( such as in advanced structural components of space shuttles, aircraft, and civil engineering structures ). For these and other applications, information about the spatial distribution of species of interest can be obtained when signals are multiplexed from different discrete point sensors or when the measurements are taken along the entire length of a single continuous extended-length sensing element. This brief review traces the evolution of approaches to such measurements. It is demonstrated that the use of ¢ber optics provides a unique solution to this monitoring problem. The combination of a `distributed' sensor ( featuring an extended-length continuous chemically sensitive optical ¢ber ) and optical timeof-£ight (OTOF ) techniques offers detection schemes for which there is no counterpart in conventional chemical sensor technologies.
2. Optical time-of-£ight approaches for spatially resolved analyte mapping For spatially resolved chemical measurements, a variety of techniques can be adapted from mature approaches for diagnosing ¢ber-optic communication networks and physical sensors [ 1 ]. In these areas, OTOF techniques are the most widely used because they offer strong advantages over other methods. These advantages include improved quantitation, operation with multimode ¢bers, simple signal-gener-
0165-9936/98/$ ^ see front matter PII: S 0 1 6 5 - 9 9 3 6 ( 9 8 ) 0 0 0 6 4 - 8
ß 1998 Elsevier Science B.V. All rights reserved.
TRAC 2460 30-10-98
594
trends in analytical chemistry, vol. 17, no. 10, 1998
Fig. 1. Con¢gurations of optical time-of-£ight chemical sensors for spatially resolved measurements. ( A ) Star con¢guration for multiplexing point sensors in transmission mode, ( B ) star con¢guration for multiplexing point sensors in backpropagation mode, ( C ) serial con¢guration for multiplexing point sensors in backpropagation mode, ( D ) distributed sensor con¢guration. FOC, ¢beroptic coupler; DL, ¢ber-optic delay line; PD, photodetector; MS, monitoring site.
ation and processing methods, and relatively inexpensive opto-electronic components. These advantages of OTOF methods are transferable to the ¢eld of chemical sensing. With OTOF-based sensing, it is possible to locate zones along a ¢ber where analyte-induced changes in absorption or £uorescence take place, to determine the magnitude of these variations, and to relate the magnitude of the variations to the concentration of an analyte. The different OTOF approaches for spatially resolved analyte mapping shown schematically in Fig. 1 have a common concept. In OTOF methods, a pulse of light is periodically launched into a ¢ber or ¢bers and the analytical information is derived from the amplitude and temporal characteristics of the light which interacts with the sensing element and then
reaches a detector. The amplitude of the pulsed signal is affected by the analyte concentration at each of the point sensors ( Fig. 1A^C ) or along the continuous sensing ¢ber ( Fig. 1D ). Information about the location where sensing occurs is obtained from the measured time delay between the launched and detected light pulses. In multiplexing strategies, used with discrete point sensors, this time delay is controlled by connecting each individual sensor to a detector through a different length of a ¢ber. The signal from each of the multiplexed sensors can be obtained in either a transmission ( Fig. 1A ) or backpropagation ( Fig. 1B,C ) mode. When the OTOF technique is based on the detection of backpropagated radiation, it is known as optical time-domain re£ectometry (OTDR ) [ 2 ]. Very similar mathematical expressions describe the signals for backscatter and backpropagated £uorescence in multiplexed point and distributed sensors [ 3 ]. Thus, although backpropagated £uorescence is not strictly a backscatter signal, the term OTDR can also be applied to £uorescence detection. Point sensors are commonly multiplexed by means of star ( Fig. 1A,B ) or serial ( Fig. 1C ) arrangements; features of these multiplexed point sensors are summarized in Tables 1 and 2. The number of point sensors used in a star OTDR-based multiplexing topology can be limited by the power budget of the system; at most, 25^30 sensors can be supported by a single laser source with acceptable detection sensitivities [ 19 ]. This limitation arises from losses at each multiplexing point [ 20 ]. These losses can be reduced signi¢cantly in a serially multiplexed topology if a single extendedlength ¢ber is reclad over short regions with a chemically sensitive reagent [ 12,15^18 ]. Unfortunately, replacement of the original cladding with a chemically sensitive ¢lm in such a `quasi-distributed' sensor creates a number of problems, including reclad lengths of only a few centimeters long, non-uniformity of the reclad sections, scattering losses, and loss of mechanical £exibility. In addition, multiplexing of short sections is still limited to spatially discrete measurements ( i.e. the continuity parameter D in Table 2 is much less than unity ). All these problems can be eliminated if a single continuous extended-length sensing ¢ber is used ( Fig. 1D ). In addition, a distributed sensor can provide a more complete and cost-effective spatial pro¢le of chemical constituents. Several such distributed sensors have been reported ( see Table 3 ).
TRAC 2467 30-10-98
595
trends in analytical chemistry, vol. 17, no. 10, 1998
Table 1 Features of reported OTOF star multiplexed point sensors Analyte species or measured parameter
Detection method
Design of a sensing element
Number of point sensors ( ¢ber length, m )
Ref.
