ANALYST, JANUARY 1991, VOL. 116 49 Indirect Determination of Chloride by Gas-diffusion Flow With Amperometric Detection SneZana D. Nikolic and Emil B. Milosavljevic Faculty of Chemistry, University of Belgrade, P.O. Box 550, 1 101 Belgrade, Yugoslavia James L. Hendrix and John H. Nelson Departments of Chemistry and Chemical and Metallurgical Engineering, Macka y University of Nevada, Reno, NV 89557, USA Injection School of Mines, A rapid, indirect gas-diffusion flow injection (FI) method with amperometric detection has been developed for the selective and sensitive determination of CI-. The method is based on permanganate oxidation of CI- to chlorine. The chlorine diffuses through the micro-porous membrane and is quantified amperometrically at a platinum working electrode. Calibration graphs were linear up to the maximum concentration of CI- investigated (10 mmol dm-3).The precision of the technique was better than a relative standard deviation of 1% at 2 mmol dm-3 levels and better than 2% at 10 pmol dm-3, with a throughput of 30 samples h-1. At elevated temperatures (50 "C) and higher acidities (5 mol dm-3 H2S04), the detection limit was 0.1 pmol dm-3 (0.7 ng of CI-). The effects of temperature, sample acidity, working potential and interferents on the FI signals were studied. The method was successfully applied to the determination of CI- in natural and tap waters. Keywords : Gas-diffusion flow injection method; ampe rome tric detection; indirect chloride determination The ubiquitous nature of the chloride ion, and its importance, makes the determination of C1- one of the most frequently required analyses. It is not surprising, therefore, that a number of flow injection (FI) methods have been developed for this analyte.One of the first FI publications1 describes the spectrophotometric determination of CI- in brackish waters. Other papers utilizing spectrophotometric detection fol- lowed.2-11 An interesting approach to the determination of C1- is the combination of FI dialysis with spectrophotometric detection.2.12 Martinez-Jimenez et al. *3,14 determined C1- and mixtures of C1- and I - , with use of FI in conjunction with indirect atomic absorption spectrometric detection. The combination of FI and potentiometry for the determination of CI - has also been extensively studied. Various ion-selective electrodes'5-26 and a copper-wire indicator electrode27 were used for this purpose.Zaitsu et aZ.28 combined FI with turbidimetry in order to develop a method for the determina- tion of CI-. The same analyte was determined by FI methods, with use of chemiluminescence29 and condu~tivity~~ detection. A literature search revealed that there are only two FI amperometric methods designed for the determination of CI-. Polta and Johnson31 utilized pulsed amperometric detection to determine C1-, which alters the rate of surface oxide formation at the platinum working electrode. A triple-step potential waveform had to be used, hence the method required an apparatus more sophisticated than that required for single-potential amperometry . Also, the method suffers from various interferents and, as these workers pointed out, it is more suitable for detection in liquid chromatographic analysis where sufficient resolution of species is provided.The FI method, with direct amperometric detection of C1- at a silver working electrode, developed by Frenzel et al. ,32 is also non-selective as any species that forms either insoluble silver salts or stable complexes with Ag+ would necessarily inter- fere. Hence, special calibration methods were required. The present paper describes an approach to the use of amperometric detection for the determination of C1-. In the FI manifold developed, the injected analyte is converted on-line into chlorine, which diffuses from the donor stream through the micro-porous membrane into the acceptor solu- tion.The latter carries chlorine to the flow-through ampero- metric detector, where it is reduced at a platinum working electrode. The cathodic current measured is proportional to the concentration of C1- in the