2780
Anal. Chem. 1087, 59, 2780-2785
1-100 ppm level and 6-9% below 1 ppm. The detection limit, estimated from the concentration that yields LC-EC peak current of twice the uncertaintly of the background blank current, is ca. 4 X lo4 M or about 2-3 ppb Fe(1II). No attempt has been made in the present study to optimize the detector's dead volume and the separation/ elution hydrodynamic conditions. It is noteworthy that thoron-loaded resin proves to be very useful for the iron preconcentration and its determination in natural, namely, river, lake, or tap, waters (32). For example, preconcentration from a 250-mL water sample required only ca. 0.20 g of the resin bed and 5 mL of the HC1 solution for elution. Coupling of the chromatographic system with the electrochemical detector allowed standard, quantitative preconcentrations/determinations of Fe(II1) at levels as low 0.0001-0.005 ppm to be carried out. Metal cations present in samples of Fe(1II) are not retained by the resin in column and therefore cannot interfere with the separation; further most of them would not be even electroactive at 0.20 V on a GC/Ni-CN-Fe electrode. On the same basis, the common anions, e.g., NO;, Br-, NOz-, I-, F-, C104-, CH,COO-, Br03-, MOO^^-, and Cr042-produced no interference chromatographically and/or electrochemically. Possibility of oxygen interference was already ruled out in a previous section. Registry No. Fe, 7439-89-6; thoron, 132-33-2;carbon, 744044-0.
LITERATURE CITED (1) Pohl, C. A.; Johnson, E. L. J . Chromafogr. Scl. 1080, 18, 442-452. (2) Fritz, J. S.;Gjerde, D. T.; Becker, R. M Anal. Chem. 1080, 52, 15 19-1521. (3) Svenlch. 0. T.; Fritz, J. S. Anal. Chem. 1883,55, 12-15. (4) Ward. J. E. Anal. Chem. 1078,51, 838-838. (5) Cassldy, R. M.; Eichuk, S. J . Chromatogr. Scl. 1080, 18, 217-219.
(6) Elchuk, S.;CassMy, R. M. Anal. Chem. 1078,5 1 , 1434-1436. (7) CassMy, R. M.; Elchuk, S. J . Liq. Chromatogr. 1081, 4 , 379-382. (8) Haddad, P. R.; Alexander, P. W.; Trojanowlcz, M. J . Chromafogr. 1084,294,397-402. (9) Kissinger, P. T. Anal. Chem. 1077,49, 447A. (IO)Ruckl, R. J. Talents 1080,27. 147-156. (11) Kissinger. P. T. in Laboratory Techniques In Electroanalytical Chemisfry; Kissinger, P. T., Heineman, W. R., Eds.; Marcel Dekker: New York, 1984;Chapter 22; pp 611-835. (12) Bond, A. M.; Wallace, G. G. Anal. Chem. 1082,5 4 , 1706-1712. (13) Johnson, D. C.; Larochelle, J. H. Talenta 1073,20, 959-971. (14)Hanekamp, H. 6.; Voogt, W. H.; Bos, P.; Frei, R. W. Anal. Chim. Acta 1080. 118, 81-86. (15) Stulik, K.; Pacakova, V. J . Elecfroanal. Chem. 1981, 129, 1-24. (16) Cox, J. A.; Kulesza, P. J. Anal. Chem. 1984,56, 1021-1025. (17) Larochelle, J. H.; Johnson, D. C. Anal. Chem. 1078, 5 0 , 240-243. (18) Stutts, K. J.; Wightman, R. M. Anal. Chem. 1083, 55, 1576-1579. (19) Ravichandran. K.; BaMwin, R. P. Anal. Chem. 1083,55, 1586-1591. (20) Guadaiupe, A. R.; Abruna, H. D. Anal. Chem. 1085, 57, 142-149. (21) Neff. V. D. J . Electrochem. SOC. 1078, 125, 886-887. (22) Ellis, D.; Eckhoff, M.; Neff, V. D. J . Phys. Chem. 1981, 85, 1225-1231. (23) Rajan, K. P.; Neff, V. D. J . Phys. Chem. 1082,86, 4361-4368. (24) Itaya, K.; Ataka, T.; Toshima, S. J . Am. Chem. SOC. 1082, 104, 3751-3752, 4767-4772. (25) Itaya, K.; Akahoshi, H.; Toshima, S. J . Electrochem. SOC.1082, 129, 1498-1500. (26) Bocarsly, A. 8.; Slnha, S. J . Electroanal. Chem. 1082, 137, 157-162. (27) Bocarsly, A. 8. UNESCOINSF Workshop on Photoelectrochemical Processes and Modified Electrodes, Santa Cruz, CA, 1984. (28) Brajter, K. J . Chromafogr. 1074, 102, 385-387. (29) Tanaka, H.; Chlkuma, M.; Harada, A.; Ueda, T.; Yube, S. Taianfa 1078,23,469-494. (30) Brajter, K.; Dabek-Zlotorrynska, E. Talents 1980,27, 19-23. (31) Nakayama, M.; Tanaka, H. Talents 1084. 3 1 , 269-271. (32)Brajter, K.; Dabek-Zlotorzynska, E. Microchlm. Acta 1085, 1 1 , 179-186. (33) Marecek, V.; Samec, 2.; Weber, J. J . Elecfroanal. Chem. 1978,9 4 , 169-176. (34) Murray, R. W. Elecfroanalytlcal Chemistry; Bard, A. J., Ed.; Marcel Dekker: New York, 1984;Vol. 13.
