luminescence detection with a liquid core waveguide

Luminescence Detection with a Liquid CoreWaveguide

Purnendu K. Dasgupta,* Zhang Genfa, Jianzhong Li, C. Bradley Boring, Sivakumar Jambunathan, andRida Al-Horr

Department of Chemistry and Biochemistry, Texas Tech University, Lubbock, Texas 79409-1061

A new fluoropolymer tube is proposed as the basis of anovel class of liquid core waveguide-based luminescencedetectors. Both chemiluminescence and photolumines-cence detectors are possible. In the latter case, illumina-tion is transverse to the main axis of the tube. With sucha geometry, it is even possible to operate without mono-chromators, although limits of detection do improve withthe incorporation of monochromators. The nature of thedesign is such that it is particularly simple to fabricatedetectors in a flow-through configuration and where thelight from the cell is coupled to a photodetector by anoptical fiber. No focusing optics are necessary. A numberof applications are illustrated. Attainable limits (LODs,S/N ) 3) of detection include 150 pM fluorescein with a254-nm excitation source, 200 amol of fluorescein in acapillary electrophoresis setup with excitation by two bluelight-emitting diodes, 35 nM NH3 as the isoindole deriva-tive in a flow injection analysis system using a photodiodedetector, 50 nM methylene blue and 1 nM Rhodamine560 using respectively red and green LED arrays and anavalanche photodiode and a PMT in a FIA configuration,100 parts per trillion by volume gaseous formaldehydeas the Hantzsch reaction product with cyclohexanedioneusing a diffusion scrubber, 2.7 µM and 17 nM hypochlo-rite based on its chemiluminescence reaction with luminolwith photodiode and PMT detectors, respectively, and 1ppm SO4

2- based on nephelometric detection at 470 nm.The approach described herein leads to particularlysimple and inexpensive luminescence detectors withexcellent sensitivity.

An optical fiber carries light with minimal loss because of arefractive index (RI) difference between the core and the claddingof the fiber. Light remains trapped in the optically denser fibercore. It has long been realized that long-path absorbance measure-ments for liquid samples can only be realized if the cell behavesas an optical fiber or waveguide. Otherwise, too much light is lostto the walls and excessive noise results. For the absorbancemeasurement cell to function as a liquid core waveguide (LCW),the cell material needs to have a lower RI than the liquid. Earlywork has involved systems such as carbon disulfide (RI 1.63 forNa D-line) in glass tubes (RI typically 1.52)1 or ethanol (RI 1.36)in a fluorinated ethylene-propylene (FEP) copolymer (RI 1.34)

tube.2 However, the most important system of practical importanceremains purely aqueous solutions. Therefore, the most importantdevelopment in this regard has been the introduction of a newamorphous fluoropolymer, Teflon AF, which has an RI less thanthat of water.3 The general area of long-path absorbance measure-ments with LCW cells, with specific reference to Teflon AF-basedcells, has been reviewed.4,5 Tubes of Teflon AF and fused-silicatubes coated externally with Teflon AF are now commerciallyavailable,4-6 as are long-path cells and detectors based on suchcells.7,8

Fujiwara et al. had carried out much of the early work onabsorbance measurements in waveguide cells. It is not surprisingthat the first use of liquid core waveguides for fluorometricmeasurements is also due to these authors.9,10 The generalpreoccupation in these efforts seems to have been the use of alaser as the excitation source. As such, the authors consideredonly situations where the excitation light is launched axially (ornearly so, at a critical angle that fosters the axial propagation ofthe light into the cell) into a tube that may be maintained in alinear, a U-shaped, or a coiled configuration. The fluorescencecan be read axially from the other terminus of the tube. With thisgeometry, very effective means of rejecting the excitation lightare required. The excitation or emitted light or both can beattenuated by the medium, and as such, the maximum fluores-cence signal is reached after a finite cell length. A “side-view cell”was also investigated. A typical implementation involves a spirallyshaped tubular cell; the plane of the spiral is placed next to thephotosensitive window of a “end-on” type photomultiplier tube(PMT) while light is launched axially into the spiral. For thisgeometry, the fluorescence signal is said to be linearly dependenton the refractive index of the solution.9

In terms of continuous illumination energy, nonlaser sourcesstill have the edge, especially when one considers the great

(1) Fujiwara, K.; Fuwa, K. Anal. Chem. 1985, 57, 1013.

(2) Tsunoda, K.; Nomura, A.; Yamada, J.; Nishi, S. Appl. Scpectrosc. 1990, 44,1163.

