650
Anal. Chem. 1986, 58,650-653 0.20
lamp, can be used. Second, the minigrid provides an electrode geometry that is centrosymmetric. Therefore, it should be possible for mathematically manipulate the experimental diffraction pattern (using an inverse Fourier transform) to obtain spatial information about the distribution of chromophore within the diffusion layer. The transform relationship between the diffusion layer and the diffraction pattern has already been shown (7) although the inversion was not accomplished. For a nonsymmetric electrode geometry like that used in the previous work, the inversion requires knowing the phase of light in the diffraction pattern. This information is difficult or impossible to obtain experimentally. However, the phase factor is zero for a centrosymmetric geometry like that provided by the minigrid. This greatly simplifies the inversion. Information such as that which would be obtained by inverting experimental diffraction patterns, would be extremely useful in the study of complex electrochemical reactions.
0.15
ABSORBANCE
0.10
ACKNOWLEDGMENT We thank R. L. McCreery (Ohio State University) for helpful discussion and Michael Mickelson and Lee Larson (Denison University) for technical assistance. Registry No. Au, 7440-57-5; o-tolidine, 29158-17-6.
0.05
4jO
440
450
WAVELENGTH (nm)
Figure 6. Absorbance as a function of wavelength at 1.3, 1.0, 0.8, and 0.4 s (listed from top) during oxidation of 1.OO mM o-tolidine. Width of thin layer cell is equal to 0.017 28 crn.
tively, time averaging could be employed. (All spectra shown are single runs.)
CONCLUSIONS The results presented here show that spectra of adequate resolution can be obtained by using 1000 lines-per-inch micromesh as both electrode and diffraction grating. In fact, the spectral band-pass employed in these experiments is comparable to that provided by many commercially available spectrophotometers when operated a t high scan rates. Extremely fast acquisition of spectra would be possible using diffraction if the P M T was simply replaced with an array detector. This work extends the potential of diffractive spectroelectrochemistry in two important ways. First, any readily available white light source, such as a tungsten or xenon arc
LITERATURE CITED (1) Rossi, P.; McCurdy, C. W.; McCreery, R. L. J . Am. Chem. SOC. 1981, 103, 2524. (2) Rossi, P.; McCreery, R. L. J . Electroanal. Chem. 1983, f 5 f , 47. (3) Chwu-Ching, J.; Lavine, B. K.; McCreery, R. L. Anal. Chem. 1985. 5 7 , 752. (4) Murray, R. W.; Heinernan, W. R.;O’Dom, G. W. Anal. Chem. 1967, 3 9 , 1666. (5) Klein, M. V. “Optics”; Wiiey: New York, 1970; Chapter 7 . (6) Petek, M.; Neal, T. E.; Murray, R. W. Anal. Chem. 1971, 4 3 , 1069. (7) Rossi, P. Ph.D. Thesis, Ohio State University, 1982.
Rick A. Fair Daniel E. Ryan Peter K. Smith Paula Rossi Melaragno* Department of Chemistry Denison University Granville, Ohio 43023
RECEIVED for review July 1, 1985. Resubmitted November 15, 1985. Accepted November 15, 1985. This work was supported by grants from the Research Corporation, the Apple Education Foundation, and Denison University Research Foundation.
Potentiometric Gas Sensor for the Determination of Free Chlorine in Static or Flow Injection Analysis Systems Sir: Chlorine is extensively used in many industrial processes and as a biocide to destroy harmful bacteria in drinking water and other types of water, such as in swimming pools and plant effluents. The speciation of chlorine in such media can be highly complex ( I ) . I t is customary to classify chlorine-containing species as “free chlorine” (chlorine plus hypochlorous acid) or “combined chlorine” (chloramines, etc.). Numerous methods of widely different efficacy and convenience have been developed for the determination of chlorine in aqueous samples. In many of these methods “combined
chlorine” (total oxidizing power) is determined, rather than free chlorine alone ( I , 2 ) . The most commonly used methods involve colorimetric, potentiometric or amperometric procedures, each of which has particular advantages and limitations. A typical colorimetric method is based on the reaction of chlorine with N,N-diethyl-p-phenylenediamine producing a red color which, however, is not ideally stable ( I , 2). The so-called “residual chlorine electrode” (3) is a potentiometric sensor for iodine liberated from excess potassium iodide; many oxidizing agents interfere. Finally, a widely used amperometric
0003-2700/86/0358-0650$01.50/00 1986 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 58, NO. 3, MARCH 1986
method is based on titration of free chlorine (or iodine liberated from excess potassium iodide) with phenylarsine oxide ( 4 ) according to reaction 1 (where X = C1 or I). The lower detection limit (20 pg L-' or less) is favorable, but free chlorine tends to attack the electrode.