Acetylene
Absorption OTOF in transmission mode ( 1.528 Wm ) Two-photon £uorescence OTDR ( exc. at 0.59 Wm ) Absorption OTDR ( 0.68 or 0.78 Wm )
A 7 cm long absorption cell coupled to a single mode ¢ber
4 ( 1^25 )
[4]
Reagent-free ¢ber tip in analyte solution A ¢ber section reclad with a sol-gel ¢lm doped with phosphomolybdic acid for evanescent-wave sensing or the reagent is deposited onto the distal end of the ¢ber Reagent-free ¢ber tip immersed into monitored sample An etched cavity at the distal end of a graded index multimode ¢ber packed with sol-gel-derived silica doped with trimethinecyanine
3^10 ( 40-140 )
[5]
NRb
[ 6^8 ]
8 (NR )
[9]
3 ( 100^300 )
[ 10 ]
K-NPOa Hydrazine, nitrogen dioxide, nitrogen tetroxide
Degree-of-cure of epoxy resin pH
a b
Fresnel re£ection OTDR ( 0.904 Wm ) Absorption OTDR ( 0.85 Wm )
K-NPO = 2-1 (-naphthyl )-5-phenyloxazole. NR = not reported.
3. Transduction mechanisms and fabrication strategies for distributed chemical sensors Most distributed chemical sensors reported to date have been based on evanescent-wave detection. In this sort of device, the detected signal is dictated by analyte-induced changes in the refractive index, absorption, or £uorescence of an analyte-permeable cladding. In order to be detected, of course, the analyte must lie within the penetration depth of the evanescent ¢eld produced by a guided mode( s ) in the ¢ber ( cf. Table 3 ), requiring the ¢ber cladding to be permeable to the analyte. The use of analyte-induced variations in refractive index of the cladding [ 21,22,27 ] is limited in scope by the small number of species that can be determined using this approach. As a result, indirect [ 23 ], including reagent-based [ 3,24,25 ], sensing approaches have received more attention. The ¢rst extended-length chemically sensitive ¢bers useful for spatially resolved analyte mapping were produced by Blyler et al. [ 28,29 ] in the late 1980s. They doped special formulations of liquid silicones with analyte-sensitive reagents, applied these solutions onto the silica core of a multimode optical ¢ber to form the ¢ber cladding, and cured the cladding in-line during the ¢ber-drawing operation. Because the analyte-sensitive reagent was distributed along the entire length of the ¢ber, analyte sensing could
occur over extended regions. Unfortunately, as admitted by the inventors [ 28^30 ], the technology suffered from several serious drawbacks. First, the photocuring chemistry involved the dissolved reagent and resulted in incomplete reversibility of the sensor. Second, the technology was complex, very expensive, and limited to those who have a ¢ber-optic drawing facility. Also, an organic reagent doped in the cladding was exposed to strong UV light or high temperature required for the crosslinking of silicone during the ¢ber manufacture [ 28 ]; these conditions were likely to cause partial degradation of the reagent. To overcome these limitations, an attempt was made also to chemically immobilize reagents onto the surface of a commercially available plastic optical ¢ber [ 31,32 ]. Although the technique was useful for short ¢ber sections, it was not attractive for distributed sensing over extended regions because of scattering losses in the modi¢ed ¢bers. In addition, the indicator had to be synthesized to contain the appropriate functional groups for immobilization. Recently, a more attractive strategy for the fabrication of extended-length sensors was introduced [ 33 ]. The technique involves immobilization of a reagent directly into the original ( crosslinked ) silicone cladding of a conventional plastic-clad silica (PCS ) ¢ber. For indicator immobilization, the ¢ber is soaked in a solution containing the required indicator dissolved in a non-polar solvent. After removal from the indicator
TRAC 2467 30-10-98
596
trends in analytical chemistry, vol. 17, no. 10, 1998
Table 2 Features of reported OTOF serially multiplexed point sensors Analyte species or measured parameter
Detection method
Design of a sensing element
Continuity parameter Da
Ref.
RHb
Rayleigh OTDR ( 0.85 Wm ) Absorption OTDR
A PCS ¢ber segment reclad with microporous silica for evanescent-wave sensing A PCS ¢ber section reclad with a reagentdoped cellulose acetate ¢lm A U-shaped unclad segment of silica ¢ber with a V0.5 mm bend radius for evanescentwave sensing Supported catalytic point sensor converted analyte concentration change into temperature change A multimode ¢ber section reclad with a cobalt chloride-doped gelatin ¢lm for evanescentwave sensing A PCS ¢ber section reclad with a £uoresceindoped sol-gel ¢lm for evanescent-wave sensing A PCS ¢ber section reclad with a £uorophoredoped sol-gel ¢lm for evanescent-wave sensing
4 cmU3 / 130 m
[ 11 ]
25 cmU3 / 25 m
[ 12 ]
1 mmU2 / 500 m
[ 13 ]
NR
[ 14 ]
5 cmU4 / 100 m
[ 15 ]
NRU9 / 100 m
[ 16 ]
4 cmU4 / 9 m; 4 cmU4 / 7 m
[ 17 ]
2 mmU4 / 8 km
[ 18 ]
Alkaline and acid vapors RH ( via RIc )
Rayleigh OTDR ( 0.85 Wm )
Methane ( via temperature )
Rayleigh OTDR
RH
Absorption OTDR ( 0.67 Wm, reference at 0.85 Wm ) Fluorescence OTDR ( exc. at 0.44 Wm ) Fluorescence OTDR ( exc. at 0.424, 0.5, or 0.56 Wm ) Rayleigh OTDR ( 1.31 Wm )
pH pH, quinone RH ( via RI )
A polished single-mode half-coupler with a polyethylene oxide overlay for evanescentwave sensing
a
Continuity parameter D is given by D = lN /L, where l is the length of a sensing region, N is the number of sensors, and L is the total length of the ¢ber which includes the sensing regions and the ¢ber link. b RH = relative humidity. c RI = refractive index.