RECEIVED for review January 27, 1987. Accepted June 18, 1987.
Determination of Oxygen Concentrations by Luminescence Quenching of a Polymer-Immobilized Transition-Metal Complex J. R. Bacon* Chemistry Department, Western Carolina University, Cullowhee, North Carolina 28723 J. N. Demas* Chemistry Department, University of Virginia, Charlottesville, Virginia 22901 Oxygen quenching of the luminescence of the tris(4,7-dip h e n y l - l , l ~ ) r u t h e n i u m ( I I )perchlorate hwnoblized in a silicone rubber is shown to be an accurate and precise method for measurlng oxygen concentrations in solutions and In the gas phase. Quenching can be quantitated by either lifetime or intensity quenching measurements. Aqueous strong adds, bases, compkxing agents, oxldants, and reductants do not penetrate the hydrophobic polymer and, therefore, do not affect the response. Gaseous interferents, such as H#, andhetk gatms (e.g., N20,halothane), and fluorocarbons do not affect the response. Chlorine and erpeclaUy SO2are strong, but fully reversbb, interterents. A system was devdoped wlth a response t h of less than 0.2 8, whlch Is adequate for the monltorlng ol breathing subJects.
The determination of oxygen concentrations in gaseous samples, aqueous samples, and biological fluids has important ramifications in medicinal, environmental, and analytical
chemistry. Today most oxygen measurements are based on modifications to the Clark electrode (I),although the Winkler titration is also widely used (2). The Clark electrode is easily calibrated, is relatively rapid in response, and requires relatively inexpensive instrumentation. However, Clark electrodes consume oxygen and are easily poisoned by HzS, proteins, and various organic compounds. In operating room use, they register spuriously large oxygen concentrations in the presence of certain anesthetics, which is a potentially fatal shortcoming (3, 4). The Winkler titration is slow and cumbersome. Further, since it is based on oxidation-reduction chemistry, interferences are numerous. The luminescence intensities of a variety of organic species are quenched (deactivated) by oxygen (5-8). The degree of oxygen quenching is dependent on the oxygen concentration and has formed the basis of an elegant method of oxygen analysis (5, 9-14). The paper by Peterson (5) describes a fiber-optic system. These authors used a silica gel bound luminescent dye separated from the solution being measured by a porous mem-
0003-2700/87/0359-278080 1.50/0 0 1987 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 59, NO. 23, DECEMBER 1, 1987
brane. Oxygen diffusion through the membrane affected the dye within the fiber. The system had limited stability and slow response and suffered severe interference from the anesthetic halothane. An elegant triple fiber optic sensor (10) based entirely on luminescence has been developed for simultaneously determining COz and O2 concentrations and pH. For improved signal to noise ratio, the system uses a ratiometric system based on a pulsed flashlamp. Organic dye sensors are used, and the system works well in vivo. The oxygen sensor dye and COz evaluation algorithms are proprietary. Exploiting the sensitivity of some dyes to quenching by halothane, a system has been developed for quantitating both halothane and oxygen (11). Dual sensor dyes are used. Covalent attachment of the sensor dyes or microencapsulation have also been used to minimize leaching effeds and to isolate the dye from the environment (12,14). Other sensor systems have included cyclodextrin-enhanced, room-temperature, phosphorescent bromonaphthalenes (13). Also, a pyrene-based excimer system has been utilized that provides a dual emission that can be used in a ratiometric mode to eliminate source fluctuations (13). The goals of the current work were to prepare new, stable, luminescence-quenched systems for oxygen analyses. Further, we wished to have a system that was amenable to measurement by either intensity or lifetime quenching. Lifetime systems avoid the need for ratiometric measurements. Also, we wished for the system to be largely immune to interference from important gaseous or solution interferents. Luminescence quenching methods of analysis are based on decreases in the emission intensity or luminescence lifetimes in the presence of the quencher. If luminescence quenching is entirely diffusional, the excited-state lifetimes or luminescence intensities are related to the quencher concentration by the Stern-Volmer equations