original sample or standard. To the best of our knowledge, there are only two FI publica- tions33.34 that combine gas diffusion with amperometric detection. This is surprising, as the inherent sensitivity of amperometry and selectivity of the gas-diffusion processes render this combination a powerful analytical tool. Experimental Reagents and Materials All the chemicals used were of analytical-reagent grade. The aqueous reagent and standard solutions were stored in polyethylene bottles. De-ionized water was used throughout. A saturated solution of KMn04 served as the oxidizing agent.It was prepared by boiling the saturated solution with subsequent filtering in order to remove MnOz and any excess of KMn04 that might be present. A stock solution of 0.1 mol dm-3 NaCl was prepared from BDH (Poole, Dorset, UK) concentrated volumetric standards, which are certified to have an accuracy within the factor limits of 0.999 and 1.001. Standard C1- solutions, which in most of the experiments were made in 3 mol dm-3 H2S04, were prepared by diluting aliquots of the stock solution to the appropriate volumes. Instrumentation and Apparatus The FI manifold is illustrated in Fig. 1. Two peristaltic pumps were used. One was a Model Mini S-840 (Ismatec, Zurich, Switzerland) and the other was a Model HPB 5400 (Iskra, Kranj, Yugoslavia).The injection valve was a Model 5020 (Rheodyne, Cotati, CA, USA) equipped with a 200 vl sample loop. The gas-diffusion unit, which was obtained from Shenyang Film-Projector Reflector Factory (Shenyang, China), is similar in construction to the Tecator (Hoganas, Sweden) Chemifold V gas-diffusion cell. The membrane used, which was of Teflon, was supplied with the unit. All connections were made with 0.5 mm i.d. Teflon tubing except for the long mixing coil (MC1), which was made from a 0.8 mm i.d. Teflon tube. The flow-through amperometric cell (Dionex, Sunnyvale, CA, USA), described earlier,35 consisted of platinum working and counter electrodes. The reference electrode was an Ag-AgCI (1 mol dm-3 NaCI) electrode and it was separated from the flowing stream by an ion-exchange Nafion mem-50 ANALYST, JANUARY 1991, VOL.116 RF I P 7 7 - U ' Fig. 1 F1 manifold used for the indirect determination of chloride: C, carrier (3 rnol dm-3 H2S04); R, reagent (saturated KMnO,); AS, acceptor solution (0.01 rnol dm-3 H2S04); P, peristaltic pump; PD, pulse damper; I, injection valve; MC1, long mixing coil (2 m x 0.8 mm i.d.); D, diffusion cell; CTB, constant-temperature bath; MC2, short mixing coil (0.3 m x 0.5 mm i.d.); EC, electrochemical flow-through cell; PO, potentiostat; RE, recorder; and W, waste. Flow-rates are given in ml min-1 brane (all electrode potentials are reported versus this reference electrode). The platinum working electrode was polished occasionally with a small amount of toothpaste and a paper tissue.The potential to the flow-through amperometric cell was applied and currents were measured with a Model MA5450 polarograph (Iskra, Kranj, Yugoslavia); the result- ing F1 signals were recorded on a Model 61 Servograph (Radiometer, Copenhagen, Denmark) strip-chart recorder. The measurements were made with both donor and acceptor streams continuously flowing. Temperature regulation was achieved with a constant- temperature bath , type NBE (VEB Prufgerate-Werl, Medingen, Germany). Results and Discussion The rate of oxidation of CI- to chlorine by permanganate is slow. Hence, in order to apply this reaction in FI, steps must be taken to increase the reaction rate. The logical step is to use the saturated KMn04 solution as the oxidant, as has been carried out recently in the non-FI method developed for the determination of C1- by flame infrared emission .36 The effects of several parameters on the performance of the FI system, illustrated in Fig.1, designed for the indirect determination of CI-, were studied. I n order to find a suitable acceptor solution, several potential candidates were tested (H20, NaOH, Na2C03, KN03 and H2S04). It was found that the optimum signal to noise ratio was obtained with 0.01 mol dm-3 