(3) DuPont Fluroproducts. Teflon AF Amorphous Fluoropolymers. H-16577-1,Wilmington, DE 19880-0711, December 1989.

(4) Altkorn, R.; Koev, I,; Gottlieb, A. Appl. Spectrosc. 1997, 51, 1554.(5) Dasgupta, P. K., Zhang, G., Poruthoor, S. K.; Caldwell, S.; Dong, S.; Liu,

S.-Y. Anal. Chem. 1998, 70, 4661-4669.(6) http://www.polymicro.com/tsu.htm. Information on series TSU fluorocarbon-

coated capillaries, Polymicro Technologies, Phoenix, AZ.(7) http://www.wpiinc.com/WPI_Web/Spectroscopy/LWCC.html.(8) http://www.thermoseparation.com/Set_02.html.(9) Fujiwara, K., Ito, S. Trends Anal. Chem. 1991, 10, 184.

(10) Fujiwara, K., Ito, S.; Kojyo, R.-E., Tsubota, H.; Carter, R. L. Appl. Spectrosc.1992, 46, 1032.

Anal. Chem. 1999, 71, 1400-1407

1400 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999 10.1021/ac981260q CCC: $18.00 © 1999 American Chemical SocietyPublished on Web 02/18/1999

diversity of wavelengths that can be economically obtained fromthem. Thus, to consider only axial illumination in fluorometricmeasurement with a LCW cell unnecessarily limits one’s vision.Many illumination sources are available in the linear format(including linear flash lamps) that can be used to illuminate a tubetransversely along its long axis. Many other sources provide acircular spot of light that can conveniently provide transverseillumination to a tube that has been coiled into a spiral.

To a first approximation, light incident on the surface of anideal fiber (transverse illumination) simply passes radially throughthe fiber and out. It does not propagate down the lumen. However,light can be generated within the lumen of the fiber itself. Thismay occur through excitation by the transversely passing radia-tion, due to elastic or inelastic scattering. Or light may begenerated through a chemiluminescent reaction. A substantialfraction of this light, the precise extent depending on thenumerical aperture of the fiber and the radial position at whichthe light is generated, will be emitted within the launch angle ofthe fiber and will propagate down the fiber. “Scintillating” opticalfibers were originally developed as a simple means of detectingnuclear radiation.11,12 The cores of these fibers is typically dopedwith an organic scintillator, often in conjunction with a fluorescentdye such as fluorescein, serving as a “wavelength shifter”. A fiberlike this can be immersed or imbedded in a medium producingnuclear radiation. Radiation that passes through the fiber has ahigh probability of interacting with a scinitillator molecule andproducing visible radiation. Again, a substantial fraction of the lightemitted propagates down the fiber and is harnessed for conven-tional photometric detection. Such fibers, with a polystyrene coredoped variously with fluorescent dyes that emit in the visible andclad with a lower RI poly(methyl methacrylate) polymer, arereadily available in round or square cross sections.13,14 In examin-ing such a fiber, for example, one doped with fluorescein, theremarkable intense green fluorescence one observes at the terminiis qualitatively the same, regardless of whether one is underdaylight, fluorescent light, or tungsten light. Of course, this isbecause the nature of the emitter is the same. However, the resultsare dramatically different if the surface of the fiber is covered upand the same suite of excitation lights is launched, one at a time,through one of the termini and the resulting light, the sum of thefluorescent emission and the remnant excitation light, is viewedthrough the other. The striking conclusion is that when fluores-cence is axially observed in a transversely illuminated optical fiber,it can be observed by itself, with little interference by the excitationlight. If we translate the same situation to a liquid core opticalfiber, axial fluorescence detection should be possible with a“white” light source illuminating the surface of such a fiber,without any excitation or emission monochromators.

In this paper, we examine the utility of this uniquely simplegeometry for fluorescence and other related luminescence detec-tion methods.