Xz + C6H,As0
+ 2Hz0
-+
2X-
2H+
+ C,jH,AsO(OH)Z
(1)
Potentiometric gas sensors in which the analyte gas is allowed to diffuse through a hydrophobic gas-permeable membrane into a test solution containing the potentiometric sensor have been developed for a number of gases, such as ammonia and carbon dioxide ( 3 ) . Such sensors have the important advantage that only gases producing those particular species in the test solution to which the sensor responds can interfere. Preliminary information on a sensor for chlorine was published by Ross et al. in 1973 ( 5 ) ,but the lower detection limit was M) and the sensor apparently was not unfavorable ( 5 X developed further. We describe here a sensor with a lower M free chlorine. detection limit of 5 X A recent development in the determination of gaseous analytes is the incorporation of a gas diffusion cell in a flow injection analysis system. Such arrangements have been described by Meyerhoff (6) for ammonia (potentiometric detection) and by Pacey (7) for ozone (colorimetric detection). We describe here a system for free chlorine with potentiometric detection.
EXPERIMENTAL SECTION Reagents. The source of free chlorine was sodium hypochlorite added to 0.1 M sodium hydrogen sulfate or sulfuric acid. Solutions of sodium hypochlorite were standardized by adding excess potassium iodide and titrating with sodium thiosulfate solution which, in turn, had been standardized against potassium iodate. Apparatus. The potentiometric sensor for chlorine was constructed using an Orion Model 95-10 ammonia electrode body. Hydrophobic membranes with a range of pore sizes (Gore-Tex, donated by W. L. Gore and Associates, Inc.) were tested. Larger pore sizes gave short response times but relatively poor (high) lower detection limits. Smaller pore sizes gave better detection limits but longer response times and substantially less sensitivity (less steep slopes of calibration curves). As a trade-off, 0.45 pm Gore-Tex PTFE on polyethylene or polypropylene scrim was chosen. The sensing electrode was a silver disk coated electrolytically with silver chloride. Both flat and slightly convex disks were tested. The sizes of these disks were also varied and a diameter of 8 mm was chosen. Other chloride ion sensors were also tested. These were AgCl + AgzS and Hg2Cl2+ HgS electrodes with membranes fabricated by pressing 1:l mole ratio mixtures of the components into pellets. The reference electrode was a double junction silver-silver chloride electrode with the same buffer as that used for the internal solution as salt bridge. A flow injection analysis system incorporating a gas diffusion cell was constructed as shown in Figure 1. This system is similar to that described by Meyerhoff for ammonia (6). Optimum experimental conditions were as follows: buffer stream, 0.1 M Na2S04+ sufficient NaHSO, to give a pH value of 4.5, flow rate 0.36 mL m i d ; analyte stream, 0.1 M HZSO4 carrier + NaOCl sample, flow rate 0.62 mL min-'; tubing diameter, 0.5 mm; mixing coil length, 6 ft; gas diffusion cell length, 8 cm; sample size, 100 wL. Throughput was 20 samples h-l. RESULTS AND DISCUSSION Potentiometric Sensor. Chlorine gas diffuses from the analyte solution through the gas-permeable hydrophobic membrane and reacts with the internal buffer according to reaction 2. The silver-silver chloride electrode immersed in Clz
+ H2O G C1- + H++ HOCl
(2)
the buffer measures the activity of chloride ion produced. The equilibrium constant of reaction 2 is approximately 3.9 x M2 (I- 0, T = 298 K), as measured by several workers using
651
-1
L-'
FJ
s
MC
Flgure 1. Flow injection analysis system: A, buffer stream; B, analyte stream; P, pump; S, sample injection valve: MC, mixing coil; GC, gas diffusion cell; W, waste; I, electrode cell compartment (see insert); M, meter; R, recorder. Insert: SB, sak bridge; CF, ceramic frit; TC, Teflon connector; TS, Teflon spacer; IE, indicator electrode;CL, connecting lead.