H2SO4, and in all the subsequent experiments this medium was used as the acceptor solution. The effect of the applied potential at the working platinum electrode was investigated in the range +0.10 to +0.70 V versus an Ag-AgC1 reference electrode. The hydrodynamic voltammogram for a 2.50 mmol dm-3 sodium chloride standard in 3 mol dm-3 H2SO4 is shown in Fig.2. As can be seen, the optimum potential is +0.30 V. However, if selectivity concerns dictate otherwise, slight variations of the applied potential are possible, bearing in mind that at potentials lower than about +O. 15 V, the background current becomes too high, probably as a result of the onset of oxonium ion reduction. The effect of H2SO4 concentration on the peak currents was studied by injecting the same CI- standard (2.50 mmol dm-3) while varying the concentration of the acid from 1.0 to 5.0 rnol dm-3. The data obtained are shown in Fig. 3. As can be I r 0 0 0 0 0 0 I I I I 1 I 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 PotentialN versus Ag-Ag CI Fig. 2 2.50 mmol dm-3 sodium chloride standard Hydrodynamic voltammogram for a 200 p1 injection of a 2.8 1 2.4 2.0 a 2 5 1.2 z u 0.8 0.4 0 1.0 2.0 3.Q 4.0 5.0 [H2SO$/mol dm-3 Fig.3 Variation of peak current as a function of sulphuric acid concentration. 0, Experimental data points; and -, calculated according to the equation i = exp(-3.50 + 0.86~). For details see text ,4min, Scan - Fig. 4 Response of the amperometric detector to: (a) nine repetitive injections of a 2.00 mmol dm-3 chloride standard; and ( b ) five repetitive injections of a 10.0 pmol dm-3 chloride standard seen, the current increases exponentially with an increase in H2S04 concentration. The simple equation of the type: i = exp(-3.50 + 0 . 8 6 ~ ) (where i is the current in PA, and c is the concentration of &SO4 in rnol dm-3) fits the data very well. The corresponding correlation coefficient was found to be 0.9989.The temperature effects were studied by injecting a 2.50 mmol dm-3 sodium chloride standard made in 3 rnol dm-3 H2S04, while varying the temperature in the interval 30-60 "C. A linear relationship was obtained with an increase in sensitivity of (0.187 -t 0.009) yA "C-1, with a correlation coefficient of 0.9975. This finding is interesting, as the change in temperature not only affects the oxidation rate, but also the solubility of the chlorine formed, the diffusion process and theANALYST, JANUARY 1991, VOL. 116 51 Table 1 Solutions tested for their possible interference* Compound NHjSCN CH3COONa NazHPOl Na2EDTAt NaF NH4N03 Concentratiodmol dm-3 0.01 0.1 0.1 0.01 0.1 0.1 * The response of the amperometric detector to 200 pl injections of the solutions tested could not be distinguished from the background noise.T Na2EDTA = Disodium ethylenediaminetetraacetate. Table 2 Comparison of FI results for the determination of C1- in the presence of Br- and I- [all samples contained 2.50 mmol dm-3 (88.6 pg ml-1) C1-] Sample c1- C1 -/Br- 'r c1 -/I - $ C1- /Br- 8 FI/pg ml- Difference (%)* 88.6 5 0.3 0 96.5 k 0.9 +8.9 88.2 -t 0.7 -0.45 84.3 k 0.3 -4.8 * Compared to a pure C1- (88.6 pg ml-1) standard. t C1- + 1 pg ml-1 of Br-. $ C1- + 1 pg ml-* of I-. Q C1- + 1 pg ml-1 of Br- + 0.5 mmol dm-3 103- (bromine expelled by boiling the solution for 10 min). Table 3 Comparison of C1- determination by FI and argentimetric titration in three water samples Conccntratiordpg ml- ~ Sample Argentimetric FI Label Belgrade tap water 22.0 k 0.1 21.2 k 0.2 Prolom" 4.60 k 0.08 4.50 k 0.1 32.0 Knjaz MiloS" 14.0 2 0.2 12.9 -t 0.2 12.0 * Commercial mineral waters. reduction of chlorine at a platinum working electrode.The last process is by itself a complicated one. Temperature changes affect the working potential (they change the potential of the reference electrode), and also the diffusive and migratory properties of the analyte, etc. As satisfactory sensitivity is achieved even at the lowest temperature studied (30 "C), most of the subsequent experiments were carried out at this temperature. The linearity studies were conducted by injecting in triplicate a total