EXPERIMENTAL SECTIONMaterials and Equipment. Teflon AF 2400 capillaries in

different sizes were obtained from Biogeneral Inc. (San Diego,

CA). Teflon AF 1600-coated fused-silica capillaries (360-µm o.d.,100-µm i.d., 15-µm coating) were obtained from PolymicroTechnologies (Phoenix, AZ). Light sources used in this workinclude light-emitting diodes (LEDs, Panasonic high brightnessLNG992CFBW emitting at 470 nm with a half bandwidth of 40nm, Nichia superluminescent 590S, emitting at 495 nm, GilwayE201, emitting at 660 nm with a half bandwidth of 60 nmstheseare of only modest brightness), low-pressure mercury pen lampsemitting at 254 nm and a phosphor-coated version of the same(∼0.25-in. diameter, illuminated lengths 1, 2, and 7 in., BHK Inc.,Claremont, CA), miniature versions of similar 254-nm lamps inboth glass and quartz envelopes (JKL Components Corp., Pa-coima, CA), mercury black light 365-nm lamps in customary blackglass envelopes in standard F4T5 (0.5 × 4.25 in.) and subminiaturesizes (3-mm diameter, 5 mm and 20 mm long, JKL Components),linear Xe flash lamp source (model FLS 01, Chapparal Technolo-gies, Albuquerque, NM; up to 10 J discharged/flash), andhousehold-type white fluorescent lamps. Detectors used rangedfrom very inexpensive silicon photodiodes (3 × 3 mm, type S2007,Electronic Goldmine, Phoenix, AZ) and various photodiodes(Hamamatsu Corp. and others). These were typically used withsimple home-built operational amplifier based current to voltageconverters; in some experiments, a commercial picoammeter(Keithley model 480) was used. For higher sensitivities, we useda miniature PMT equipped with its own high-voltage (HV) powersupply and current to voltage converter (H5784) and a tempera-ture-controlled avalanche photodiode (APD) module equippedwith its own HV supply (S5460, both from Hamamatsu). Weshould like to emphasize that with the exception of the pulsedXe source and the commercial current amplifier (neither of whichis involved in the best results presented in this work), the partscost to build any complete luminescence detector including a flow-through cell, light source, and all necessary power supplies andsignal processing electronics, ranges from under $100 to $1000.

The general design shown in Figure 1 was used for experi-menting with the detection scheme. An opaque PEEK tee T for1/8-in. tubes (Upchurch Scientific) constitutes one end of the cell.In the center of the tee, an LCW tube AF butts against an 1-mmcore acrylic or silica optical fiber F that is coupled to a photode-tector. The liquid enters through the perpendicular arm of thetee, enters the AF tube through the gap that unavoidably remainsbetween AF and F, and proceeds through AF to the other endwhere a connecting tubing is put in with a chromatographic male-male union U. A light source L illuminates the LCW as shown inthe figure. For subminiature fluorescent lamp or LED excitationsources (in single or arrayed configuration), isolation from externallight is provided by an opaque tubular shell S that fits within thehub of the nut in fitting T and within the inner half of U (whichis drilled out to accommodate it). Electrical leads to L are broughtout through the walls of S. The inside of the shell S was polishedto improve excitation light throughput. Larger sources requireda larger shell, put in on the exterior of the nut at T and that of U,with a spacer element, if necessary. Many experiments wereconducted where light shielding was simply provided by a darkcloth draped over the experimental arrangement, or the setup wasput inside a cardboard box.

For optical filtering, interference filters were used in a fewexperiments but most commonly we used inexpensive colored

(11) Binns, W. R.; Israel, M. H.; Klarmann, J. Nucl. Instrum. Methods Phys. Res.1983, 216, 475.

(12) White, T. O. Nucl. Instrum. Methods Phys. Res. 1988, A273, 820.(13) www.edsci.com.(14) http://www.bicron.com/fibers.htm.

Analytical Chemistry, Vol. 71, No. 7, April 1, 1999 1401

plastic sheets available from Edmund Scientific (Catalog No.E60403), and filter numbers quoted in the text refer to thenumbers from this vendor.

Capillary electrophoresis experiments were conducted usinga homemade CE instrument that has been briefly describedpreviously.15 A 60-cm-long Teflon AF-1600-coated 100-µm-i.d.capillary (TSU100360, Polymicro Technologies) was used and thehardware for a radial path homemade capillary absorbancedetector16 was used to hold a pair of blue LEDs on either side ofthe capillary, ∼10 cm from the grounding end. The groundingend of the capillary was put in a subminiature polypropylene teeand a 1-mm core silica fiber was butted against it. The other endof the fiber addressed the PMT to perform fluorescence detection.Fluid communication with the background electrolyte (BGE, 4mM Na2B4O7) in the vial is maintained through the tee arm, andthe ground electrode is maintained in the vial. Injection washydrodynamic (∆h ) 10 cm, t ) 5 s, which resulted in an injectedvolume of 24 nL), and electrophoresis was conducted at an appliedvoltage of +20 kV.

