Anal. Chem. 1995, 67,2224-2230
Chemiluminescent Chemical Sensors for Oxygen and Nitrogen Dioxide Greg E. Collins* and Susan L. RossPdwsson Chemistly Division, Code 61 10, Naval Research Labotatoly, Washington, 0.C., 20375-5342
Chemiluminescent chemical sensors for oxygen and nitrogen dioxide have been investigated via the immobilization of 3-aminophthalhydrazide (luminol) within a hydrogel or polymeric, sorbentcoalingthat is positionedin h n t of a photomuhiplier tube. Some selectivityis tailored into these devices by careful selection of the polymer type, pH, and metal catalyst incorporated within the film. Oxygen levels as low as 2.4 ppm in nitrogen have been detected using the oligomer fluoropolyol as the support matrix for immobilizing luminol, KOH, and the metal catalyst, F a (so& For the detection of NO%@),the use of sorbent, polymer coatings such as fluoropolyol or polyethylenimine resulted in strong humidity dependencies, exhibiting no chemiluminescence under dry conditions. Incorporation of the chemiluminescentreagents within a hydrogel (e.g., poly(viuyl alcohol) or superabsorbing polymer Waterlock) eliminated this effect. A survey of various hydrogel immobilization matrices and metal catalysts found that, for the determinationof N02(g), Cu(II) and the superabsorbing polymer Waterlock provided the best film performance with respect to sensitivity (0.46ppb) and stability. Discussions center about the complete characterization of these chemiluminescent chemical sensors and an examination of specific methods used to extend the sensitivity and selectivity of these devices. Various chemiluminescence schemes have been successfully applied to the detection of metals, polycyclic aromatic hydrocarbons, and numerous oxidizing and reducing agents.' These methods typically rely upon the oxidation of a chemically reactive species, e.g., luminol or lucigenin, and the subsequent emission of a photon from an electronic, excited-stateintermediate. Chemiluminescent techniques are free of Raman and Rayleigh scattering, interferences associated with fluorescence techniques, and, therefore, permit the operation of a photomultiplier tube at maximum sensitivity. The primary attraction of chemiluminescence detection is the excellent sensitivity obtainable over a wide dynamic range using simple instrumentation. The vast majority of work done in the chemiluminescence field has been devoted to solution-based,flow injection analysis formats. One of the most widely investigated chemiluminescence reagents is 3-aminophthalhydrazide, or luminol. Luminol has been effectively used for the detection of many different metals and oxidants in solution, due to the catalytic behavior these metals have on the chemiluminescent oxidation of luminol. Examples include the detection of Cr(III),2 CO(II),~Fe(ID,4 HzO~(aq),~ (1) Robards, IC;Worsfold, P. J. Anal. Chim. Acta 1992,266,147-173.
2224 Analytical Chemistry, Vol. 67,No. 13, July 1, 1995
NO(aq) !and ClO-(ag) ? These methods require adequate mixing of the reagent and unknown solutions prior to passage before a photomultiplier tube (PMT), accurate control of the solution delivery to the PMT, and the consumption of a substantial amount of reagents. Different instrumental variations, based upon the chemiluminescent oxidation of luminol, have been devised for the detection of gas-phase oxidants. Maeda et al. designed a chemiluminescent reaction compartment for the detection of NO&) which collected a pool of alkaline luminol solution directly below a PMT and introduced the sampled airstream into the region above the solution! Excess luminol solution and air flowed continuously out of the reaction compartment through an outlet port at the base of the system. This system was very sensitive to movements of the compartment and had a relatively slow time response. An alternate design for the chemiluminescent detection of NO2@) was subsequently presented by Wendel et al., wherein a length of filter paper was positioned adjacent to a PMT, and a flow of alkaline luminol solution was directed down the paper in a fine film.9 This system has also been used for the measurement of trace levels of ambient ozone, utilizing the chemiluminescent dye eosin Y.lo A further variation on the instrument described by Wendel et al. led to the development of a commercially available instrument devoted to the measurement of NOz(g), the Luminox LMA-3 (Scintrex/Unisearch); a detailed description of this instrument was presented by Schiff et al." The reaction cell consists of a fabric wick positioned in front of a PMT that is continually wetted with fresh luminol solution delivered via a peristaltic pump, and whose surface is exposed to a stream of the ambient air pumped through the cell. In recent years, this instrument has been utilized in combination with various pretreatment stages for the trace detection of organic nitrate~.'~J~ Still another variation on the luminol chemiluminescent detector for NOz@) was introduced by MikuSka and Vefei-a; this configuration employs a ~~~
~~
~~
~~
~
~~
(2) Escobar, R; Lin, Q.; Guiraum, A; de la Rosa, F. F. Analyst 1993,118,643-
647. (3) Cheng, C. A; Patterson, H. H Anal. Chem. 1980,52,653-656. (4)Seitz, W. R;Hercules, D. M. Anal. Chem. 1972,44, 2143-2149. (5)Shaw, F.Analyst 1980,105,11-17. (6)Kikuchi, IC;Nagano, T.; Hayakawa, H.; Hirata, Y.; Hirobe, M. Anal. Chem.