Ksv = 4 7 0 where 7's and I's are luminescence lifetimes and intensities, respectively. The subscript "0" denotes the value in the absence of quencher. Ksv is the Stern-Volmer quenching constant, and k z is the bimolecular quenching constant. Plots of 70/7or Io/I versus [Q] will be linear with identical slopes of Ksv. In principle, either Stern-Volmer equation could be used to analytically determine concentrations of a variety of quenchers. In practice, an excited state that is susceptible to quenching by one species is generally susceptible to quenching by a variety of materials, and the number of interferents is large enough to make solution luminescence quenching nearly unusable. In order for luminescence quenching to be a viable analytical method, potential interferent quenchers must be excluded from the sensor while rapid penetration and equilibration of the quencher to which the sensor must respond were allowed. Demas and co-workers have shown that a variety of luminescent Ru(II), Os(II), and Ir(II1) complexes are very susceptible to oxygen quenching in solution (15). Further, they have found that the complexes are quenched by oxygen to yield singlet oxygen both in homogeneous media and when immobilized on polymers (15,16).Such systems show promise as luminescence quantum counters (17) or as singlet oxygen generators (15,16) for synthesis. We have examined these complexes to determine whether, when immobilized in an appropriate substrate, they would be useful as monitors of oxygen concentration by means of luminescence quenching.
2781
We report here a detailed study on the use of a luminescence-quenched oxygen sensor based on a luminescent metal complex, tris(4,7-diphenyl- 1,lO-phenanthroline)ruthenium(II) perchlorate, [Ru(PhZphen),](C104)2,immobilized in a solvent impermeable-gas permeable polymer film. The degree of quenching of the excited complex was related to the partial pressure of O2in contact with the film. A variety of potential interferents were examined, and the limitations of the sensor are discussed.
EXPERIMENTAL SECTION Chemicals. The silicone rubber RTV-118 was obtained from General Electric Co., Inc. [Ru(Phzphen),](C104)2was prepared and purified as described earlier (18). Instrumentation. Excited-state lifetime measurements were made on a microcomputerized apparatus (19). Samples were excited at 337 nm with a nitrogen laser. Luminescence intensity measurements were made with a microcomputerized SLM Model 8000 spectrofluorimeter (450-nm excitation). Lifetime and intensity measurements were made from the front surface because the sample thinness precluded the customary 90" viewing. The emission intensity was monitored at the uncorrected emission maximum of 610 nm. Absorption spectra were measured by using a Varian Model 634 spectrophotometer. Homemade manometers were used in measurements involving partial pressures. All measurements were made at room temperature (23 f 3 "C). Polymer Films. A number of polymer films were examined. These included the silicone rubbers GE 361, GE RTV-118, and GE RTV-615 and a hardware store variety of Dow-Corning. We also tried Plexiglas, poly(viny1 chloride), polystyrene, and polycarbonate. All data reported here are for the silicone rubber RTV-118, which is a one-part polymer similar to clear bathtub sealer. The cured polymer does not stick to Plexiglas, and this characteristic was exploited to form thin films with good optical quality. The uncured polymer was clamped between two Plexiglas plates. Teflon (0.01 in.), shim stock (0.004 in.), or aluminum foil spacers (0.001 in.) were used to form uniform films of different thicknesses. The curing in water took several days since exposure was limited to the edges. The cured films were washed twice in CHzClzprior to use in subsequent experiments. Film Doping. In order to impregnate the films with the sensor, we exploited the fact that CH2ClZdoes not dissolve the cured f h but does penetrate and greatly expand them (twofold). In a typical experiment, the dry films were immersed in 1 X M [Ru(Phzphen),](C104)2in CHZCl2.The concentration in the fiim was controlled by adjusting the concentration of the complex. The films rapidly swelled and took up the complex. In 10 min the films were removed, rinsed rapidly in CHzC12solvent to remove any surface contamination, and air-dried. This preparation procedure yielded very rapid solvent evaporation and caused film curling and water condensation on the surface. This degraded the optical quality