Nephelometric experiments were conducted with a 470-nmLED as the transverse illumination source located 1.5 cm from a1-mm core silica fiber in a cell built with 0.79-mm-i.d. Teflon AFtubing. The detection of sulfate by the classical, albeit relativelyinsensitive technique BaSO4 turbidimetry was studied in a flowinjection configuration. The sulfate sample (65 µL) was injectedinto a water carrier (160 µL/min) and merged with an equal flowrate of a reagent stream containing 5% BaCl2, 10% ethanol, and10% ethylene glycol. The detection cell was placed after a 60-cmlength of 0.7-mm-i.d. reaction coil.

Phosphorimetric detection was studied with the pulsed Xe lightsource (arc length 13 cm, 3.5 J/pulse used for the data reportedhere). This linear flash lamp source was located within a reflectiveelliptical cavity, the lamp being located on one of the foci and a0.79-mm-i.d. Teflon AF tube (40 cm) running through the otherfocal point. One side of the AF tube is connected to a waste line;the other side is connected to a tee for solution input and opticalconnection to a 1-mm core silica fiber that terminates in a UV-sensitive photodiode (Hamamatsu HUV1000 BQ). The photodiodeoutput was converted to a voltage signal by a rapid response i toV converter (Keithley model 427) and read by a storage oscil-loscope (Tektronix, model 5103N).

Except as stated, reagents were obtained from Aldrich Chemi-cal and used without further purification.

RESULTS AND DISCUSSIONRejection of Excitation Light. In a conventional fluorometric

setup, fluorescence is viewed at right angles to the excitationbeam, typically using 10-mm path length cells. Due to reflectionand scattering at the windows and at the interfaces, some lightreaches the detector even with the latter placed at right angles.Using 5-mm circular apertures with a conventional cell filled withwater and a divergent source, we measure that 0.06-0.1% of thesource light is scattered into the detector. When the sameexperiment is conducted with a LCW cell, with the cell illuminatedover a ∼3-mm length, ∼100 mm away from the detector, 0.001-0.002% of the source light can be measured to be axiallypropagating through the cell. It is this degree of rejection of theexcitation light that allows fluorescence measurement without anyor with only rudimentary wavelength discrimination in the lightpropagating through the lumen of the LCW cell. It is also thischaracteristic that makes possible the uniquely simple design ofsuch fluorescence detectors.

Light Conduction to a Photodetector. Initial experimentswere conducted with a design in which the detector end of theLCW cell ended in a thin transparent glass window made of amicroscope slide cover glass. This was placed directly on top ofthe PMT window. Studies with the design described in theExperimental Section showed, however, that light is just asefficiently and far more conveniently conducted to the detectorwhen a high numerical aperture fiber (a silica fiber of the sameor greater core diameter as the inner diameter of the LCW cellor a larger diameter acrylic fiber) is butt-jointed to the LCW tube.This arrangement was used henceforth. When used, coloredplastic filters could be directly attached to the detector end ofthe fiber optic by UV-cure adhesive or a suitably cut circle of thefilter put on the face of the photodetector.

LCW Cell Geometry. With a core RI of 1.33 (water) and apolymer RI of 1.29, a Teflon AF 2400 tube has a numerical apertureof ∼0.32 and an acceptance angle of 37°. Coiling any fiber into asmall-diameter coil causes loss of light from the lumen; the exactcoil diameter at which the loss becomes significant depends onthe numerical aperture. However, theoretical computations aloneare of limited help because this is also greatly influenced by theimperfections in a fiber. With water-filled Teflon AF 2400 tubes,we find that loss of light becomes significant only below coildiameters of 5 cm. Not only light is lost from the lumen in small-diameter coils; the ability of transverse illumination to propagatethrough the axial mode increases. The signal-to-background ratiois generally unacceptably poor if, for example, the LCW cell iscoiled on a conventional 3.7-cm-diameter fluorescent lamp (as

(15) Liu, S.; Dasgupta, P. K. Anal. Chim. Acta 1992, 268, 1.(16) Boring, C. B., Dasgupta, P. K. Anal. Chim. Acta 1997, 342, 123.

Figure 1. Typical arrangement for fluorescence detection. F, acrylate or silica optical fiber butted against LCW tube AF in the center of teefitting T. The other end of AF is connected to waste tubing WT by compression fitting union U. A tubular shell S houses the tubing AF and thelight source L.

1402 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999

commonly used in overhead lighting applications). Most lampsare smaller than this in diameter. So coiling an LCW cell directlyon a lamp source (in many cases heat dissipation would also bean additional problem) is not practical. In all further work, wetherefore used the LCW cell in a linear configuration.