1993,65,1794-1799. (7)Marino, D.F.;Ingle, J. D., Jr. Anal. Chem. 1981,53,455-458. (8) Maeda, Y.;Aoki, IC; Munemori, M. Anal. Chem. 1980,52,307-311. (9)Wendel, G. J.; Stedman, D. H.;Cantrell, C. A; Damrauer, L. Anal. Chem. 1983,55,937-940. (10)Ray, J. D.; Stedman, D. H.; Wendel, G. J. Anal. Chem. 1986,58,598-600. (11) Schiff, H. I.; Mackay, G. I.; Castledine, C.; Harris, G. W.; Tran, Q. WaterAir Soil Pollut. 1986,30,105-114. (12)Blanchard, P.; Shepson, P. B.; Schiff, H. I.; Drummond, J. W. Anal. Chem. 1993,65,2472-2477. (13)Hao, C.; Shepson, P. B.; Dmmmond, J. W.; Muthuramu, IC Anal. Chem. 1994,66,3737-3743. 0 1995 American Chemical Society 0003-2700/95/0367-2224$9.00/0
continuous spray of luminol solution, directed immediately below the PMT, and generated from a stream of the analyzed gas.14Each of the systems referenced above requires accurate and continual pumping or delivery (e.g., peristaltic pump) of an aqueous luminol solution, be it across a piece of filter paper, a wick, or mixed with an additional solution or gas containing the analyte of interest. The systems described above are not amenable to remote sensing or personal dosimeter applications because of constraints with respect to size, weight, power, or portability. The industrial community is continually striving to develop sensor systems that are simpler, less expensive, and more compact and offer the possibility for remote sensing or personal dosimeter applications. In order to address some of these constraints, efforts were focused upon the development of a solid-phase, chemiluminescent, chemical sensor. These sensors are based upon the incorporation of the chemiluminescent reagent luminol and an appropriate metal catalyst within a polymeric thin film or hydrogel that is supported in direct view of a PMT. The objective of this work is twofold (1) to further simplify and miniaturize the instrumental requirements for chemiluminescent measurements and (2) to introduce some selectivity, sensitivity, or both through the incorporation of the chemiluminescent reagents withii a polymeric network bearing solubility characteristics altogether different from that of water. The successful incorporation of chemiluminescentreagents withii polymeric matrices could lead to the formation of very compact sensor systems which consist of simply the chemiluminescent gel, a PMT, and a small pump for sampling air across the surface of the polymer or dfision membrane. The latest development by Hamamatsu, a miniature photomultiplier tube supported on a TO-8 type metal can package (15 mm in diameter and 10 mm in height) that requires simply a 15 V power supply, is ideally suited to the development of a compact, low-power sensor system which has the possibility of being adapted for remote sensing and portable dosimeter applications. Previous authors have demonstrated the feasibility of utilizing luminol in a reagent-less fashion, i.e., using a solid substrate support. Agranov and Reiman attempted the direct application of luminol, sodium carbonate, and copper sulfate onto indicator tape for the detection of hydrogen peroxide but were plagued by humidity and stability pr0b1ems.I~Freeman and Seitz determined hydrogen peroxide concentrations in solution by immobilizing luminol and peroxidase within a polyacrylamide gel held on the end of a fiber-optic probe.16 The co-immobilization of dehydrogenase enzymes and luminol withii a poly(viny1 alcohol) matrix has been proposed for the detection of ATP or NAD(P)H.17 Luminol has also been covalently immobilized onto silica particles for the detection of hydrogen peroxide by flow injection analysis.'* Selectivity is one of the major drawbacks limitingthe analytical applicability of the luminol chemiluminescencereaction. This is not true for all chemiluminescent reactions; e.g., peroxyoxalate chemiluminescence is based on the very selective reaction of an aryl oxalate ester or oxamide with hydrogen peroxide, wherein the energy produced from this chemical reaction is transferred to a fluorescent species added to the matrix and emitted in the (14) MikuSka, P.; Vetefa, Z. Anal. Chem., 1992,64, 2187-2191. 1533-1538. (15) Agranov, Kh. I.; Reiman, L. V. Zh.Anal. Khim. 1979,34, (16) Freeman, T. M.; Seitz, W. R Anal. Chem. 1978,50, 1242-1246. (17) Coulet, P. R; Blum, L. J.; Gautier, S. M. Sem. Actuators B 1993,21, 5761. (18) Hool, IC; Nieman, T. A. Anal. Chem. 1988,60, 834-837.
p=pqpjq
" Supply
Ammeter
Figure 1. Experimental setup for monitoring solid-phase chemiluminescence.