solution, the ¢ber is washed with pure solvent and dried [ 33,34 ]. In contrast to earlier reported methods for fabricating extended-length indicator-based sensing elements [ 28,29,31,32 ] the new strategy does not suffer from photo- or temperature-induced degradation of the indicator, offers reversible sensor response, simpli¢es the fabrication process, lowers the construction cost of the sensing ¢ber, does not induce scattering losses upon indicator immobilization, and permits immobilization of several reagents in adjacent sections of a single optical ¢ber for multianalyte determinations. The silicone cladding of a conventional PCS optical ¢ber, a copolymer of vinyl-terminated poly( dimethylsiloxane ) and poly( dimethyl-methylhydrosiloxane ), is an attractive immobilization matrix for a wide variety of reagents and opens up new avenues of sensor design [ 34 ]. The silicone cladding material offers a number of advantages over other common silicones used as host matrices for immobilization of indicators. These advantages accrue from both the composition and crosslink density of the silicone cladding. First, the cladding makes dynamic £uorescence quenching of an immobilized £uorophore more
ef¢cient than is provided by common room-temperature-vulcanized (RTV ) silicones. For example, the Ru( phen )3 cation exhibits the highest oxygenquenching constant reported when immobilized in a silicone elastomer [ 35 ] ( Fig. 2 ). Also, the high network density in the silicone cladding material leads to reduced aggregation of immobilized indicator molecules. In a chemical sensor that employs dynamically quenched £uorescence, reduced reagent aggregation results in more effective quenching. Second, the high crosslinking density in the cured elastomer affords the cladding a high degree of resistance to aggressive media ( such as alkaline solutions ) and therefore protects immobilized reagents from leaching or decomposition [ 38 ]. The resistance of the reagent-doped silicone cladding to highly alkaline solutions is important in sensors intended for high ammonia concentrations ( cf. Fig. 3 ). Previous attempts to develop a sensor useful for such determinations, without sample dilution, suffered from leaching of the immobilized indicator [ 39 ]. Longterm stability tests of a PCS ¢ber modi¢ed with a pH indicator in aqueous solutions of 14 M NH4 OH and 10 M NaOH revealed no observable washout
TRAC 2467 30-10-98
597
trends in analytical chemistry, vol. 17, no. 10, 1998
Table 3 Features of reported OTOF distributed sensors Analyte species or measured parameter
Detection method
Design of a sensing element
Length of the sensing region ( m )
Ref.
Any substance via RI
Rayleigh OTDR
NR
[ 21 ]
Oil ( via RI )
Rayleigh OTDR ( 1.31 Wm )
2000
[ 22 ]
Water
Rayleigh OTDR
60
[ 23 ]
pH, moisture
Absorption OTDR ( 0.85 Wm, reference at 1.30 Wm ) Fluorescence OTDR ( exc. 0.532 Wm )
A multimode ¢ber with analyte permeable cladding for evanescent-wave sensing Specially drawn single-mode eccentric core ¢ber with silicone coating on top of the glass ¢ber cladding for evanescent-wave sensing A multimode optical ¢ber in intimate contact with a hydrogelcoated rod by a helically wound thread. Hydrogel swelling induces the microbending loss of light within the ¢ber through a change in strain caused by the con¢ning thread Specially drawn multimode ¢ber with dye-doped polymer cladding for evanescent-wave sensing Conventional PCS ¢ber with chemically modi¢ed original silicone cladding for evanescent-wave sensing Conventional PCS ¢ber with chemically modi¢ed original silicone cladding for evanescent-wave sensing Conventional PCS ¢ber with chemically modi¢ed original silicone cladding for evanescent-wave sensing
20
[ 24 ]
48
[ 3,25 ]
40
[ 26 ]
40
[ 26 ]
Oxygen, atmospheric pressure Ammonia Ammonia
Absorption-modulated £uorescence OTDR ( exc. 0.532 Wm ) Bidirectional OTDR ( exc. 0.532 Wm )
of the immobilized dye over a period of several months [ 38 ]. Dye-doped silicone-cladding sensors for continuous humidity monitoring, possibly useful for early detection of moisture-induced corrosion in structural components in civil, aerospace, and electrical engineering, have been examined [ 35 ]. While liquid water has a very low solubility in silicone resins, such resins are highly permeable to water vapor [ 40 ]. The dynamic response curve for a Nile-redmodi¢ed PCS ¢ber is compared with that of a commercial humidity monitor in Fig. 4. The sorption and desorption of water vapor in the silicone ¢ber cladding was completely reversible, as veri¢ed by monitoring the hypochromic effect at 522 nm. More than 30 indicators of different sorts ( polycyclic aromatic hydrocarbons, rhodamines, coumarins, cyanides, sulfonaphthaleins, azo-dyes, porphyrins, and others ) have been successfully immobilized into the silicone cladding of PCS ¢bers, by utilizing a variety of non-polar solvents [ 34 ]. The molecular weight of the immobilized indicators spanned almost an order of magnitude from 140 ( nitrophenol ) to 1200 ( a Ru
complex ). It is possible that this range could be broader; however, it encompasses most £uorescence and absorption indicators.