Effect of Illumination Length. This was investigated by usinga 17-cm-long Hg pen lamp emitting at 254 nm. The LCW cell (0.79-mm i.d.) was placed along the length of the lamp, a few millimetersabove its top. Fluorescence intensities from a 10, 30, 50, 100, and200 nM fluorescein solution and the absolute level of thebackground were measured in a FIA configuration. The lamp wascovered with a metal sheet. By moving the sheet, different lengthsof the lamp could be exposed. The tee end with the optical fiberwas 10 cm distant from the exposed terminus of the lamp.Exposed lengths of 1, 4, 8.5, and 17 cm were studied. In eachcase, the response was linear with concentration with an interceptthat was statistically indistinguishable from zero and the linear r2

value ranged from 0.9994 to 0.9998. The response slope increasedwith increasing length, being in the ratio 1: 5.0:7.3:15 for the fourabove illumination lengths. Considering that the intensity ofillumination is not identical in different portions of such a lampand it is difficult to have a very well defined length of the lampproviding the illumination, these results do indicate that theobserved fluorescence intensity increases in direct proportion tothe illumination length. The background signal, on the other hand,increases in a proportion less than linearly with the illuminationlength, being in the ratio of 1:2.6:3.8:6.0 for the 1:4:8.5:17 cmillumination lengths. Moreover, at least under these conditions,the detector noise does not increase in proportion to thebackground signal, the ratio being 1:1:1.5:2 in the order of

increasing illumination lengths. Thus, the LOD improves withincreasing illumination length.

Figure 2 shows replicate injections of a 10 nM fluoresceinsolution (illuminated volume 39 µL). The LOD (S/N ) 3) iscomputed to be 150 pM. The inset shows a similar experimentwith a commercial fluorescence detector for HPLC (cell volume20 µL) made by a major manufacturer, set at the optimumexcitation and emission wavelengths for the measurement offluorescein. The superior performance of the present detector isreadily apparent. Admittedly, the excitation source of the com-mercial detector is no longer in its prime, S/N measurement forquinine sulfate indicates a performance 3 times worse relative tomanufacturer’s specifications. However, the performance differ-ence between the present detector and the commercial detectoris considerably greater even after accounting for this and thedifference in illuminated volume (which accounts for a factor of21/2; see below).

It was also of interest to us to determine that if light intensityis increased, what type of LODs may be attainable with aninexpensive photodiode detector. The linear flash lamp sourceilluminated a 0.79-mm-i.d. LWC, placed in its close proximity,protected by a roll of blue plastic (No. 35,135). A silica fiber ledthe emitted light to a very inexpensive photodiode covered withNo. 877 green plastic. The flash lamp was fired with 5-J pulses.The actual flash duration was of the order of 1 µs, but the highcapacitance of the photodiode broadened the observed responseto a half-width of 40 µs. The S/N was found to increase directlywith the square root of the number of pulses accumulated, with100 pulses, the LOD was 12 nM.

Effect of Illuminated Volume. A more detailed study of theeffect of varying both the illumination length and the diameter ofthe tube was conducted. From basic considerations, S/N shouldimprove with the square root of the illuminated volume, whichshould therefore be linearly related to the LOD. This proves tobe the case. Figure 3 illustrates this for tubes of three differentdiameters and different illumination lengths. A clear establishment

Figure 2. Response of repeated injections of 10 nM fluorescein ina FIA system. LCW cell i.d. 0.79 mm, illuminated length 80 mm, 254-nm Hg lamp unfiltered illumination, and PMT detector with No. 877green plastic filter. The upper inset shows the response from acommercial HPLC detector for the same sample.

Figure 3. Reciprocal of the limit of detection plotted against thesquare root of illuminated volume.


of this basic behavior is very useful because it allows one tocalculate a priori what changes in performance can be expectedon the basis of changing cell volume or tube diameter.

Illustrative Examples. It appears to us that the real merit ofthe proposed approach is not in general studies of luminescencespectrometry but in actual applications to design a dedicateddetector that may constitute an inexpensive module for a completemeasurement system.