form of light.lg Our aim was to introduce some selectivity into the luminol-based Chemiluminescent, chemical sensors discussed here by proper choice of the chemiluminescent reagent, catalyst, pH, and polymeric substrate. We have examined a number of different polymers whose sorbent properties have been characterized with respect to their hydrogen bond acidity and basicity, dipolarity, polarizability, and ability to distinguish between members of a homologous series.20Careful choice of the polymer used to incorporate the chemiluminescent reagents helps determine the types of analytes concentrated within the organic thin film. The selectivity of these chemical sensors can be further extended by choosing a chemiluminescentreagent, catalyst, and pH that is more responsive to the analyte of interest. This paper will discuss the development of chemiluminescent chemical sensors for oxygen and nitrogen dioxide, each of which employs specific hydrogels or polymeric thinfilms of luminol incorporating different catalytic metals and pH conditions. EXPERIMENTAL SECTION Apparatus. The instrumental setup consisted of a small cell supporting the polymercoated, glass substrate behind a glass window in full view of a PMT (see Figure 1). Inlet and outlet tubes (Teflon (DuPont) , PTFE)allow a vacuum pump WSA FlowLite Turbo) to sample air (2 Wmin) directly across the surface of the chemical sensor. For those applications in which a hydrogel substrate was utilized for immobilization of the chemiluminescent reagents, the gel was sandwiched between the glass window and a Teflon PTFE membrane (0.47 pm, Gelman Sciences). The purpose of the Teflon membrane is to permit the diffusion of analytes from the airstream into the gel while preventing the loss of water from the hydrogel. A Keithley 485 picoammeter was used to monitor the cathodic current, and National Instruments' LABView for Windows controlled the data acquisition. Chemicals and Gas Mixtures. All chemicals were used as received from the suppliers. The following is a list of the (19) Stigbrand, M.; Pontbn, E.; Irgum, K. Anal. Chem. 1994,66, 1766-1770. (20) Rose-Pehrsson, S. L.; Grate, J. W.; Ballantine, D. S.; Jurs, P. C. Anal. Chem. 1988,60,2801-2811.
Analytical Chemistry, Vol. 67, No. 13, July 1, 1995
2225
polymeric sorbent coatings and hydrogels examined, the abbreviation used, and the corresponding manufacturer name: polyethylenimine (PEI, Phase Separations);fluoropolyol (FPOL, synthesized in-house,21with the structure [OCHZCH(OH)CHZOC(CF3)2(CsH4)C(CF3)20CH~CH(OH) CHzOR-l,, where n = 5- 10 and R = CH~(CFZ)~CH~ for FPOL1, R = (CF3)zCCHCHCH for FPOU, and R = (CHZ)~ for FPOU; FOMBLIN 2-DOL, perfluoropolyether diol (Aldrich); polyacrylamide (PAC, synthesizedfrom N,M-methylenebis(acrylamide) and acrylamide (Aldrich) according to a previously published procedure); 4%poly(viny1 alcohol) @VAL,Aldrich, 99%hydrolyzed) cross-linked with sodium borate; poly(ethy1ene oxide) (PEO, Union Carbide Corp.); SuperAbsorb ing Polymer (SAP, Absorbent Technology Co.); WaterLock SuperAbsorbent polymer D-242 (Grain Processing Corp.). The 3-Aminophthalhydrazide (luminol) (97%) was purchased from Aldrich and used to prepare a 0.01 M stock solution in 0.1 M KOH. The metal salts used for the catalytic studies include Fe~(S04)3, CuC12, MnS04, CoC12, &Fe(CN),j, Na~S03,colloidal platinum prepared by reducing hydrogen hexachloroplatinate(n? with ascorbic and colloidal platinum prepared by reducing PtC4 with sodium citrate.23 Gas mixtures (86 ppm NO2 in air, 103ppm 02 in He, 110 ppm NO in air, 99.8 ppm SO2 in air, and 450 ppm NH3 in nitrogen) were obtained from either Scott Specialty Gases or Matheson and diluted with air using Matheson 8200 mass flow controllers to give varying concentrationlevels. The hydrazine vapor generation system has been described elsewhere and was used to generate hydrazine levels from 10 to 500 ppb in Humidity was controlled through the use of bubblers and line mixers and quantitated using a Hygrodynamics hygrometer. Film Preparation. The polymeric thin films were prepared by exhaustively spraying a basic methanolic solution of the polymer, luminol, and metal catalyst onto a glass substrate (8 cm2) using an air brush (Badger Model 200.3). In general, the hydrogels were prepared by adding 1mL of M luminol (PH 13 in KOH) to 0.02 g of the resin. RESULTS AND DISCUSSION Shown in Figure 1 is the experimental apparatus utilized for
measuring the chemiluminescence signals arising from the luminol-immobilized,polymeric films. A small, hand-held,vacuum pump was used to sample the air (2 L/min) directly across the surface of the film, and a photomultiplier tube positioned behind a glass window sensitively determined the emission of photons from the chemiluminescent film. Because the sensitivity of these sensors is directly linked to the background chemiluminescence, it was critical that all light leaks through the sampling ports be eliminated. Providing that the inlet and outlet tubes are long enough (>30 in.) and sufficiently masked with black, heat-shrink tubing, the background chemiluminescence in the absence of any analytes (i.e., in an atmosphere of ultrahigh-purity nitrogen) can be reduced to the room-temperature baseline response of the photomultiplier tube (0.5 nA). The oxidation of luminol by oxygen results in the emission of a photon of light (-440 nm) according to the chemical reaction J. G.; Griffith, J. R; Reines, S. A. J. Paint Technol. 1971,43,113119. (22) Collins, G. E.; Latturner, S.; Rose-Pehrsson, S. L. Talanfa,in press. (23) Matheson, M. S.: Lee, P. C.; Meisel, D.; Pelizzetti, E.J. Phys. Chem. 1983, 87,394-399. (24) Grate, J. W.; Rose-Pehrsson, S.; Barger, W. R Langmuir 1988,4, 12931301. (21) Orear.