a variety of techniques (8-10). While the detailed mechanism of reaction 2 is complex (9),the kinetics of the overall reaction are fast enough for the present purpose. For example, in 1 M NaC10, a t 298 K, the rate constants for the forward and back reactions were found to be 8.7 f 0.2 smland (2.66 0.03)
*
x 104 M-2 (10). The equilibrium potential of a typical potentiometric gas sensor will be reached when the partial pressure of the gas becomes the same over the internal (sensing) solution as it is over the external (analyte) solution. If the sensor is subsequently placed in a solution containing no analyte gas, the gas will diffuse from the internal to the external solution, so that the sensor quickly reverts to its original physical state and potential. This is the case, for example, for the ammonia sensor utilizing ammonium chloride as the internal solution and a pH electrode as sensor. In the present case, the chemistry is less favorable. The equilibrium concentrations of chlorine and of chloride ion can be calculated from eq 2 as a function of total free chlorine concentration, C, and of pH by substituting the stoichiometric relationships [Cl,] = C [Cl-] and [HOCl] = [Cl-1. From such calculations and additional considerations two main conclusions can be drawn. First, as far as the analyte solution is concerned, a t even the lowest reasonable p H values, [C12]
159
-
E 125
LL
100
75
'
25 . 3 2.3
3.0
4.2
5 9
6 .3
-Log (Ci
Figure 2. Calibration curves for potentiometric sensor: 0 , flat electrode; A, convex electrode. Slopes and correlation coefficients for linear sections are as follows (mean of 10 runs): flat, -49.7, 0.999; convex, -58.5,0.998. Lower detection limits are as follows (intersections of linear sections of each curve): flat, 5 X lo-' M; convex, 2 x 10-5 M.
SO,), uH+ = 3.16 x fcl- = 0.73 (calculated from DebyeHuckel equation with I = 0.13 and u = 3 A); inside solution (mainly 0.1 M Na2S04),I 0.3, a H t = 3.16 X loT5,fcr = 0.66. The response ranges of the two electrodes were also different, as shown in Figure 2. The flat electrode responded to lower chlorine concentrations but its response slope was less steep than that of the convex electrode; i.e., it had a better lower detection limit but poorer sensitivity. The cause of these differences is not known. What is known is that convex electrodes in gas sensors tend to respond faster, presumably because of better overall contact of the inevitably rough electrode surface with the hydrophobic membrane so that the layer of solution in which the activity of the sensed ion is measured is thinner and the equilibrium potential is reached more quickly (11). However, our two electrodes had similar M approximately 10 response times, requiring for C = min to reach potentials that remained constant within A 1 mV for 30 min. Both electrodes could be used for another run
-
\
25 0
"Adjusted by adding sufficient NaHS04 to 0.1 M Na2S04, except for pH 0.8 and 1.5 for which 2.5 and 0.1 M NaHSO, were used. *In mV per decade in concentration of free chlorine over a M. All potenitals were mearange of approximately to sured after 10 min (see text).
1.00 x 10-2 1.00 x io-3 1.00 x 1.00 x 10-5
a'
a 0 0 1 0
2 3
3 0
-Log
4 3
5 3
6 3
(C)
Flgure 3. Calibration curve for flow injection analysis system: linear between lo-' and 3 X M chlorine, slope -57.8, correlation coefficient 0.999.