Formaldehyde. Formaldehyde is a simple compound ofconsiderable environmental interest that can be selectively andsensitively determined by a fluorogenic reaction with cyclohex-anedione and ammonium acetate; a FIA adaptation of this reactionhas been published.17 The optimum excitation and emissionwavelengths are 395 and 465 nm. However, both are broad andthis allows considerable latitude in choosing excitation wave-lengths and in the choice of detectors/filters. The signal-to-noiseperformance of different types of photodiodes were as follows (therelative S/N is indicated): G1115 (GaAsP) and S1226-5BQ (UV-

enhaned silicon) 11, S2007 (silicon) and G3273 (large-area high-performance silicon) 17, G118 (GaAsP, diffusion type) 85, andG1742 (GaAsP, Schottky type) 100. With direct fiber coupling toa photodiode, a large active sensing area is not necessary, anysensing area beyond the fiber optic diameter merely contributesto noise. Lens ends are not desirable either. In this particular case,best results are obtained with small active area high quantumefficiency GaAsP photodiodes where the active surfaces are closeto the top of the case. The two best performers cost less than $10each. Variations in the lamp characteristics also affect theperformance. Table 1 lists the attainable LODs with a G1118photodiode, a light blue (No. 856) plastic filter, and severaldifferent types of lamps. Of course, in most applications, a choicehas to be made based on the cost/performance ratio. Here theperformance attainable with thin (3-mm diameter), inexpensive(less than $20, including power supply) blacklight (365 nm) lampsis particularly attractive. The detector geometry shown in Figure1 is such that it is readily amenable to clustering several lampsaround the LCW tube. Up to four lamps clustered around the LCWtube were studied. As may be intuitive, the response improvesand the LOD decreases with the number of lamps. Of course, amuch greater increase in the LOD is achieved with a PMTdetector. Figure 4a shows the response from 100 nM HCHO witha 3 × 50 mm 365-nm lamp as the excitation source and a PMTdetector. Figure 4b shows the response from repeated injectionsof 2.5 parts per billion by volume levels of gaseous HCHO (a keymarker molecule in atmospheric photochemistry studies) in zeroair, using a Nafion membrane diffusion scrubber as a collectorand the same detection arrangement.17

Ammonia. Ammonia is another simple compound the deter-mination of which is needed over a very wide range in differentapplications. The sensitivity of applicable methods is taxed when(17) Fan, Q.; Dasgupta, P. K. Anal. Chem. 1994, 66, 6, 551.

Figure 4. (a) Response from 100 nM formaldehyde, PMT detector with No. 856 (blue) plastic filter; (b) long-term response stability fromrepeated exposures to 5 ppbv level of gas-phase formaldehyde collected by a Nafion membrane diffusion scrubber and same detector.

Table 1. Lamp Performance in Fluorometric LCWMeasurement of Formaldehyde

lamp typeLOD,nM

4-cm active length Hg pen lamp 254 nm 342.5-cm active length Hg pen lamp 406 nm phosphor coated 804-cm active length F4T5 1.25-cm-diameter blacklight,

365 nm150

4-cm active length 11 × 90 mm glass envelope Hg lamp,254 nm

13

4-cm active length 8 × 100 mm quartz envelope Hg lamp,254 nm

3

3 × 50 mm miniature Hg blacklight lamp, 365 nm 11


one needs, for example, to measure the concentration of ammoniaover remote oceans.18 One of the more sensitive methods is thederivatization of ammonia with sulfite and o-phthalaldehyde toform 1-sulfonatoisoindole; FIA adaptations19 and application to themeasurement of gaseous ammonia20 have been described. In thiscase, the optimum excitation wavelength is 365 nm, perfectlymatched by a Hg blacklight lamp. The emission is centered at425 nm. The detection limit with a commercial fluorescencedetector with a PMT detector equipped with a long-pass excitationfilter was 20 nM.19 With the same excitation arrangement as forformaldehyde above, the fluorescence in this case is sufficientlyintense that a detection limit of 35 nM could be obtained with$15 detector (blue/UV-sensitized photodiode detector with integralamplifier (Burr-Brown OPT 301) with blue (No. 856) plastic filter).

Methylene Blue and Rhodamine. Methylene blue (MB) isan intensely colored dye that absorbs in the deep red (absorptionmaximum 664 nm). It is widely used as a biological stain and alsoas a cationic ion-pairing agent for measuring anionic surfactantsin water and wastewater. MB fluoresces with an emissionmaximum at 690 nm. Although the emission is not particularlystrong, the fluorescence is selectively quenched by purine nucle-otides; this has important applications in biological analysis.21 Thedye or its ion pairs/conjugates are usually determined by absorp-tion spectrometry; the fact that it fluoresces is less widelyexploited. Spaziani et al.22 recently described a diode laser-based