of the films; there were obvious visual nonuniformities in the Ru(I1) concentration. We have subsequently found that this nonuniformity could be avoided by slow evaporation of the CHzCl2in a jar covered by a loose-fitting cardboard lid and containing a small amount of CH2C12. Film uniformity was improved to the point that no visual variations in the sensor concentration were detectable. Regardless of the method of sample preparation,all doped or undoped GE RTV-118 films were slightly cloudy. Stern-Volmer Calibration Curves. The formed silicone sensor films stick tenaciously to glass surfaces. In the quenching measurements, doped polymer films were placed on the insides of small square bottles, and the bottles were connected to a vacuum pump and manometer. The luminescence lifetimes and luminescence intensity versus partial pressure were measured in two separate experiments. In addition to Stern-Volmer data, experiments were performed to determine hysteresis and reproducibility. We determined the variation of the sensor response depending on whether the sensor was in a gas stream or an aqueous solution. This was done by comparing the lifetime of sensor films in an evacuated, air- or oxygen-filledbottle with that of the same sensor film in a water-filled bottle that was bubbled with a vigorous
2782
ANALYTICAL CHEMISTRY, VOL. 59, NO. 23, DECEMBER 1, 1987 Gos Inlet
0 Cisposable Plost i c
Figure 1. Diagram of breathing cell. Gas flows in through the tube at the top of the cell.
stream of nitrogen, air, or oxygen. Interferences. A variety of solutions are potentially capable of quenching the luminescence, destroying the complex, or extracting it from the film. Interference studies were carried out in two ways. First, to monitor leaching or destruction of the complex, we used a spectrophotometric method. The absorbance at the absorption maximum was measured before and after a 24-h exposure to the selected solution. Interferents included a surfactant, a strong oxidant, a reducing agent, a strong acid, a strong base, a strong complexing agent, and organic solvents. In the second set of studies the films were exposed to one excellent aqueous solution quencher and several gases of physiological or chemical importance. Interference was monitored by changes in luminescence lifetime or intensity on introduction of the interferent to the evacuated or N2-purged sample cell. The ionic quencher was Fe3+. The gases were the anesthetics NzO, cyclopropane, and halothane, the refrigerant Genetron 118, the sterilent Clz from commercial bleach solutions, and the industrial pollutants SO2 and H2S. Because of the possible interference of back diffusion of pump oil and because of the interest in petroleum products, we also immersed the sensor in vacuum pump oil. In each case where quenching occurred, the sensor was tested for recovery by reevacuating or repurging the cell. Sensor Response Time. A 0.004-in. film stuck to a glass wall was tested for response time by the lifetime method. The cell was initially evacuated, and the lifetime was determined. The sample was then abruptly opened to the air, and the lifetime was measured after fixed delay times. The experiment was repeated to obtain a lifetime versus time profile. The uncertainty in the time display was less than 1 s. The primitive breath monitoring cell of Figure 1 was constructed in order to measure rapidly changing concentrations, such as might occur in a patient's breath. To enhance response time, a 0.001 in. film was mounted so that both sides of the film were exposed to the gas flow. The spectrofluorimeter was set to its fastest integration time (0.02 9). Two gas-flow configurations were used. A two-way stopcock was set up so that rapid flows of either N2 or air could be switched to flow over the sensor f i b . Intensity measurements were started, and the gas flow was switched. This process was repeated several times. In the breathing experiments the T was disconnedd and one of us breathed through a 3/s in. wide, 2 ft rubber tube while the spectzofluorimeter recorded luminescence intensities as a function of time. In one experiment the breath was held for about 17 s and then normal breathing resumed. Temperature Effects. Due to a lack of suitable equipment, we were only able to measure temperature effects on the excited-state lifetime rather than on the quenching constant. These studies were done by filling the cell with cooled or heated deoxygenated water and quickly making the lifetime measurements.