over the concentration range from to M without changing the internal buffer after leaving the sensors in water for 10 h. This recovery rate is much slower than for such other gas sensors as those for ammonia and carbon dioxide and is the main limitation of using the sensors in static solutions. Flow Injection Analysis System. Results are shown in Figure 3. Peak heights are reported as AE = E(base line) - E(peak1. Reproducibility of individual peak heights was always better than f l mV. In an attempt to improve the lower detection limit we replaced the silver-silver chloride sensing electrode with a mercury(1) chloride-mercury(I1) sulfide ion selective electrode, which responds to somewhat lower chloride ion activities in solutions containing only potassium chloride and supporting electrolyte. However, the lower detection limit for chlorine remained the same, while at chlorine concentraM the electrode became passivated, tions greater than probably as a result of surface oxidation. In preliminary studies of interferences, it was found that hydrogen sulfide interferes under the experimental conditions described here. It appears that the surface of the silver chloride of the sensing electrode is quickly converted into silver sulfide, which responds to the small fraction of hydrogen sulfide present as sulfide ion at a pH value of 4.5. It is to be noted, however, that hydrogen sulfide and chlorine cannot coexist. We are nevertheless studying this system further in an attempt to develop a hydrogen sulfide gas sensor. We are also investigating possible interference by chlorine dioxide, which coexists with chlorine in certain samples. In preliminary studies of the applicability of the flow injection analysis system to real samples, we have measured the chlorine content of a sample of swimming pool water. Such water normally contains only free chlorine (mainly as hypochlorous acid) and no combined chlorine. The free chlorine content found by flow injection analysis was 1.12 f 0.02 ppm, as compared to 1.054 f 0.005 ppm found by standard iodometric titration; the agreement is satisfactory, considering that this free chlorine concentration falls in the curved region of the FIA calibration plot. CONCLUSIONS
Of the two methods described here for the determination of free chlorine, flow injection analysis provides faster sample throughput and better reproducibility. When the potentiometric sensor is used in static solutions, the lower detection limit is improved but response and recovery times are long. Registry No. Clz, 7782-50-5; ClHO, 7790-92-3. LITERATURE CITED (1)
Jolley, R. L.; Carpenter, J. H. "Aqueous Chemlstry of Chlorine: Chemistry, Analysis, and Environmental Fate of Reactive Oxidant Species": Oak Ridge National Laboratory, 1982.
Anal. Chem. 1986, 58,653-654 (2) "Standard Methods for the Examination of Water and Wastewater"; American Public Health Association, 14th ed.; American Public Health Association: Washington, DC, 1975. (3) "Handbook of Electrode Technology"; Orion Research, Inc.: Cam-. bridge, MA, 1982. (4) Morrow, J. J.; Roop, R. N. J.-Am. Water Works Assoc. 1975, 67, 184-186. (5) Ross, J. W.; Riseman, J. H.; Krueger, J. A. Pure Appl. Chem. 1973, 36,473. (6) Meyerhoff, M. E.; Fraticelli, J. M. Anal. Lett. 1981, 74, 415-432. (7) Straka, M. R.; Gordon, G.; Pacey, G. E. Anal. Chem. 1985, 57, 1799-1803. (8) Connick, R. E.: Chia, Y.-T. J . A m . Chem. S O C . 1959, 87, 1280-1 284. (9) Eigen, M.; Kustin, K. J . A m . Chem. SOC. 1982, 8 4 , 1355-1361 (IO) Drougge. L.; Elding, L. I . Inorg. Chem. 1985, 2 4 , 2292-2297.
653
(11) Bailey, R. L.; Riley, M. Analyst(London) 1975, 700,145-156
J. F. Coetzee* C. Gunaratna Department of Chemistry University of Pittsburgh Pittsburgh, Pennsylvania 15260
RECEIVED for review August 29, 1985. Accepted November 14,1985. This research was supported by the National Science Foundation under Grants No. CHE-8106778 and CHE8408411.
