- Email: [email protected]
Unsegmented flow analysis with ion-selective electrodes by use of a large-volume wall-jet cell Continuous electrode reactivation in the determination of fluoride and chloride
Talonto, Vol. 38, No. 12, pp. 139S1398, 1991 Printed in Great Britain. All rights reserved
Copyright 0
0039-9140/91 $5.00 + 0.00 1991 Pergamon Press plc
UNSEGMENTED FLOW ANALYSIS WITH ION-SELECTIVE ELECTRODES BY USE OF A LARGE-VOLUME WALL-JET CELL CONTINUOUS ELECTRODE REACTIVATION IN THE DETERMINATION OF FLUORIDE AND CHLORIDE J&f LEXA Chemical Laboratories, Central Research Institute, Skoda Works, 316 00 P&en, Czechoslovakia
Department of Analytical Chemistry, Charles University, Albertov 2030, 128 40 Prague 2, Czechoslovakia (Received 30 July 1990. Revised 3 June 1991. Accepted 11 Jzme 1991)
Summary-A simple, large-volume wall-jet cell was designed for unsegmented flow analysis. The working electrode is immersed in a solution that reactivates the electrode surface. During the sample measurement, the working electrode is screened from the reactivation solution by the streaming sample solution; between the individual samples, air is pumped through the jet and agitates the reactivating solution at the electrode surface. The properties of the cell were investigated with the fluoride and chloride ion-selective electrodes and the method was applied to determination of fluoride and chloride in steel corrosion products and in a reference sample of the fly dust from electric-arc furnaces. The method permits about 90 measurements per hour, the results are reproducible and the limits of determination (6.3 x lo-* and 2.3 x lo-‘M for fluoride and chloride, respectively) are substantially lower than the values commonly obtained in batch experiments. At least 1.50measurements can be carried out without significant changes in the reliability of determination and the reactivation solution can then be replaced.
Various techniques of flow analysis have rapidly gained importance, as they exhibit many advantages that are thoroughly discussed in the literature.1*2 In addition to the most common spectrophotometric detection, some electrochemical methods are being progressively used more (primarily potentiometry, amperometry and coulometry), in view of their high sensitivity and selectivity that can, to a certain extent, be varied by judicious control of the experimental conditions.3 Potentiometry with ion-selective electrodes (ISE’s) is very popular because it is highly selective and simple. However, problems may arise in flow detection due to rather sluggish response of many BE’s which, to make it worse, is dependent on the analyte activity (the response time generally increases with decreasing activity). M Another difficulty, common to all the electrochemical methods except for high-frequency impedance measurements, stems from the interactions between the electrode surface and the test solution and leads to electrode passivation, memory effects and enhanced noise.3,6 TN.
3*,1*--D
On the other hand, an advantage of flow measurements with BE’s lies in the fact that the measuring sensitivity is often somewhat higher than in batch experiments6 and that many operations required for the sample pretreatment and electrode reactivation are easy to carry out. The cells for flow potentiometry are mostly low-volume thin-layer, wall-jet and tubular cells or open wall-jet systems with the working electrode placed above a solution containing a reference electrode.3 All these cells often suffer from noise and random interruptions to the electric circuit, due to air bubbles trapped in small cells or breakage of the thin film of liquid connecting the working electrode with the solution containing the reference electrode in the open systems. Moreover, if the working electrode is to be reactivated, it must either be taken out of the cell, or a different solution must be pumped through the cell, which is awkward and time consuming. A large-volume wall-jet cell, in which the test solution is fed from a jet onto the working electrode surface and the whole system is placed
1393
JIIU LEXAand KARELSTIJL~K
1394
in a large vessel containing a suitable solution, was first described by Horvai et al.’ and later studied in detail by Gunasingham et aL8-” The detector has primarily been used for amperometric detection in HPLCal’ and recently also for flow-injection potentiometric stripping analysis,” flow-injection potentiometry13-l5 and potentiometric detection in ion chromatography.16s’7 It has been shown”” that the liquid beam coming from the jet remains intact even at lo-mm distances of the jet orifice to the electrode surface, that the effective working volume of_the cell is small (less than the volume of the hydrodynamic boundary layer at the electrode surface) and that the impinging test solution shields the electrode from the effects of the liquid in the cell, so that it is, e.g., possible to perform amperometric reductive measurements without removal of oxygen from the solution in the cell. The present paper describes a large-volume wall-jet cell of a very simple design and its application to unsegmented flow ISE potentiometry. The favourable properties of the cell are utilized for continuous reactivation of the working electrode surface and the method is applied to determination of fluoride and chloride in the products of steel corrosion and in a reference sample of fly dust from electric-arc furnaces. EXPERIMENTAL
Apparatus
The potentiometric measurements were performed with an ION-85 instrument and a REC80 chart recorder with a REA-120 module (all from Radiometer, Copenhagen, Denmark). The fluoride and Chloride ISE’s and the saturated silver/silver chloride reference electrode were Crytur types 09-27, 17-27 and RAE-l 12, reobtained from Monokrystaly, spectively, Turnov, Czechoslovakia. In order to suppress the noise caused by the high impedance of the chloride ISE, an impedance convertor based on a WSH-220 operational amplifier (Tesla, LanBkroun, Czechoslovakia) was placed inside the electrode body.” The test solutions were propelled by a Type 3 15 peristaltic pump (Zalimp, Warsaw, Poland). In the measurement of the detector time constant dependence on the solution flow-rate, an ABU-80 autoburette (Radiometer, Copenhagen, Denmark) was used for pumping.