fluorescence detector in which sulfide is determined via theformation of MB.23 In our experiments, to excite MB, a 29-mm-long LED array consisting of 14 LEDs, connected in parallel (eachwith its own current limiting resistor), was constructed byremoving much of the epoxy molding from each LED (both fromthe sides and top), cementing them together with epoxy adhesive,and polishing the top of the array to create a flat surface. Thearray was placed in close lateral proximity of the LCW tube. Anavalanche photodiode is well suited for the detection of fluores-cence in this case. While red-sensitive PMTs are expensive, siliconAPDs have usable sensitivities at such wavelengths. Moreover,while the active area of a PMT tends to be a minimum of 1 cm indiameter, APDs are readily available with active area diametersof the order of 1 mm, which is perfectly adequate for coupling toa fiber optic as used in the present cells and obviates the needfor more expensive APDs with larger active areas. A response toan injection of 670 nM MB in a FIA system with water carrier isshow in Figure 5a, the S/N at this level is such that an S/N ) 3LOD of 50 nM can be estimated.

A similar experiment is shown for Rhodamine 560, anotherfluorescent dye widely used as a tag. In this case, much brightergreen LEDs, two arrays (of six LEDs each) were deployed onopposite sides of the LCW tube, providing an illuminated volumeof 12 µL. The results for 3 nM injections with a PMT detector areshown in Figure 6b; the baseline appears between the two setsof injections and the fluctuations are largely due to pumppulsations. Even then, the LOD is equal to at least 1 nM. With a(18) Genfa, Z.; Uehara, T.; Dasgupta, P. K.; Clarke, T.; Winiwarter, W. Anal.

Chem. 1998, 70, 3656.(19) Genfa, Z.; Dasgupta, P. K. Anal. Chem. 1989, 61, 408.(20) Genfa, Z.; Dasgupta, P. K.; Dong, S. Environ. Sci. Technol. 1989, 23, 1467.(21) Dunn, D. A.; Lin, V. H.; Kochevar, I. E. Photochem. Photobiol. 1991, 53,

47.

(22) Spaziani, M. A.; Davis, J. L.; Tinani, M.; Carroll, M. Analyst 1997, 122,1555.

(23) Kuban, V.; Dasgupta, P. K.; Marx, J. N. Anal. Chem. 1992, 64, 36.

Figure 5. (a) Response to 670 nM aqueous methylene blue in a FIA system, 5 mM HCl carrier, LWC illuminated by a 660-nm red LED array,illuminated volume 15 µL, avalanche photodiode detector, and No. 35136 plastic filter. (b) Response to 3 nM Rhodamine 560, LWC illuminatedby dual array of green LEDs, illuminated volume 12 µL, No. 806 plastic filter, and PMT detection.


pulseless pumping arrangement, the LOD is expected to be inthe high picomolar range.

Capillary Electrophoresis. Detection of molecules taggedwith fluorescent taggants is in extensive use in the practice ofcapillary electrophoretic separations, especially in biological andbiomedical research. The separation and detection of amino acidsas derivatized with fluorescein isothiocyanate (FITC) is in wideuse, for example. The limits of detection for FITC-derivatizedamino acids are typically in the low-attomole range with laser-induced fluorescence detection.24

When a Teflon AF-coated fused-silica tube filled with anaqueous liquid is illuminated, both the water core and the silicawall can act as waveguides. If such a tube is filled with water andtransversely illuminated with the blue LEDs as in the presentexperiment, dark-field microscopic examination of the tip of thecapillary shows that a small amount of blue light is transmittedthrough the silica wall. If on the other hand the tube is filled witha fluorescein solution, it can be clearly observed that thefluorescence emission is confined to the liquid core. To reject thestray excitation beam transmitted through the silica wall, one caneither provide an opaque coating to the tip of the capillary or usea collection fiber that is matched to the core diameter of the LCWsuch that it collects only the light conducted by the liquid core.

We chose the first alternative. The face of the capillary wasmade opaque by heavy silvering and the emitted fluorescent lightcarried through the liquid core was transmitted to a large (1-mm)-diameter silica optical fiber. The results are shown in Figure 7.The injected sample amount is 2.5 fmol. We estimate an LOD of∼200 amol under these conditions. The LEDs were very modestlydriven in these experiments, if they are operated in a pulsed modemuch higher illumination intensities can be obtained. We note inthis context that if four different colored LEDs are staggeredaround the capillary to selectively excite four different coloredfluorescent tags that themseleves selectively attach to fourdifferent nucleotide bases, an excitation pulse train sequence can

be used to perform four-color sequencing very effectively andinexpensively.