Comments on Determination of Sulfuric Acid and Ammonium Sulfates by Means of a Computer-Controlled Thermodenuder System Sir: Slanina et al. (1)have made an important advance in the measurement of sulfate aerosols with their highly sensitive thermodenuder system. In this instrument air is drawn through 120 "C and 220 "C denuders connected in series, and particulate sulfuric acid and the ammonium sulfates are converted into sulfuric acid vapor, which is sorbed on the denuder walls. At the end of the sampling period each denuder is individually heated to 800 "C, and the evolved SO,, which results from decomposition of the sorbed sulfuric acid, is measured by a flame photometric detector. The purpose of this note is to clarify the speciation capabilities of the instrument and particularly the difference between volatilizable and titratable sulfuric acid. It will be assumed that only sulfuric acid, ammonium sulfate, and intermediate stoichiometries are present. Metal sulfates, organosulfur compounds, and other acidic and ammonium species (e.g., nitrates) are taken to be absent. The acidic and ammonium fractions of sulfate aerosol can exist either as internal or external mixtures, and the response of the thermodenuder system will depend on the type of mixture. In an internal mixture both the acidic and ammonium fractions are present in the same particle, and the composition of such an aerosol can be represented formally as (NH4),HyS04where x + y = 2. (In such a representation x and y need not be integers.) An internal mixture can occur when H2S04droplets resulting, for example, from the atmospheric oxidation of SO, are partially neutralized by NH3. The only case in which the acidic and ammonium fractions exist as a pure external mixture is the aerosol consisting of HZSO4 droplets and (NH4),S04 particles. Such a situation could result when SO, is emitted into an aerosol containing (NH4)&?04 and is then oxidized to H2SO4. The response of the thermodenuder system to these two types of mixtures is similar to that of other thermal speciation-flame photometric monitors for aerosol sulfur and has been discussed previously (2-4). For the situation in which all particles in an internally mixed aerosol are more acidic than NH4HS04(i.e., y > x ) , the following process occurs in the 120 "C denuder: ("JxH,SO4
-.+[(y -
x ) / ~ ] H , S O+ ~ ~xNH4HS04
(1)
The volatilized HzS04 is trapped on the walls of the CuO/Cu denuder, and the NH4HS04is transmitted to the 220 "C denuder where it is thermally decomposed as follows: 0003-2700/86/0358-0653$01 SO/O
xNH4HS04
-+
xH2S04t + xNH3t
(2)
The volatilized HzS04from this process is sorbed on the walls of the 220 OC denuder. (Kiyoura and Urano (5)point out that other sulfur products (SO, and SO3) are possible in eq 2, but these will also be scavenged in the denuder.) Subsequent thermal desorption a t 800 "C of the 120 and 220 "C denuders produces flame photometer signals Slz0= (y - x)/2 and Szzo = x (per mole of total sulfate sampled). Thus the amount of titratable sulfuric acid (y/2) corresponds to ( S l z 0 + S220/2). For an internally mixed aerosol in which all particles are less acidic than NH4HS04 (i.e., x > y ) , the reaction shown in eq 3 occurs in the 120 "C denuder, and there is no volatilizable
("JXHyS04
-
[ ( x - Y ) / ~ I N H , ?+ [ ( x
+ ~)/21"4HS04
(3)
H2S04a t this temperature (i.e., Slzo= 0). As a result, the thermodenuder system provides no information on the degree of acidity for internally mixed sulfate aerosols with stoichiometries between NH4HS04and (NH4),SO4. In the 220 "C denuder the decomposition of NH4HS04proceeds as indicated in eq 2. For the external mixture consisting of (NH,),S04 particles and H,S04 in relative molar amounts x/2 and y/2 respectively, of the 120 "C denuder corresponds to y/2 the response (Slzo) (per mole of total sulfate sampled). In this case the amount of volatilized H2S04in the 120 "C denuder is equal to the amount of titratable HzS04. Unfortunately, the state of mixing of the sulfate aerosol is usually not known, and only upper and lower limits for the amount (y/2) of titratable sulfuric acid can be determined. The lower limit corresponds to the external mixing interpretation; Le., y/2 = s,,,,. The upper limit is given by the internal mixing assumption for y > x ; i.e., y/2 = SI20 + Szz0/2. Similar considerations apply to other aerosol sulfur analyzers employing thermal speciation (2-4, 6-9). Registry No. H2SOI, 7664-93-9; (NH,),S04, 7783-20-2; NHdHSO4, 7803-63-6.
LITERATURE CITED (1) Slanina, J.; Schoonebeck, C. A. M.; Klockow, D.; Niessner, R. Anal. Chem. 1985, 57, 1955-1960. (2) Huntzicker, J. J.; Hoffman, R. S.; Ling, C.-S. Afmos. Envlron. 1978, 12, 83-88, (3) Huntzicker, J. J.; Hoffman, R. S.; Cary, R. A. Environ. Scl. Techno/. 1984, 78, 962-967.
0 1986 American Chemical Society