The sample solutions were aspirated from polypropylene beakers and the sample size was controlled by controlling the aspiration time. The solutions were aspirated through a PTFE tubing (1 mm in internal diameter) and led to the detection cell by silicone rubber tubing (40 cm long and 2 mm in internal diameter). Unless stated otherwise, the flow-rate was 3 ml/min. The detection cell is schematically depicted in Fig. 1. The cell consists of a cylindrical glass vessel (1) with an opening for the working electrode (2) and reference (3) electrode and for the jet (4) which are fixed in place by polyethylene O-rings. The horizontal position of the ISE and the jet prevents gas bubbles from adhering to the ISE surface. The polyethylene jet is 0.3 mm in internal diameter and its distance from the ISE can be varied. The vessel is provided with a solution outlet stop-cock (5) and is placed in the neck of a waste bottle (7). Tube (6) maintains a constant solution level (a solution volume of 50 ml). Reagents
Analytical-grade chemicals were employed as received from Lachema, Bmo, Czechoslovakia. Stock solutions of the analytes, O.lM, were prepared by dissolving solid sodium fluoride and sodium chloride in water, and used as primary standards. Buffer solution A [0.5M in sodium nitrate, acetic acid and sodium acetate and O.lM in diaminocyclohexanetetraacetic acid (DCTA)], pH 4.3, was prepared as follows. A 36-g amount of DCTA was dissolved in ca. 500 ml of an aqueous solution of 15 g of sodium hydroxide,
Fig. 1. The large-volume wall-jet cell (for description see test).
Unsegmentedflow analysiswith ion-selectiveelectrodes then 43 g of sodium nitrate and 30 ml of acetic acid (p = 1.060) were added and the mixture was made up with water to 1000 ml. Buffer solution B (pH 4.3) was 0.5M in sodium nitrate, acetic acid and sodium acetate, and was prepared by dissolving 6.0 g of sodium hydroxide in water, adding 30 ml of acetic acid (p = 1.060) and 43 g of sodium nitrate, and diluting with water to 1000 ml. Solutions A and B were filtered before use. Buffer solutions A and B were mixed with the sample solutions at a volume ratio of 1: 5. A reactivation solution, 0.M in A13+ and nitric acid, was prepared by dissolving 3.3 g of Al,(SO,),. 18H,O in 100 ml of O.lM nitric acid. General measuring procedure
The detection cell is prepared for measurement by directing the jet to the centre of the working electrode and adjusting its distance from the electrode. The jet is placed excentritally in a plastic stopper, so that its proper position is attained by turning the stopper, while pumping water onto the electrode surface through the jet in the empty vessel. The vessel is then filled with 50 ml of an appropriate reactivation solution and the ISE potential is allowed to stabilize. The test solution is pumped for 5-10 set, thus obtaining sample solution volumes from 0.25 to 0.5 ml at a flow-rate of 3 ml/min. When a sample is not aspirated, air is pumped through the jet and the reactivation solution is thus stirred at the electrode, hastening the reactivation process. Procedure for determination of fluoride and chloride in the steel corrosion products and in the j?y dust from electric-arc furnaces
A sample of lo-25 mg is fused in a platinum crucible with 0.5 g of sodium carbonate; 20 mg of silica may be added to assist the sample decomposition. The melt is extracted with 5 ml of cold water in an ultrasonic bath, the extract acidified with 2 ml of 5M nitric acid, degassed in an ultrasonic bath and diluted with water to 10 ml. A part of this solution is measured in a polyethylene beaker and diluted with water to 20 ml. A volume of 5 ml of buffer A (determination of fluoride) or buf?‘er B (determination of chloride) is added to the beaker, the mixture is stirred and the determination is carried out according to the above measuring procedure. The vessel contains the appropriate
1395