of eight standards between 0.10 and 10 mmol dm-3 made in 3 mol dm-3 H2S04. The linear regression equation for a typical calibration run was: i = (-5.05 k 0.04) x 10-3 + (0.164 k 0.007) x c (i is the peak current in PA, and c is the concentration of C1- in mmol dm-3), with a correlation coefficient of 0.9992 (all the statistics were calculated for a 95% confidence level).The repeatability of the analytical system is illustrated in Fig. 4. For example, the relative standard deviation for a 2.00 mmol dm-3 standard was found to be 0.8% ( n = 9). The detection limit under these experimental conditions (30 "C; 3 mol dm-3 H2SO4), calcu- lated according to the recommended procedure,37 was 5 pmol dm-3 of CI-. At elevated temperatures (50 "C) and higher acidities (standards were made in 5 mol dm-3 H2S04) the detection limit was 0.1 pmol dm-3, which corresponded to 0.71 ng of CI- (the sample loop volume was 200 pl).It has been established previously that PTFE membranes used in the FI gas-diffusion studies are effective barriers for ionic species.38.39 Nevertheless, a number of anions were tested. The concentrations of these species given in Table 1 are the maximum concentrations at which they were tested. As expected, in all these examples the response of the amperometric detector could not be distinguished from the baseline. It has been established previously34 that anions such as NOI-, S032-, C03"-, S2032-, CN- and S2-, which, when acidified, form acidic gases, could potentially interfere when a particular amperometric flow-through cell is used. These anions at sufficiently high concentrations, even if they are not electroactive at the working potential, could interfere in an indirect manner.The Ag-AgC1 reference electrode in the configuration used is separated from the flowing stream by the ion-exchange Nafion membrane. If the buffer capacity of the acceptor solution is too low, there will be a significant change in pH when the acidic gases, formed by on-line acidification in the FI manifold, diffuse through the Teflon membrane and are trapped in the acceptor solution. This pH change alters the potential of the reference electrode assembly, which is probably induced by a shift in the ion-exchange equilibria at the Nafion membrane. However, in the method developed in this work, the standards and samples are acidified off-line as they are made in 3 mol dm-3 H2S04, so that the aforemen- tioned anions would not interfere in the determination of CI- , as they are evolved prior to injection.(Danger! If some of the aforementioned anions are present, the acidification of the samples should be performed with all due precautions as in some instances poisonous gases are formed.) Other potential interferents are the anions that can be oxidized on-line by permanganate to form molecular species. If these species diffuse through the membrane and are reducible at the platinum electrode at the applied potential, they would cause a positive error in the determination of CI-. Likely candidates are Br- and I-. It has been established that Br- interferes, whereas I- does not. Under the experimental conditions used for the determination of C1-, it is probable that Br- is mainly oxidized to bromine, while I- yields higher oxidation states, which form ionic species.Bromide and I- can be present in natural waters up to levels of 1 and 0.1 pg ml-1, respectively. As can be seen from Table 2, a 2.50 mmol dm-3 C1- standard spiked with Br- at 1 pg ml-1 levels increases the signal by 8.9%. Kubala et a1.,36 in their flame infrared emission method for the determination of CI-, utilized the iodate pre-treatment method, which is usually applied to the determination of CI- by classical argentimetric procedures: 103- + 6Br- + 6H+ -+ I- + 3Br2 + 3H20 This pre-treatment produces bromine, which can be boiled out of the solution, and I - , which was shown not to interfere with the FI method for the determination of CI-. The applicability of this pre-treatment method is also illustrated in Table 2, from which it can be seen that iodate treatment decreases the absolute