2226 Analytical Chemisfry, Vol. 67,No. 13, July 1, 1995
TH*
e
PH2
e
Figure 2. Chemiluminescence reaction arising from the oxidation of luminol by O*(g). Table 1. Sensitivities Obtained for the Detection of Oxygen in Nitrogen Utilizing a PolyethyienimlneFilm Incorporating Co(ll) and Varying Levels of KOH onto a alas8 Slide (8 cm2)
[KOH] on glass slide (mol/cm2) 0 2.23 1.15
10-5 10-4 2.06 x 10-4 2.97 10-4 4.49 10-4
sensitivity to 02 ) (A/ppm x
0.00 0.00 0.308 0.593 0.586 0.200
shown in F i r e 2. This reaction occurs under basic conditions and is catalyzed by numerous metal and metal cation species. The luminol-immobilized,polymeric thin films were initially characterized with respect to oxygen exposure in order to determine those substrate conditions necessary for suppressing any chemiluminescence arising from OZ@) interactions with the film. The sensitivity of these chemiluminescent chemical sensors for NOz @) is dramatically improved by lowering the background chemiluminescence evident from the oxidation of luminol by the high concentration of OZ@) in air. In addition, by suppressing the reaction of luminol with oxygen, the lifetime of these sensors is improved by maintaining the luminol and metal catalyst concentration within the polymeric film. As the next section will demonstrate, appropriate control of the polymer, relative pH, and metal catalyst can be exploited for the sensitive determination of trace levels of oxygen. Our interest in this study was to determine those conditions necessary for either completely suppressing or enhancing the OZ@) chemiluminescence reaction with luminol. The polymeric, luminol-immobilized, chemiluminescent films were prepared by using an air brush to exhaustively spray the contents of a methanolic solution containing the chemiluminescent reagents and polymer onto a small quartz slide (8 cm2). Initial studies focused on the effect of KOH on the responsivity of luminol-immobilized (6.2 x lo-* mol/cm2) polyethylenimine (6.2 x 10-3 g/cm2) films incorporating C O O as the metal catalyst (1.2 x mol/cm2). Shown in Table 1 are the sensitivities [A/ppm OZ@)in N2@)] observed with respect to the detection of oxygen for several different polymeric films, each of which was prepared identically, with the exception of the quantity of KOH deposited onto the glass slide during spraying. It is evident from this table that the optimum concentrationof KOH for the detection of oxygen is -2 x 10-4 mol/cm2, with the sensitivity dropping off significantly on either side of this concentration level withii the polymeric film. As will be discussed in more detail later, films prepared for the detection of nitrogen dioxide utilize a much lower concentration of KOH (e.g., 2 x mol/cm2) within the film, in order to adequately suppress the chemiluminescence background signal arising from the direct oxidation of luminol by oxygen. The particular metal catalyst chosen for immobilization within the polymeric film has an important impact upon the sensitivity
Table 2. Chemiluminescence Sensitivities Obtained for the Detection of Oxygen While Using Various Metal Catalysts immobilized within a PoiyethylenimineThin Film
wE-0081 h
v)
a
5
v
polymer typeb metals Fez(SO&
PEI
0.797 0.593 Mn(I1) 0.518 CUOI) 0.323 Pt(asc0rbic) 0.280 &Fe(cN)6 0.00
FPOL2 FOMBLIN FPOLl FPOU 1.13
1.11
2.22
2.61
none 0.448
that &n exhibits for oxygen. Table 2 summarizes the sensitivities obtained for a series of polyethyleniminefilms grown onto quartz slides, each incorporating identical concentrations of catalyst (1.2 x 10-8 mol/cm2), KOH (2 x mol/cm2), polymer (6.2 x loW3 g/cm2>, and luminol (6.2 x mol/cm2). It is interesting to note that the sensitivity for Oz(g) obtained from a film containii Fe(CN)$ is nearly unmeasurable, while the sensitivity seen using Fez(SOJ3 is nearly 8 x 10-13 Mppm Oz(g) in nitrogen. Ferricyanide is certainly a catalyst to consider for situations requiring a low oxygen background, while Fez(SOd3 appears to be the catalyst of choice for the detection of Oz(g). McGill et al. have investigated the sorbent properties of a number of different polymeric coatings (including FPOL and PED, demonstrating their ability to preferentially adsorb particular classes of analytes on the basis of the functionality groups incorporated within the polymeric backbone.25In fact, predictions can be made a priori concerning the type and extent of analyte adsorption into a given polymeric film by elucidating the solvation parameters for the polymer and the analyte (e.g., polarizability, dipolarity, hydrogen