RESULTS AND DISCUSSION General Considerations. Many Ru(1I) complexes with a-diimine ligands (e.g. [ R u ( b p ~ ) ~ ][R~(Phzphen)~]~+) ~+, exhibit single, charge-transfer luminescences, high photochemical stability, and rather long-lived, excited-state decays (microseconds). [R~(Phzphen)~]~+ was selected as the sensor material for several reasons. Its absorption spectrum peaks a t 460 nm
and extends to beyond 500 nm, which gives flexibility in the excitation region. Its molar absorptivity at the visible maximum is 30000 and is relatively broad, which permits low concentrations to be used and further decreases the stringency of the excitation wavelength requirements. Its emission (>600 nm) is well-separated from the absorption, which eliminates reabsorption problems and minimizes spectral filtering difficulties. The emission spectrum is also a good match for the sensitivity curves of silicon photodiodes. Further, in aqueous solutions the unquenched photon yield 0.5 (20). Given the similar of [ R ~ ( P h ~ p h e n )approaches ~]~+ solution and polymer-immobilized lifetimes, the luminescence yield in the RTV-118 is probably very nearly the same. The high luminescence efficiency in the sensor reduces its sensitivity to scattering interferences; this permits the use of less sensitive detectors. Finally, of all the Ru(I1) complexes examined, [ R ~ ( P h ~ p h e n )also ~ ] ~has + the longest excited-state lifetime (>5 ps) and the largest solution Stern-Volmer quenching constant (15). The high photochemical stability of the Ru(I1) complex is in contrast to most phosphorescent organic luminophores (21, 22). The single emission of Ru(I1) complexes is also in marked contrast to phosphorescent organic molecules, which invariably exhibit some degree of fluorescence and phosphorescence, each with a very different oxygen-quenching sensitivity. Thus, with many organic luminophores the total emission and decay curves are superpositions of two components of different lifetimes. This greatly complicates data acquisition and reduces the accuracy of concentration measurements based on intensity or lifetime quenching. RTV-118 was selected as the polymer support after a variety of systems were examined. Several had better optical properties or were easier to form (two-part polymers), but the oxygen Stern-Volmer constant was greater for our complex in the RTV-118 than in any other polymer that we tested. These results are consistent with the high sensitivity of silicone rubbers used with fluorescent organic dye sensors (10,11,23). Further, some were difficult or impossible to dope or fabricate into desired shapes (e.g., PVC, polystyrene, polycarbonate), were soluble in common solvents (e.g. polystyrene, Plexiglas), or had lower stability. Temperature Effects. The unquenched excited-state lifetimes at different temperatures were 5.8 (0 "C), 5.9 (25 "C), 4.8 (38 "C), and 3.3 1.18 (60 "C). These results are consistent with the well-known temperature effects on the solution luminescence properties of Ru(I1) complexes (21,22). With a simple temperature correction scheme, the quenched sensor should be quite useful up to the physiological temperature of 37 "C. Higher temperatures would result in a significant drop in lifetime and yield. Quenching Curves. The lifetime in deaerated, aerated, and oxygenated water were the same (less than 2-3%) as that obtained for a vacuum, in air, and in oxygen, respectively. This would be expected if water vapor were not a quencher and the sensor only responded to the partial pressure of the oxygen. Figures 2 and 3 show intensity and lifetime Stern-Volmer quenching plots for the RTV system. Unlike solution Stern-Volmer measurements, the Stern-Volmer calibration c w e s for the polymer f i i s always exhibited some downward curvature. The film calibration plots were, however, highly reproducible, which is the primary consideration for an analytical technique. In spite of the nonlinear Stern-Volmer plots, intensity and lifetime measurements gave calibration curves that were indistinguishable (Figure 2). The sensors exhibited no hysteresis as shown in Figure 3, where the data was taken as the partial pressure of O2 decreased ( X ) and then increased (+).