Fig. 2. Stern^Volmer calibration plots for quenchingbased oxygen sensors based on the Ru( phen )3 complex immobilized in different matrices: ( A ) silicone cladding of a PCS optical ¢ber, ( B ) GE RTV 118 silicone ¢lm ( experimental points taken from Ref. [ 36 ]), ( C ) single-component RTV acetic acid releasing silicone ¢lm ( experimental points taken from Ref. [ 37 ]).
TRAC 2467 30-10-98
598
trends in analytical chemistry, vol. 17, no. 10, 1998
These requirements include high signal levels, a fairly uniform detection limit over the length of the sensing ¢ber, and high spatial resolution. Recently, however, several signal generation and processing methods have been devised to address these requirements. 4.1. Raising signal levels in OTOF distributed sensors
Fig. 3. Calibration curve for a solid-state sensor with the phenol-red-modi¢ed optical ¢ber, a green light-emitting diode, and a silicon photodiode. The 95% con¢dence intervals are given for three replicate measurements during a month of experiments. The total exposure time to 0.006^0.6 M ammonium ion solutions in these measurements was 4 h. Concentration ranges of ( I ) environmental and ( II ) industrial interest.
4. Recent progress in OTOF methods in spatially resolved analyte mapping The analytical signal in an OTOF measurement system with a distributed sensing region was ¢rst formulated for Rayleigh backscatter and then for Raman backscatter. If the relationship is generalized also for £uorescence detection, the backpropagated impulse response from a distance l along the ¢ber excited by a N-pulse is given by [ 3 ] Z l 1 P l rPo S lexpf3 Kf z Kb zdzg 0
Scattering-based OTDR suffers from extremely low signal levels. The Rayleigh backscattered signal is typically 104 ^105 times weaker than the forward traveling pulse [ 2 ]; Raman backscatter is about 103 times weaker still [ 41 ]. As a result, recording the backscattered signal is often a problem since it requires sophisticated detection schemes, high-power lasers, and time-consuming signal averaging techniques. The S /N-enhancement methods devised for distributed physical sensors ( e.g. based on pseudonoise, polarimetric techniques, and nonlinear optical effects [ 42,43 ]) are limited to applications involving singlemode optical ¢bers and, thus, are not useful for most distributed chemical sensors. A new approach to signal generation and processing was recently introduced to enhance signal levels in distributed chemical sensors [ 26 ]. In this method, an analyte-sensitive absorbing dye and an analyte-insensitive £uorophore were immobilized together into the ¢ber cladding. These reagents were chosen so their absorption bands overlap and the excitation wavelength was selected to be within this overlap region. The £uorescence intensity of an analyte-insensitive £uorophore was then monitored as a function of the evanescent-wave absorption of an analyte-sensitive
where Po is the excitation power in the input pulse coupled into the ¢ber, r is the ratio of the transmitted to the re£ected optical power in the ¢ber-optic coupler ( cf. Fig. 1D ), S is a constant that depends on the local numerical aperture (NA ) of the ¢ber, the £uorescence quantum yield of the immobilized £uorophore ( in a £uorescence-based sensor ), or on the Rayleigh and Raman scattering parameters of the ¢ber core and cladding ( in a scattering-based sensor ), and Kf ( z ) and Kb ( z ) are the attenuation coef¢cients of the forward and backward traveling pulses, respectively. In £uorescence-based sensors, both Kf and Kb have two components, one each for the excitation and emission wavelengths. Traditional signal-processing techniques used with OTOF-based chemical sensors do not meet the requirements for spatially resolved chemical sensing.
Fig. 4. Dynamic response to change in relative humidity: ( b ) PCS optical ¢ber modi¢ed with Nile red and ( a ) a commercial humidity monitor.
TRAC 2467 30-10-98
599
trends in analytical chemistry, vol. 17, no. 10, 1998
Fig. 5. Spatially resolved distributed measurements with OTOF absorption-modulated £uorescence. Regions 2, 3, 4, and 9 of a continuous 40 m long distributed ammonia sensor were exposed to 213 ppm of ammonia while the rest of the ¢ber was in air. ( A ) Collected intensity waveform I is normalized by the intensity waveform I o of the whole ¢ber in absence of analyte. ( B ) Analyte-induced increase in the local absorption coef¢cient in decibels as a function of ¢ber length. The waveform in B is smoothed by a running-average function.