bond acidity, hydrogen bond basicity, etc.) . In an effort to introduce some selectivity to these chemiluminescent chemical sensors, a number of different polymer coatings were investigated and compared for the detection of oxygen (see Table 2). Each of the films was prepared with identical concentrations of polymer (2.5 x g/cm2), KOH (2 x mol/cm2), metal catalyst (Fe(IID, 1.2 x lo-* mol/cm2), and luminol (6.2 x mol/cm2), with the exception of the control case, in which no polymer was used. In each case, the presence of a polymeric sorbent coating on the glass slide for the immobilization of the chemiluminescent reagents resulted in an enhancement of the sensitivity. The sensor utilizing FPOU was found to be nearly 10 times more sensitive to oxygen. We attribute this effect to the increased chemical sorption of oxygen by the polymer on the glass slide. From the optimization studies, a chemiluminescent chemical sensor was prepared for the sensitive determination of trace levels of oxygen via the incorporation of luminol(6.2 x mol/cm2), KOH (2 x loT4mol/cm2> and Fe2(S04)3 (1.2 x mol/cm2) g/cm2). Figure 3 within a thin film of FPOL3 (2.5 x illustrates the chemiluminescent response of this sensor to increasing levels of oxygen in nitrogen. A detection limit of 2.4 ~~
:i
i
P
~
2.4 ppm
I
8 3.1 ppt
1.6
800 PP
100 ppm
Effect of the polymer sorbent coating on the chemiluminescence sensitivity obtained for the detection of oxygen (Fe$II) used as the metal catalyst). b Sensitivity given in N p p m x 10-
~
8.00E-009
6.2 P
~
~
(25) McGill, R A; Abraham, M. H.; Grate, J. W. CHEMTECH 1994.24, 2737.
O.OOE+OOO!
0
.
, 50
I
I
I
100
150
.
I
200
.
, 250
.
I
300
Time (sec)
Figure 3. Introduction of varying levels of oxygen in a nitrogen carrier stream to a thin film of fluoropolyol immobilizing luminol, Fe(Ill),and KOH. The inset is an example of the signavnoise ratio seen for the detection of 2.4 ppm oxygen in nitrogen.
ppm O2 in nitrogen (signal/noise = 2:l; see inset of Figure 3) was obtained with a response time of several seconds (r > 0.99). In a fashion similar to the films prepared for the detection of NOT (g) ,which will be discussed in the following section, the sensitivity was observed to slowly decrease (l%/h) with exposure to 20 ppm 02(g) . We attribute this gradual change in the film’s response to the continual and irreversible change in luminol concentration occurring at the surface of the polymeric thin film. The usable lifetime of these films is strongly dependent upon the concentration and extent of exposure to oxygen. This drawback will likely limit the applicability of this type of chemical sensor to the trace detection of oxygen within an inert atmosphere, e.g., monitoring low levels of oxygen during the fabrication of microsemiconductor devices. For monitoring low-ppm quantities of oxygen, we have found that, in general, the polymeric films should be replaced on a daily basis in order to maintain an accuracy of 10%. Nitrogen dioxide is a strong oxidizer whose reaction with luminol results in the emission of a photon of light. Initialattempts at preparing a chemical sensor for NOz(g) relied upon the same polymeric sorbent coatings and conditionsinvestigated for oxygen, with the exception that a much lower concentration of KOH (2 x mol/cm2) was employed in order to sigdlcantly lower the background chemiluminescence evident from the oxidation of luminol by OzCg). These polymeric films exhibited a strong humidity dependence for the detection of NO&). Figure 4 is a plot of the plateau chemiluminescenceresponse observed for the introduction of 1ppm NOz(g) in air to two different films, FPOU and PEI, while varying the relative humidity of the airstream from 0 to 100%. Both of these films immobilized identical concentrations of colloidal platinum (1.2 x lo-* mol/cm2) and luminol (6.2 x mol/cm2). Under dry conditions, no chemiluminescence response was seen for either film to the introduction of NOz@). As the humidity level was increased, the two polymeric matrices exhibited dynamically different behavior. For the FPOU film, the chemiluminescence response increased steadily with increasing relative humidity. The PEI film, on the other hand, showed a maximum in chemiluminescence at -20% relative humidity, which decreased with increasing humidity. It is interesting to note that the peak chemiluminescence intensities observed were identical for both films, albeit at two different relative humidities. Analytical Chemistty, Vol. 67, No. 13, July 1, 1995
2227
I.