ANALYTICAL CHEMISTRY, VOL. 59, NO. 23, DECEMBER 1, 1987
1IUTENS:TY A N D LIFETIME OUENCHING 7 +
= Lifetime
%
Table I. Sensor Absorbance Changes for a 24-h Exposure to Possible Interferences
X
5
x = Intensity
.'
t
3
X
5 '
I
chemical
+x
+
c
8
\
* L 3
I
.t
1
F0
0
t
w
0
90
60
30
120
150
2783
type
absorbance before after % change
0.05 M NaLS 0.1 M EDTA-0.5 M NaOH 0.1 M Na2Sz04 0.1 M K2Cr207-0.1 M
surfactant complexing
1.46 1.10
1.46 1.15
0.0 +4.5
reducing oxidizing
1.27 1.22
1.27 1.21
-0.8
1.0 M NaOH 1.0 M HCl acetone 95% ethanol
base acid organic organic
1.36 1.30 1.23 1.45
1.67 1.12 0.42 1.47
+22.8 -13.8 -67.2 +1.4
0.0
Oxygen Pressure (Torr)
SULFUR D I O X I D E QUENCHING
Figure 2. Comparison of lifetime (+) and intensity (X) quenching plots for 0.004-in. sensor fiim.
3.5
5
HYSTERESIS
+
5
p \ x t
I
31
X
2
I1
t
+
.5'
t
-
+
+++
1.5
xt
T
t
+
++
1(
I
0
t
0
40
80
120
163
Oxygen Pressure (Torr)
Figure 3. Hysteresis plot for luminescence of sensor fllm exposed to decreasing (X) and then increasing (+) oxygen partial pressures.
Sensor reproducibility was apparently limited by instrument drift and inhomogeneity of the complex in the sensor films. For ten air-to-N2-to-aircycles, the relative standard deviation (RSD) for the emission intensity in air was 3.4% and that in N2 was 3.7%, but the RSD for paired nitrogen-to-air intensity ratios was 0.54%. Thus, instrumental drift was indicated. With lifetime measurements the N2 values yielded an RSD of 0.38% (n = 5 ) when the sample was not moved and 2.51% (n = 5 ) when the same film was repositioned between each reading. This film also showed an RSD of 2.0% (n = 5 ) when it was in air and was not repositioned. We attribute the higher noise for the air sample to the fact that the relative noise increases with the decreasing luminescence intensity of the quenched sample. Because of the inhomogeneity of the films used in this study, we would expect that the enhanced preparation procedure described in the Experimental Section would produce a more uniform sensor. We attribute the nonlinearities of the Stern-Volmer plots to inhomogeneities in the binding sites in the polymer and, thus, differences in the local oxygen quenching constants. This conclusion is supported by several facts. The downward curvature of the Stern-Volmer plot is characteristic of multiple species. The decay curves appear to have some degree of nonexponentiality in the more heavily oxygen-quenched samples. Sulfur dioxide quenched samples exhibit gross nonexponentialities. The observed nonexponentialities are consistent with SO2being less mobile and bulkier than oxygen, which amplifies lifetime differences for different binding regions. Interferents. The solution stability tests are summarized in Table I. All gases were at 1-atm pressure except chlorine
(bleach), SO2 (Figure 4),and halothane, which was at its maximum room-temperature vapor pressure. The surfactant and the reducing agent are without detectable effect on the complex concentration. Both the EDTA and the NaOH solution are basic and cloud the polymer surface. This increases light scattering, which appears as increased absorbance. Within our experimental error, however, there was no discernible destruction of the complex. The HC1 results suggest that the acid might leach the complex; however, the K2Cr2O7solution was nearly as acidic but showed no leaching. We, therefore, attribute the apparent loss of complex to nonuniformity in the concentration of the complex across the film or in film thickness and the difficulty of reproducibly positioning the fiim in the spectrophotometer. There was no quantitative evidence for leaching of the complex into ethanol, even though ethanol is an excellent solvent for the complex. Similarly, methanol does not extract any [ R ~ ( P h ~ p h e n )from ~ ] ~ +the film as judged by the absence of color or luminescence in the methanol in which the films have soaked. Acetone, however, swelled the polymer and leached the complex out of the film as indicated by coloration and luminescence of the solvent. Table I1 summarizes the quenching studies. The excellent solution quencher Fe3+is without effect. The physiologically and medically important gases COB,nitrous oxide, cyclopropane, and halothane are all without effect at concentrations well above those normally encountered. The halocarbon refrigerant Genetron is without effect as is pump oil and the common petroleum industry pollutant H2S. The halothane and nitrous oxide results are significant as halothane is the operating room anesthetic of choice in the United States,and nitrous oxide is the primary gaseous dental anesthetic. Halothane, because of its ease of reduction, is a severe interferent for the Clark electrode. Indeed, at phys-
2784
ANALYTICAL CHEMISTRY, VOL. 59, NO. 23, DECEMBER 1, 1987
RESPONSE TIME (0.004 i n c h
Table 11. Effect of Interferents on Excited-State Lifetime or Intensity chemical
typea
Fe(N03hb bleach (C1.J
Q
SO2
Q
cyclopropane Gentron
co2 H2S N2O
pump oil
halothane
Ox, Q An An G G G, An 0 R
lifetime ( p s ) or intensity before during after 6.48 5.95 5.69 5.77 5.79 5.51
6.31 5.45 0.94 5.32 5.92 5.33
5.55 5.42 5.49
5.46 5.19 5.36
1.oooc
f i In)
2
1.5
3 sec = 5. 4
1
sec
+,
*.. 5.83 5.72 5.84 5.81 5.50
+
+
+
1.000c
5.46
...