indicator. In essence, the analyte-sensitive absorber `steals' excitation light from the £uorophore, so the emission of the £uorophore is inversely dependent on the analyte concentration. The low £uorophore concentration does not contribute appreciably to the ¢ber attenuation at the excitation and emission wavelengths and therefore does not lead to a loss in dynamic range of the sensor. The £uorophore can be chosen with a £uorescence lifetime shorter than the pulse width of the source in order not to degrade spatial resolution ( see Section 4.4 ). This concept was experimentally veri¢ed with a distributed sensor constructed from a 40 m long continuous chemically sensitive optical ¢ber. This sensing element was produced by immobilization of an ammonia-sensitive absorbing reagent ( phenol red )
and an analyte-insensitive £uorophore ( rhodamine 640 ) into the silicone cladding of a PCS ¢ber. Typical results of these experiments are plotted in Fig. 5A as the ratio of the measured £uorescence intensities in the presence and in the absence of the analyte. By plotting the analyte-induced increase in the local absorption coef¢cient ( in decibels ) as a function of time ( distance along the ¢ber ) it is possible to spatially resolve analyte concentrations ( Fig. 5B ). When this data representation is employed, ¢ber regions that cause an increased attenuation ( because of higher ammonia concentration ) exhibit noticeable dips below the constant background. The vertical scale on a plot such as that in Fig. 5B could be easily converted by calibration from analyte-induced ¢ber attenuation to analyte concentration.
TRAC 2467 30-10-98
600
trends in analytical chemistry, vol. 17, no. 10, 1998
The resulting signal levels in absorption-modulated £uorescence OTOF detection were well above those found in OTDR methods that rely upon backscattering for spatially resolved detection [ 2,24 ]. The degree of enhancement was between 10- and 120-fold and was dependent on the concentration of immobilized £uorophore and on its quantum yield. With this technique, the range of species detectable with distributed £uorescence sensors might be expanded to include analytes previously detectable only with attenuationbased methods, and therefore dif¢cult to monitor with an OTOF distributed sensor. By using £uorescence rather than backscatter detection, it is possible also to increase the speed of data collection by reducing the number of collected waveforms that must be averaged, and to use more compact and less powerful light sources. 4.2. Improvements in S /N
Methods have recently been introduced that help overcome the loss in S /N that usually accompanies the use of a long ¢ber [ 26 ]. The loss in S /N with ¢ber length [ 2 ], caused by the ¢nite attenuation coef¢cient of the ¢ber material, usually limits the useful length of an OTDR sensor. To overcome this limitation, bidirectional detection was used. The light was launched sequentially into the two ends of a 40 m long continuous chemically sensitive optical ¢ber and the backpropagated portion was collected from the same end into which the respective pulse was launched [ 26 ]. Because the two pulses provided complementary information, only the ¢rst half of each of the collected waveforms was used for analyte quantitation. As shown in Fig. 6A, the S /N degraded rapidly with distance from the respective launch end of the ¢ber. The relative standard deviation (RSD ) dropped from its initial level of 0.5% to between 12 and 15% at the distant ends of the ¢ber. When bidirectional detection was used, the RSD at the far end of the chemically sensitive ¢ber was reduced by 3^4.5-fold. The result of combining the ¢rst halves of the waveforms collected from the different ¢ber ends is illustrated in Fig. 6B. The middle section of the ¢ber had an unchanged 3% RSD whereas the two ends yielded the highest S /N. The bidirectional pulse-launching method to improve S /N for measurements over the length of a sensing ¢ber might be especially useful for sensors based on analyte-induced changes in ¢ber attenuation; in this type of sensor the S /N decays rapidly with distance from the launch end of the ¢ber.
4.3. OTOF measurements with dynamically quenched £uorophores
A limitation of absorption-based OTOF distributed sensors is the accumulative analyte-induced light attenuation that they suffer. A local increase in evanescent-wave absorbance by the immobilized reagent leads to a drop in light intensity available for propagation farther down the ¢ber. The dynamic range can therefore be limited by high analyte concentrations along the ¢ber. Sensors based on immobilized dynamically quenched £uorophores are free from this limitation. Such a sensor does not exhibit a change in attenuation coef¢cient upon exposure to a quencher. Rather, the £uorescence lifetime is a function of the quencher concentration, as described by the Stern^Volmer equation. The variable £uorescence lifetime then changes the emission intensity of the immobilized £uorophore detected with an OTOF system [ 3 ]. The ¢rst distributed OTOF sensor based on dynamically quenched £uorescence was developed by immobilizing a far-red emitting £uorophore ( tetraphenylporphyrin ) onto a 48 m long PCS ¢ber and was used for spatially resolved oxygen mapping [ 25 ] ( Fig. 7A ). The £uorophore was selected to have a short £uorescence lifetime ( 10 ns ) to provide adequate spatial resolution. Numerical simulations were performed using known parameters of the immobilized indicator and of the detection system to predict the observed intensity waveform in spatially resolved measurements ( Fig. 7B ). 4.4. Spatial resolution of distributed OTOF sensors
Spatial resolution in a distributed sensor should be as great as possible to provide the maximum number of resolved local variations of analyte concentration along the ¢ber length and to make the distributed sensor as cost-effective as possible compared to multiplexed point sensors. Spatial resolution of an absorption-based sensor is limited by the pulse width of the primary light source. For example, a laser having a 1-ns pulse width can provide roughly 10-cm spatial resolution in an absorption-based sensor [ 24 ]. In £uorescence-based sensors, spatial resolution is limited by the pulse width of the source, the impulse response ( excited-state lifetime ) of the immobilized £uorophore, and the bandwidth of the £uorescence that is detected [ 44 ]. Thus, for the best spatial resolution, it is desirable to use a narrow-bandpass inter-
TRAC 2467 30-10-98
601
trends in analytical chemistry, vol. 17, no. 10, 1998
Fig. 6. Spatially resolved analyte monitoring with OTOF by means of a bidirectional distributed sensor. Regions 2, 3, 6, and 7 of a continuous 40 m long distributed ammonia sensor ( similar to that in Fig. 5 ) were exposed to 100 ppm of ammonia while the rest of the ¢ber was in air. The returned waveforms were collected sequentially from end 1 and end 2 of the ¢ber for three ammonia-air cycles. The waveforms collected from end 2 were reversed and plotted as if they had been measured from end 1. ( A ) Collected intensity waveforms I are normalized by the waveforms I o of the whole ¢ber in absence of analyte. ( B ) Combined ¢rst halves of each of the normalized waveforms. Every 10th data point in ( A ) and ( B ) is shown with an error bar, which represents one standard deviation from the mean.