percentage difference from the pure C1- standard by about 50% (see Table 2).In order to illustrate the potential of the indirect FUgas- diffusiodamperometric method for the determination of CI- , three water samples have been analysed. Table 3 compares the values obtained with those given by argentimetric titrations. It is interesting to note that FI results always gave slightly lower values than those of argentimetric titrations. This could be explained by the fact that argentimetric titrations are influ- enced by the presence of anions such as phosphate, which might be present in the water samples analysed. Therefore, considering that the titration method is not error free, the agreement between the results obtained by the two methods is very good.The authors acknowledge the financial support of the United States Bureau of Mines under the Mining and Mineral Resources Institute Generic Center programme (Grant num- ber G1125132-3205, Mineral Industry Waste Treatment and Recovery Generic Center) and the Serbian Republic Research Fund.52 ANALYST, JANUARY 1991, VOL. 116 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 References RgZiEka, J., Stewart, J. W. B., and Zagatto, E. A. , Anal. Chim. Acta, 1976,81, 387. Hansen, E. H., and RgiiEka, J., Anal. Chim. Acta, 1976, 87, 353 * Slanina, J., Bakker, F., Bruyn-Hes, A., and Mols, J. J., Anal. Chim. Acta, 1980, 113, 331. Krug, F.J . , Pessenda, L. C. R., Zagatto, E. A. G., Jacintho, A. O., and Reis, B. F., Anal. Chim. Acta, 1981, 130, 409. Rosner, B., and Schwedt, G., Fresenius Z. Anal. Chem., 1983, 315, 197. Alexander, P. W., and Thalib, A., Anal. Chem., 1983,55,497. van Staden, J. F., Fresenius Z. Anal. Chem., 1985,322, 36. Faizullah, A. T., and Townshend, A., Anal. Chim. Acta, 1986, 179, 233. Whitman, D. A., Christian, G. D., and RfiiiEka, J., Analyst, 1988, 113, 1821. Frenzel, W., Fresenius Z. Anal. Chem., 1988, 329, 668. Eremina, I. D., Shpigun, L. K., and Zolotov, Yu. A., Zh. Anal. Khirn., 1989, 44, 399. Sundqvist, U., Fresenius Z. Anal. Chem., 1988, 329, 688. Martinez-Jimenez, P., Gallego, M., and Valcarcel, M., J . Anal. At. Spectrom., 1987, 2, 211. Martinez-Jimenez, P., Gallego, M., and Valcarcel, M., Anal.Chim. Acta, 1987, 193, 127. Miiller, H., Anal. Chem. Symp. Ser., 1981, 8, 279. Virtanen, R., Anal. Chem. Symp. Ser., 1981, 8,375. Gao, Z., and Lu, M., Huanjing Kexue, 1981, 2, 376. Trojanowicz, M., and Matuszevski, W., Anal. Chim. Acta, 1982, 138, 71. Trojanowicz, M., and Matuszevski, W., Anal. Chim. Acta, 1983, 151, 77. Ilcheva, L., and Cammann, K., Fresenius 2. Anal. Chem., 1985, 322, 232. 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 van Staden, J. F., Anal. Chim. Acta, 1986, 179, 407. van Staden, J. F., Anal. Lett., 1986, 19, 1407. van Staden, J. F., Analyst, 1986, 111, 1231. Alegret, S., Alonso, J., Bartoli, J., Garcia-Raurich, J., and Martinez-Fabregas, E., J . Pharm. Biomed. Anal., 1988,6,749. Cammann, K., Fresenius 2. Anal. Chem., 1988, 329, 691. Cardwell, T. J., Cattrall, R. W., Hansen, P. C., and Hamilton, I. C., Anal. Chim. Acta, 1988, 214, 359. Alexander, P. W., Haddad, P. R., and Trojanowicz, M., Anal. Chem., 1984,56, 2417. Zaitsu, T., Maehara, M., and Toei, K., Bunseki Kagaku, 1984, 33, 149. Cooper, M. M., and Spurlin, S. R., Anal. Lett., 1986, 19,2221. Lach, G., and Bachmann, K., Anal. Chim. Acta, 1987,196,163. Polta, J. A., and Johnson, D. C., Anal. Chem., 1985,57,1373. Frenzel, F., Almhofer, N., Citroni, G., and Hager, W., Fresenius Z. Anal. Chem., 1988, 330,489. Granados, M., Maspoch, S. , and Blanco, M., Anal. Chim. Acta, 1986, 179, 445. MilosavljeviC, E. B., Solujie, L., Hendrix, J. L., and Nelson, J. H., Anal. Chem., 1988, 60, 2791. Rocklin, R. D., and Johnson, D. C., Anal. Chem., 1983,55,4. Kubala, S. W.. Tilotta, D. C., Busch, M. A., and Busch, K. W., Anal. Chem., 1989,61,2785. Analytical Methods Committee, Analyst, 1987, 112, 199. van der Linden, W. E., Anal. Chim. Acta, 1983, 151, 359. Hollowell, D. A,, Pacey, G. E., and Gordon, G., Anal. Chem., 1985,57,2851. Paper 01028641 Received June 26th, 1990 Accepted August lst, 1990