...
"Abbreviations: Ox, oxidant; Q, quencher; 0, organic; A, anesthetic; R, refrigerant; G, gas. bO.O1 M in 0.1 M "OB. CRelative emission intensity. Estimated accuracy f l % .
20
10
0
I
30
Response Time (sec)
Flgure 5. Time response of a 0.004-in. film stuck to a glass wail on release of the vacuum to air. The initial evacuated iifetime was 5.4 PS.
iological concentrations, halothane produces signals that correspond to oxygen partial pressures in excess of an atmosphere, even in the corhplete absence of oxygen (3, 4 ) . We feared that because the Ru(I1) excited states are powerful oxidants and reductants (24), easily reduced gaseous species such as HzS would interfere. Our results, however, show that there is no detectable quenching by HzS even at 1 atm. Only Clz and SOz were effective quenchers. We consider the small degrees of apparent quenching in the remaining cases as arising from small air leaks in our vacuum system or the inability to exclude traces of air when we introduced the potential quencher. Chlorine introduced from common household bleach quenched the lifetime. The recovery of the emission and lifetime, however, was complete after the bleach was removed by rinsing with water. For both SO2 and C12we suspect that excited-state oxidative quenching of the complex is occurring. Sulfur dioxide is an especially severe interferent. A liietime Stern-Volmer plot is shown in Figure 4. Not only is quenching severe at low SOz partial pressures, but also the decay curves are extremely nonexponential. Thus, it would be virtually impossible to use the sensor in the presence of appreciable amounts of SO2. We attribute the quenching to an excitedstate electron-transfer reaction with the SOz. On the basis of lifetime data, the sensor film exhibited the same oxygen quenching constant before and after exposure to all the gaseous species. Thus, even though the gaseous species quenched, their complete removal restored the sensor to the original state. We did not evaluate the oxygen quenching constants after exposure to the solution quenchers. However, our sensor complex is extraordinarily sensitive to quenching of several of the species tested and to subtle changes in environment (25,26). Thus, the absence of any effects on emersion in the solutions virtually excludes the possibility of any penetration to the sites around the metal complex. In the absence of alteration of the local environment, we assume that the oxygen sensitivity of our sensor is unaffected by exposure to these nonpenetrating species. In summary, quenchers, solutions, or solvents that do not penetrate into the polymer will not affect the sensor concentration, degrade the sensor, or distort the response. Good solvents for the complex such as acetone and CHZClP,which penetrate and swell the film, destroy the sensor. Somewhat surprising is that relatively polar organic solvents such as ethanol are also barred from penetrating into the polymer and, thus, do not degrade the sensor. While virtually all small gaseous molecules will penetrate the sensor film, we have found few that interfere.
Nitrogen
t0.01 0.0
"
"
'
I
"
"
"
"
1.0
05
J
15
Time ( s ) Flgure 6. Response of 0.001-in. sensor in cell of Figure 1 when subjected to step changes in the oxygen concentration.
Exhale
hot
00
n
Held
Normal B r e a t h l n a
1 '
0
15
30
45
Tlme ( s ) Flgure 7. Response of sensor system of Figure 1 when subjected to breathing. Normal breathing would correspond to the rightmost three breaths.
Response Time. Figure 5 shows the response time of the 0.004-in. film when air is suddenly introduced into the evacuated cell. The response time to 95% of the final value is 3 s. Figure 6 shows the response of the 0.001-in. thick film to changes in the oxygen concentration. The terminal intensity is reached in