ference ¢lter and a £uorophore with a short £uorescence lifetime. This approach can be useful with a sensor that utilizes static quenching [ 26 ]. Unfortunately, this strategy does not work with a sensor utilizing a dynamically quenched £uorophore; the sensitivity of such a sensor usually drops when a £uorophore of very short lifetime is used. However, spatial resolution in a sensor that utilizes a reagent with a long £uorescence lifetime can be improved to its theoretical limit by using appropriate signal processing [ 44 ]. This processing involves deconvolution of the returned waveform with the laser pulse width and with the analyte-dependent £uorescence lifetime of the immobilized £uorophore. The method retrieves information about the analyte-modulated £uorescence
lifetime of the immobilized £uorophore at any location along the ¢ber. Several practical limitations often make deconvolution dif¢cult, noise usually being the most serious one [ 45 ]. If deconvolution is not used, the spatial resolution vlmin of a detection system is the smallest resolved distance between two ¢ber locations given by [ 1] vlmin cdb = nf nb
2
where c is the velocity of light in vacuum, nf and nb are the refractive indices of the ¢ber core at the wavelengths of the forward ( excitation wavelength ) and backward ( analytical wavelength ) propagated pulses,
TRAC 2467 30-10-98
602
trends in analytical chemistry, vol. 17, no. 10, 1998
Fig. 7. Spatially resolved oxygen determinations with the distributed oxygen sensor. In experiments I^IV several regions of a continuous 48 m long distributed oxygen sensor were exposed to nitrogen or oxygen, while the rest of the ¢ber was in air: I, regions 2 and 4 in nitrogen; II, regions 2 and 4 in oxygen; III, region 11 in nitrogen; IV, regions 11 and 13 in oxygen. ( A ) Collected intensity waveforms I are normalized by the intensity waveforms I air of the whole ¢ber in air. ( B ) Comparison of experimental ( data points ) and calculated ( solid lines ) results for spatially resolved oxygen monitoring.
and db is the width of the backpropagated pulse. The number of sensing regions N resolved with such a distributed sensor of length L is given by [ 3 ] N nf nb L=cdb
3
Several OTDR systems designed speci¢cally for diagnostic applications in ¢ber-optic communications
( spectral region from 0.85 to 1.55 Wm ) have been adapted for spatially resolved analyte mapping [ 11,13,18,23,24 ]. Typical pulse widths of conventional OTDR systems used for chemical detection range from 1 to 100 ns [ 13,18,24 ], which corresponds to a spatial resolution from 0.1 to 10 m. Such systems require near-infrared reagents or refractive-index detection.
TRAC 2467 30-10-98
603
trends in analytical chemistry, vol. 17, no. 10, 1998
5. Conclusions and future prospects Chemical detection represents a strong potential market for distributed sensors, after those for temperature, pressure, strain, displacement, and intruder detection [ 46 ]. However, activity in this direction has been limited by dif¢culties in fabricating sensing ¢bers of extended length, by low signal levels in backscatter-detection schemes, and by poor spatial resolution in £uorescence-based sensors. As new fabrication technologies for extended-length ¢bers are invented, as methods for advanced signal processing are designed, and as techniques for improving the spatial resolution of £uorescence-based sensors are devised, a speedier implementation of this technology can be expected. The basic and improved OTOF methods can be combined to advantage with other techniques. One of the possibilities is to couple spatially resolved OTOF detection with determination of the modal power distribution in a multimode waveguide [ 47 ]. Schemes based not only on OTOF techniques but also on other strategies also could be developed on the basis of amplitude-modulation and phase-modulation approaches. In the future, it should be possible to use optical ¢bers with alternative cladding materials, to allow different reagents to be immobilized, so a wide range of species in gaseous and aqueous phase can be determined over extended, remote areas. Spinoffs of this technology might be useful also in other areas of chemical sensing. For example, extended-length sensing elements could be used not only for distributed sensing, but also in point sensing; sensing segments of a desired length could be cut from an already prepared and calibrated ¢ber spool and used without further calibration.
Acknowledgements This work was supported in part by the National Institutes of Health through Grant GM 53560. R.A.P. acknowledges support from a McCormick Science Grant provided by the College of Arts and Sciences, Indiana University.