2.5~10~1
g
h
v)
v
2.ox10-8
..
Table 3. Compll8tlon of the Sensltlvltles Observed for the Detectlon of NO&) and Other Possible Interfennts While Uslng Various Hydrogels for lmmoblllzlng Lumlnol, KOH, and Colloidal Platinum (Cltrate Reduced)
V
.
sensitivity (A/ppm x 10-9) polymer
SAP PEO PAC Waterlock Waterlocka 0
20
40
60
80
100
120
Relative Humidity (%) Figure 4. Effect of humidity on the chemiluminescence response of polyethylenimine and fluoropolyol thin films (immobilizing luminol and colloidal platinum) to 1.O ppm NOp.
These results can be better understood by examining the solubility characteristics of these polymers with respect to the adsorption of water, a property that has been previously quantitated using surface acoustic wave devices. Polyethylenimine is a polymer with hydrogen bond accepting functionalities that interact specifically with water. Fluoropolyol, on the other hand, is considered to be a hydrogen bond acid due to the fluorinated carbon atoms found along its polymer backbone and, as a result, adsorbs several times less water vapor. From Figure 4, it is evident that water absorption into these two different polymers plays a critical role in the chemiluminescent reaction scheme involving NO&). One explanation may be that the solvating properties of water permit the formation of free OH- ions from KOH crystallites contained withii the polymeric film. This effect is most clearly seen for the FPOL film, which exhibits a nearly linear increase in its chemiluminescenceresponse to 1ppm NOT (g) with the increase in relative humidity. The PEI film, on the other hand, strongly adsorbs water vapor into the film even at much lower relative humidities and, for this reason, exhibits a peak in the chemiluminescence at a much lower level (20%versus nearly 100%).As the relative humidity exceeds 20%,the overall Chemiluminescencesignal to 1 ppm NO2 (g) decreases due to the competition the film expresses for the chemisorption of water vapor, as opposed to NOz(g). From a sensor standpoint, the strong intluence of humidity on the sensitivity of these chemiluminescent films limits the usefulness of these devices. In order to eliminate this effect, the chemiluminescent reagents were immobilized withii polymeric hydrogels sandwiched between a quartz glass window and a Teflon PTFE membrane. The membrane serves two purposes: (1) prevents the rapid evaporation of water from the hydrogel and (2) permits the diffusion of analytes across the membrane into the hydrogel. The immobilization of luminol and a metal catalyst within a hydrogel provides a suitable matrix for supporting the chemiluminescent reaction in either dry or wet air with identical sensitivity. Shown in Table 3 is a compilation of the sensitivities observed in the quantitation of NO&) and a number of other gases of analytical interest, for a series of hydrogels immobilizing the chemiluminescent reagents. Although fh-to-film reproducibility was not extensively investigated for the purposes of this report, initial studies have indicated that for a given set of three films 2228 Analytical Chemistty, Vol. 67,No. 13, July 1, 1995
PEI (20% RH) FPOL (88%RH)
NOz@ 76.5 52.6 4.6 186 389 48.5 806 11.2 21.2
NZH4 k) SOZGZ)
NO@
0.7 0.6 0.00
0.03 0.00 0.00
0.3
0.3
4.5 0.2
0.03 0.05 0.03 0.00 0.00
0.00 0.00 0.00 0.00 0.00 0.00
no data 0.06 0.00
NH3 @)
no data 0.00
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
0.00
0.00
0.01 M Na~S03.
Tabk 4. Sensltlvltles Observed for the Detectlon of NO&) and Other Porslble Interfennts Whlle Using a Series of Transltlon Metal Catalysts lmmobllized within a pH 13, Poly(vlny1alcohol) Film
sensitivity (A/ppm x 10-9)
metals Fez (SO413 &Fe(CN)G CUOO Pt(citrate) Pt(ascorbic) COOO
NO&) 110 76.3 95.0 48.5 44.2 42.9
NZH4k)
SOzk)
NO&)
NH&)
2.2 1.6 66.2 0.2 2.0 20.9
0.08 0.06 0.01 0.05 0.01 0.6
0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00 0.00
prepared under identical conditions, the h - b f hreproducibility is within 6-8%. We believe the reproducibilitycould be improved through careful control of the film thickness and placement withii the sensor compartment cell. A detailed description of each hydrogel and the immobilization procedure utilized is given in the Experimental Section. Also listed for comparison are the sensitivities of two polymeric coatings, PEI and FPOL2, at their optimum relative humidity. Note first that the detection of NOz k) using PEI and WOL2 was inferior to the sensitivitiesobserved for the majority of hydrogels examined. Just as was seen for the case of oxygen, the detection of NOz(g) is strongly dependent upon the polymeric composition of the hydrogel used to support the chemiluminescent reagents. Of the different hydrogels examined, the water-soluble resin Waterlock provided the highest sensitivity with respect to the detection of NO&). Wendel et al. have demonstrated that the addition of NaZS03 to a solution of luminol and metal catalyst will result in a significant enhancement of the chemiluminescencesignal seen for the detection of nitrogen dioxide. Our results support this observation for luminol-immobilized hydrogels, wherein the addition of sodium sulfite to an otherwise identical Waterlock hydrogel more than doubled its sensitivity to NO2 (g). In Table 4, a series of sensitivities are tabulated for the hydrogel, poly(viny1alcohol),immobilizing luminol(5 x moll L) and a series of different metal catalysts mol/L), all prepared under identical conditions. Whiie Fez(S04)3seemingly provided the best sensitivity to NO&), the chemiluminescence response obtained using this metal catalyst was unstable, with the plateau response slowly decreasing with time. For this reason,
1 . 8 ~ 1 -0 ~
4.5 ppm
2.0~10~
E 8E m
-$'
8.0~10~
I'\ppm
3.6 ppm
1.6~10~1.4~10'-
v
m
1.2~10~-
2.3 ppm
1.0x10-~-
3
8.0x108-
.-$ .-
8.OX10"
1.4 ppm
c
2
4.0~10" 2.ox10~
0.0 I
9
'
I
10
.