Chemiluminescence (CL) Detection. Aqueous chlorine andhypochlorite are important analytes that can be measured by theirreaction with luminol that results in CL. Under carefully optimizedconditions, the LOD has been reported to range from 0.3 26 to 12µg/L25 for a custom-built luminometer (S/N ) 3). In ourexperiments, the same detection arrangement as used for thefluorometric measurement of ammonia and/or formaldehyde wasused, except the lamp was not turned on and no filter was used.Using 2 mM luminol in a 0.2 M carbonate buffer (pH 11.2) in aflow injection analysis (FIA) configuration, the LOD was 2.75 µMClO- with a photodiode detector (as used for ammonia above)and 15 nM ClO- for the PMT detector (as used for formaldehyde).System output in the latter case is shown in Figure 6.

It is known that ammonia rapidly reacts with hypochlorite toform chloramine; the latter is incapable of eliciting CL fromluminol.27 Injection of ammonia into a luminol-hypochloritesystem (that provides a CL background signal) thus producesnegative CL signals. Using this principle and a ammonia-hypochlorite reaction time of 90 s, we were able to attain LODsof 9.5 µM NH3 for the photodiode detector and 0.34 µM for thePMT detector. These results will be published in greater detailin a future paper.

(24) Mattusch, J.; Dittrich, K. J. Chromatogr., A 1994, 680, 279.

(25) Gonzalez-Robledo, D.; Silva, M.; Perez-Bendito, D. Anal. Chim. Acta 1990,228, 123.

(26) Marino, D. F.; Ingle, J. D., Jr. Anal. Chem. 1981, 53, 455.(27) Balciunas, R. T. An Automated Reagent Preparation System for Fast Reaction-

Rate Analyses. Ph.D. Dissertation, Michigan State University, 1981.

Figure 6. Chemiluminescence signals recorded by a PMT detectorfor 500 nM ClO- in a luminol reaction system in a FIA configuration.

Figure 7. Response from a mixture of 100 nM fluorescein (tm ∼3.5min) and 100 nM carboxyfluorescein (tm ∼5 min) in a capillaryelectrophoresis system. Computed injection volume 25 nL. See textfor details.


Nephelometric Detection. Formation of particles in anysystem causes scattering. Consequently, precipitation can bedetected by the present detection arrangement. The length of anephelometric cell, specifically, the distance between the excitationsource and the receiving fiber, must be relatively short, else thescattered light will be attenuated by the flowing scattering particlesresponsible for the nephelometric signal. Using the arrangementdescribed in the Experimental Section, the LOD for injected sulfatewas 0.9 mg/L (S/N ) 3). This is quite comparable to LODsattainable with the current practice of BaSO4 turbidimetry.

Phosphorimetric Detection. An interesting room-tempera-ture phosphorimetric detection technique, applicable to solutions,has recently been described by Segura Carreto et al.28 For severalnaphthalene derivatives, phosphorescence signals can be elicitedby adding large concentrations of a heavy atom salt (e.g., KI) inthe presence of an oxygen scavenger such as Na2SO3. Using theexperimental arrangement described, the LOD was found to belimited primarily by the variations in the blank signal. In a mediumof 0.25 M KI and 12.5 mM Na2SO3, we found the S/N ) 3 LODfor naphthoxyacetic acid to be ∼1.2 µM (∼150 µg/L) with the

pulsed Xe lamp excitation system. Without the KI-Na2SO3 matrix,the signal from even a 1000-fold more concentrated analytesolution was indistinguishable from the blank.

Raman Scattering. It is obvious that Raman scattered lightwill also be guided by the LCW cell and can be detected. This isan important outcome on its own and such results will appear ina separate paper.29

CONCLUSIONSWe have described here some very simple and inexpensive

but nevertheless sensitive arrangements to perform various typesof luminescence detection, made possible by LCW cells fabricatedfrom a new fluoropolymer with a RI lower than that of water. Thetransverse illumination LCW fluorescence detection techniquegreatly reduces the need for high-throughput monochromators,and the general geometry is ideally suited for efficient fiber opticcoupling permitting the use of a variety of remotely locateddetectors.

Received for review November 17, 1998. AcceptedJanuary 20, 1999.

AC981260Q

(28) Segura Carretero, A.; Cruces Blanco, C.; Canabate Dıaz, B.; FernandezGutierrez, Anal. Chim. Acta 1998, 361, 217.

(29) Holtz, M.; Dasgupta, P. K., Genfa, Z., unpublished work, Texas TechUniversity, 1998.


luminescence detection with a liquid core waveguide

Documents