References [ 1 ] A.D. Kersey, in E. Udd ( Editor ), Fiber Optic Sensors: An
Introduction for Engineers and Scientists, Wiley, New York, 1991, p. 325. [ 2 ] M. Tateda, T. Horiguchi, J. Lightwave Technol. 7 ( 1989 ) 1217. [ 3 ] R.A. Potyrailo, G.M. Hieftje, Anal. Chem. 70 ( 1998 ) 1453. [ 4 ] D.R. Hjelme, S. Neegaard, J.S. Rambech, N. Ryen, Proc. SPIE ^ Int. Soc. Opt. Eng. 2507 ( 1995 ) 113. [ 5 ] R.L. Steffen, F.E. Lytle, Anal. Chim. Acta 215 ( 1988 ) 203. [ 6 ] C. Klimcak, G. Radhakrishnan, B. Jaduszliwer, Proc. SPIE ^ Int. Soc. Opt. Eng. 2367 ( 1995 ) 80. [ 7 ] B. Jaduszliwer and C.M. Klimcak, Sensor for detection of nitrogen dioxide and nitrogen tetroxide, U.S. Patent 5,567,622, 1996. [ 8 ] C.M. Klimcak, G.L. Loper and B. Jaduszliwer, Diode laser interrogated ¢ber optic reversible hydazine-fuel sensor system and method, U.S. Patent 5,610,393, 1997. [ 9 ] Y.M. Liu, C. Ganesh, J.P.H. Steele, J.E. Jones, J. Composite Mat. 31 ( 1997 ) 87. [ 10 ] V. Murphy, B.D. MacCraith, T. Butler, C. Mcdonagh, B. Lawless, Electron. Lett. 33 ( 1997 ) 618. [ 11 ] K. Ogawa, S. Tsuchiya, H. Kawakami, T. Tsutsui, Electron. Lett. 24 ( 1988 ) 42. [ 12 ] F. Kvasnik, A.D. McGrath, Proc. SPIE ^ Int. Soc. Opt. Eng. 1172 ( 1989 ) 75. [ 13 ] T. Takeo, H. Hattori, Proc. SPIE ^ Int. Soc. Opt. Eng. 1544 ( 1991 ) 282. [ 14 ] Y.-W. Shi, Y.-C. Wang, H.-T. Jiang, C.-S. Yao, Z.-C. Wu, Proc. SPIE ^ Int. Soc. Opt. Eng. 1572 ( 1991 ) 308. [ 15 ] A. Kharaz, B.E. Jones, Sensors Actuators A 47 ( 1995 ) 491. [ 16 ] P.A. Wallace, Y. Yang, M. Campbell, A.S. HolmesSmith, Proc. SPIE ^ Int. Soc. Opt. Eng. 2838 ( 1996 ) 153. [ 17 ] C.A. Browne, D.H. Tarrant, M.S. Olteanu, J.W. Mullens, E.L. Chronister, Anal. Chem. 68 ( 1996 ) 2289. [ 18 ] D.C. Bownass, J.S. Barton, J.D.C. Jones, Opt. Lett. 22 ( 1997 ) 346. [ 19 ] A.D. Kersey, Proc. SPIE ^ Int. Soc. Opt. Eng. 1797 ( 1992 ) 161. [ 20 ] J.C. Walker, R. Holmes and G.R. Jones, in K.T.V. Grattan and A.T. Augousti ( Editors ), Sensors VI. Technology, Systems and Applications, Institute of Physics, Bristol, 1993, p. 251. [ 21 ] J.S. Gergely and F.M. Evens, Distributed ¢ber-optic sensor with substance selective permeable coating, US Patent 5,138,153, 1992. [ 22 ] K. Nakamura, N. Uchino, Y. Matsuda and T. Yoshino, IEICE Trans. Commun. E80-B ( 1997 ) 528. [ 23 ] W.C. Michie, B. Culshaw, M. Konstantaki, I. McKenzie, S. Kelly, N.B. Graham, C. Moran, J. Lightwave Technol. 13 ( 1995 ) 1415. [ 24 ] E.A. Mendoza, J. Sorenson, A. Iossi, Z. Sun, D. Robinson, R.A. Lieberman, Proc. SPIE ^ Int. Soc. Opt. Eng. 2836 ( 1996 ) 242^249. [ 25 ] R.A. Potyrailo and G.M. Hieftje, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Atlanta, GA, 16^21 March 1997, Paper 921. [ 26 ] R.A. Potyrailo and G.M. Hieftje, M., Anal. Chem. ( 1998 ) in press. [ 27 ] J. Buërck, E. Sensfelder, H.-J. Ache, Proc. SPIE ^ Int. Soc. Opt. Eng. 2836 ( 1996 ) 250.