m
.
I
'
11
I
12
'
I
13
.
1
14
PH Flgure S. Influence of the pH of a Waterlock hydrogel on the sensitivity obtained for the detection of NOn(g) (Cu(ll) used as the metal catalyst).
subsequent studies have emphasized the use of the metal catalysts colloidal platinum (citrate reduced) and CugD for the quantitation of NOzCg). Just as was seen in the case of oxygen detection, NOz(g) determinations using these chemiluminescent chemical sensors are strongly dependent upon the relative pH of the hydrogel. For the chemiluminescent detection of oxygen, extremely alkaline conditions were necessary on the surface of the polymer film.The optimal pH for the detection of NO&), on the other hand, occurs in a pH region where the background chemiluminescence from oxygen is minimal. Figure 5 is a plot of the change in sensitivity with pH for a Waterlock polymer incorporating luminol and Cu(11)in the detection of NO&). The most striking observation to be made from this figure is the sharp onset of chemiluminescence intensity above pH 11.5. The importance of this result is that some selectivity can be tailored into these devices by simply adjusting the relative pH of the hydrogel being utilized. For example, we have been examining methods for the chemiluminescence detection of chlorinated hydrocarbons that utilize a preoxidation step with a heated Pt filament.26This chemistry is operable at pH 11 with little loss in sensitivity for the chlorinated hydrocarbons, but with a dramatic reduction in the interference effect of NOz(g). The sensitive detection of part-per-trillion levels of nitrogen dioxide in air was demonstrated in a Waterlock hydrogel incorporating luminol and the metal catalyst colloidal Pt (citrate reduced) (see Figure 6). The downward spikes evident in Figure 6 are artifacts associated with the solenoid valve within the gas flow controller adjusting to the lower flow requirement for the analyte gas. Response times were on the order of seconds, with a detection limit as low as 460 ppt (signal/noise = 3:l) and error bars for point-to-point reproducibility off%. While the linearity of response was excellent (r > 0.999), the sensitivity was observed to slowly increase or decrease depending upon the following: (1) the speciiic metal catalyst immobilized within the hydrogel and (2) the concentration, exposure time, and nature of the oxidant being examined. For example, when the Cu2+/luminol/Waterlok gel was exposed to a constant concentration of NO&) at 20 ppm, the chemiluminescence signal slowly increased at a rate of -l%/ h. This change in sensitivity is likely attributed to a change in either the catalyst or luminol concentration within the gel. The (26) Collins, G. E.; Rose-Pehrsson, S. L., document in preparation.
0.0
I 0
'
,
I
100
I
200
I
300
Time (sec)
Figure 6. Response of a poly(viny1alcohol) film immobilizing luminol and colloidal platinum (citrate reduced) to the introduction of varying levels of NOz(g) in a carrier stream of dry air.
lifetime of these chemiluminescent gels is directly related to the type and concentration of the oxidant and the extent of exposure. In order to maintain accuracy (lo%), we have found that, in general, the chemiluminescent gels should be replaced with fresh substrates on a daily basis. Tables 3 and 4 also tabulate the sensitivities exhibited by each of the different hydrogels with respect to the detection of several other chemical oxidants and reductants: hydrazine, sulfur dioxide, nitrous oxide and ammonia. Each of the different film compositions was completely unresponsive to NO (g) and NH3 (g) . We can attribute this result to the weak, oxidative properties of these molecules with respect to the chemiluminescent oxidation of luminol. Hydrazine and sulfur dioxide, on the other hand, were each detected to varying degrees, although nowhere near the sensitivity shown by nitrogen dioxide. Fortunately with respect to the selective detection of nitrogen dioxide, immobilization of the metal catalyst, Cu(II), resulted in the maximal suppression of any interference due to sulfur dioxide (no response seen for SOT (g) concentrations of < 3 ppm). Optimal sensitivity for sulfur dioxide was observed using the Waterlock hydrogel to immobilize the metal catalyst, Co(II), and luminol. The detection limit for sulfur dioxide under these film conditions was -37 ppb. The time response for sulfur dioxide is comparable to that seen for nitrogen dioxide, with 90%of full scale observed within 1min. Hydrazine is an extremely strong reducing agent whose toxic character has generated considerable interest in the development of an effective chemical sensor for this molecule. One would not expect a chemiluminescent reaction to take place between hydrazine and luminol because of the reducing character of hydrazine. Previous investigations in our laboratory with the metal catalyst colloidal platinum have suggested that hydrazine is catalytically oxidized at the surface of metal catalysts to form a radical peroxo intermediate that oxidizes luminol. The detection of hydrazine has been demonstrated to levels as low as 10 ppb in air (signal/ noise = 3:l). The response obtained from this sensor for the detection of 125 and 250 ppb hydrazine in dry air is shown in Figure 7. The detection of NO2 operated on a signiicantly faster time scale (several seconds) when compared to that obtained while monitoring hydrazine levels (10 min). The slow time response evident for the detection of hydrazine is likely attributed to the indirect reaction mechanism involved. In order to be Analytical Chemistry, Vol. 67, No. 13,Ju/y 1, 1995
2229
h
ln
Q
E
v
.-L.
9.0~10~
1
8.0~10'~
tn
8
7.0~10~
Time (sec) Figure 7. Example of the time response obtained for the detection of hydrazine using a Waterlock polymer immobilizing luminol, KOH,
and colloidal platinum (ascorbic acid reduced). detected, hydrazine vapor must dissolve in the hydrogel film and catalytically react at the surface of a platinum colloid to form a radical peroxo intermediate that can subsequently react with luminol to generate a photon of light. Nitrogen dioxide and oxygen, on the other hand, react with luminol directly to give a much more rapid chemiluminescent response. The timedomain signal acquired for any particular analyte can be an additional tool for introducing selectivity into these chemical sensors. Kolesar, for example, has demonstrated the ability to selectively differentiate between NO&) and diisopropyl methylphosphonate @IMP) using the time-and frequency-domain responses of a copper phthalocyanine-coated IGFET?7 CONCLUSIONS Chemiluminescencetechniques benefit from superior sensitivity and ease of measurement, two features that are attractive to the chemical sensor arena. We are currently pursuing efforts to immobilize these chemiluminescent gels directly onto the surface of a miniature photomultiplier tube (8 mm diameter, Hamamatsu, Inc.) in order to generate a new class of portable, chemiluminescent chemical sensors. There are a number of advantages associated with the successful incorporation of this chemistry into a miniaturized format, and they include its simplicity, low power (27) Kolesar, E. S. Proceedings Sensor Expo; Helmers Publishing, Inc.: Peterborough, NH, 1994; pp 11-30.
2230 Analytical Chemistty, Vol. 67,No. 13, July 1, 1995
requirements (+15 V), lack of liquid waste generation (absence of a peristaltic pump-a feature common to most chemiluminescence techniques), and possibility for adaptation to remote sensing and personal dosimeter applications. The primary drawbacks to the system described here are selectivity and stability. The selectivity can be improved by using an array of chemiluminescence chemical sensors employing different chemiluminescent reagents, pH, metal catalysts, and polymer substrates. We have demonstrated that simple changes in these parameters will signilicantly alter the sensitivity and selectivity of these films with respect to the detection of oxygen and nitrogen dioxide. The gradual shift in sensitivity observed in these sensors following continuous exposure to an oxidant (l%/h for 20 ppm NO&)) is a feature associated with the irreversible character of the luminol chemiluminescencereaction, changes evident in the gel composition due to the slow evaporation of water through the Teflon membrane, or changes in pH arising from the absorption of COz(g) into the film. Although it may be feasible to automatically adjust the calibration curve according to the integrated number of photons collected at the detector, the gradual change in sensitivity apparent in these films may ultimately limit the quantitative usefulness of these devices to either trace detection or alarm monitoring. While chemiluminescence systems that utilize continuously regenerated reagent solution have fewer problems with drift, these systems are neither compact nor suitable for remote or portable applications. The system described here sacrifices some with respect to quantitative accuracy and sensitivity, in deference to simplicity, portability, and remote sensing capability. We are currently pursuing the development of a portable, remote sensor alarm system whose daily maintenance consists of simply inserting a plug of chemiluminescence gel into the compartment cell, ensuring excellent sensitivity and reasonable quantitative accuracy over the course of a day of continuous monitoring. ACKNOWLEDGMENT This research was supported by NWKennedy Space Center (DLE23-24, CC-823604. Received for review January 19, 1995. Accepted March
28, 1995.@ AC950065Y @Abstractpublished in Advance ACS Abstracts, May 15, 1995.