TRAC 2467 30-10-98
604
trends in analytical chemistry, vol. 17, no. 10, 1998
[ 28 ] L.L. Blyler, Jr., L.G. Cohen, R.A. Lieberman and J.B. MacChesney, Optical ¢ber sensors for chemical detection, US Patent 4,834,496, 1989. [ 29 ] L.L. Blyler Jr., R.A. Lieberman, L.G. Cohen, J.A. Ferrara, J.B. MacChesney, Polymer Eng. Sci. 29 ( 1989 ) 1215. [ 30 ] R.A. Lieberman, Sensors Actuators B 11 ( 1993 ) 43. [ 31 ] J.B. Puschett and K. Matyjaszewski, Surface modi¢cation of plastic optical ¢bers, U.S. Patent 5,077,078, 1991. [ 32 ] B.S. Rao, J.B. Puschett, K. Matyjaszewski, J. Appl. Polymer Sci. 43 ( 1991 ) 925. [ 33 ] R.A. Potyrailo and G.M. Hieftje, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, New Orleans, LA, 6^9 March 1995, Paper 1022. [ 34 ] R.A. Potyrailo and G.M. Hieftje, Anal. Chem. ( 1998 ) submitted. [ 35 ] R.A. Potyrailo, S.J. Ray, D.L. Burden and G.M. Hieftje, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Chicago, IL, 3^8 March 1996, Paper 1080. [ 36 ] E.R. Carraway, J.N. Demas, B.A. DeGraff, J.R. Bacon, Anal. Chem. 63 ( 1991 ) 337. [ 37 ] I. Klimant, O.S. Wolfbeis, Anal. Chem. 67 ( 1995 ) 3160. [ 38 ] R.A. Potyrailo and G.M. Hieftje, FACSS XXII Annual Meeting, Cincinnati, OH, 15^20 October 1995, Paper 335. [ 39 ] T. Werner, I. Klimant, O.S. Wolfbeis, Analyst 120 ( 1995 ) 1627. [ 40 ] W. Noll, Chemistry and Technology of Silicones, Academic Press, New York, 1968. [ 41 ] R. Feced, M. Farhadiroushan, H. Rodriguez, A.J. Rogers, Proc. SPIE ^ Int. Soc. Opt. Eng. 2838 ( 1996 ) 105. [ 42 ] A.J. Rogers, V.A. Handerek, S.E. Kanellopoulos, J. Zhang, Proc. SPIE ^ Int. Soc. Opt. Eng. 2507 ( 1995 ) 162. [ 43 ] J.K.A. Everard, J. Opt. Sensors 2 ( 1987 ) 363. [ 44 ] R.A. Potyrailo and G.M. Hieftje, FACSS XXIV Annual Meeting, Providence, RI, 26^30 October 1997, Paper 371. [ 45 ] J.N. Demas, Excited State Lifetime Measurements, Academic Press, New York, 1983. [ 46 ] S.D. Crossley, Proc. SPIE ^ Int. Soc. Opt. Eng. 2294 ( 1994 ) 14. [ 47 ] V. Ruddy, J. Maguire, Sensors Actuators A 42 ( 1994 ) 525. Radislav A. Potyrailo received his Ph.D. in Analytical Chemistry at Indiana University in 1998. Earlier, he received a degree in Engineering in the ¢eld of Optoelectronics from Kiev Polytechnic Institute. In 1992^1993 he served as a visiting scientist at the University of Toronto. His doctoral work at IU focused on the development of innovative instrumentation, materials,
and signal-processing techniques for chemical and biochemical diagnostics. These studies encompass the areas of evanescent-wave chemical and biochemical sensors, signal generation and processing for spatially resolved analyte mapping, and radioluminescent light sources. His research interests cut across the boundaries of analytical chemistry, environmental science, biochemistry, optoelectronics, image processing, and materials science. He has co-authored over 20 technical papers and over 30 presentations, ¢ve patents, and two books. He was recently awarded the McCormick Science Grant from IU and the Tomas Hirschfeld Award from the Federation of Analytical Chemistry and Spectroscopy Societies. Professor Gary M. Hieftje is Distinguished Professor and Chairman of the Department of Chemistry at Indiana University in Bloomington, Indiana. He received the A.B. degree from Hope College, Holland, MI in 1964 and a Ph.D. in 1969 from the University of Illinois. His research interests include the investigation of basic mechanisms in atomic emission, absorption, and £uorescence spectrometric analysis, and the development of atomic methods of analysis. He is interested also in ¢ber-optic sensors, on-line computer control of chemical instrumentation and experiments, the use of timeresolved luminescence processes for analysis, the application of information theory to analytical chemistry, near-infrared re£ectance analysis, and the use of stochastic processes to extract basic and kinetic chemical information. He currently serves on the editorial boards of Analytica Chimica Acta, Journal of Analytical Atomic Spectroscopy, Laboratory Microcomputer, Spectrochimica Acta, Part B, Advances in Inorganic Mass Spectrometry, the Analytical Chemistry Bench Top Series from Springer Verlag, Talanta, and Spectroscopy and Spectral Analysis. In 1991, he served as President of the Society for Applied Spectroscopy, and received the gold medal of the Quality Control Academy of the Upjohn Company. In 1992, he received the Eastern Analytical Symposium Award for Outstanding Achievements in the Fields of Analytical Chemistry and was awarded a second Lester Strock Medal. In 1993, he was elected as an Honorary Member of the Golden Key National Honor Society and was the recipient of the 1993 Distinguished Faculty Award from the College of Arts and Sciences alumni of Indiana University. In 1996, he received the Humboldt Research Award for Senior U.S. Scientists and the Meggers Award for the year 1995 from the Society for Applied Spectroscopy. Dr. Hieftje is the author of more than 380 scienti¢c publications, has authored or edited 10 books, and holds 11 patents.
Commercial reprints We can supply companies with reprints of any article in TrAC that directly or indirectly support their products at very reasonable rates ( minimum order 100 ). Please apply for details to: TrAC Editorial Of¢ce, P.O. Box 330, 1000 AH Amsterdam, The Netherlands
TRAC 2467 30-10-98