reactivation solution, i.e., 0.M A13++ O.lM HNO, (fluoride) or O.lM NaNO,+O.lM CH, COOH + 0.1M CH, COONa + 0.02M DCTA, pH 4.3 (chloride). The samples are pumped for 10 see at a flow-rate of 3 ml/min; the reactivation interval is 30 sec. RESULTS AND DISCUSSION So far, when the activity of a measuring ISE in a flow cell changed, the electrode was either removed from the cell and reactivated by a suitable mechanical or chemical procedure, or a chemical reactivation was carried out directly in the cell, by passage of a reactivating solution. In both cases, the sequence of analytical measurements had to be interrupted. However, the favourable properties of the large-volume walljet cell, primarily the fact that the solution impinging on the electrode efficiently separates the electrode surface from the bulk solution in the cell, make it possible to continuously reactivate the electrode by placing the reactivating solution in the cell. During passage of a sample solution, the reactivating solution has no access to the electrode surface; between samples, air is pumped through the jet and stirs the reactivating solution at the electrode. The most efficient way of reactivating solidmembrane ISE’s is removal of a very thin layer of the electrode surface, either by mechanical polishing or chemical etching. In this work, complexation reactions are used for the purpose-an acidic solution of aluminium ions for the fluoride ISE and a weakly acidic solution of DCTA for the chloride ISE. The action of these solutions is efficient but gentle and no mechanical damage is done to the electrode surface.
Basic properties of the cell
The results are in agreement with those published earlier.p” The optimal distance of the jet orifice from the electrode surfaces lies within an interval of 2-12 mm. The higher the flow-rate, the greater is the optimal distance. At distances shorter than 2 mm, the noise sharply increases; at distances longer than ca. 12 mm, the beam of the liquid coming from the jet ceases to be intact. For the flow-rate used in further experiments, 3 ml/r&, a distance of 3-6 mm was found optimal and was maintained within this interval throughout. At this or greater distance, there was also no danger of air bubbles being trapped between the jet and the electrode.
Jti LEXAand KAREL~JL~K
1396
Table 1. Dependence of the time constant, T,., of the detector with the chloride ISE on the flow-rate of the test solution
PH
T k, see
Flow rate, mllmin
1 2 3 4 6 8
I.1
0
33
4.60 1.93 1.14 0.73 0.55 0.40
The T, value is the time required for the signal to increase to 63.2% of the maximum value on a step change in the chloride concentration; the test solution, 0.M HNO, + 2.5 x 10-j or 2.5 x lo-‘M Cl-; the cell solution, O.lM HNO,.
As can be expected, the response time decreases with increasing flow-rate (for the dependence with the chloride ISE see Table 1). From the points of view of electrode response and sample consumption, a flow-rate of 3 ml/mm is a useful compromise. Determination of fluoride and chloride
Whereas the response rate of the chloride ISE is independent of the solution pH from 1.l to 4.6 and of the chloride concentration within the range of 2.5 x 10e4-2.5 x lo-‘M Cl-, the response rate of the fluoride ISE increases with increasing concentration of fluoride and decreasing pH (see Table 2). Moreover, the linear dynamic range increases with decreasing pH (see Fig. 2). However, measurement in strongly acidic solutions is inconvenient because the potential stabilization is slow and the selectivity of measurement is poorer. Therefore, it is better to work at a pH between 4 and 5 (we further used
6.0
-20
T k,
pgl25 ml 1.9 19 1.9 19 1.9 19
see 4.8 1.5
6.2
4.6 \-
L
Fig. 2. The effect of the test solution pH on the fluoride calibration plot. The test and the cell solution were identical: O.lM sulphosalicyclic acid + 0.02M EDTA with the pH adjusted with an NaOH solution, or a O.lM HCl @H 1.1).
a pH of 4.3), and add aluminium(II1) ions to the cell solution to enhance the electrode reactivation (O.lM HN03 + O.lM Al’+). We further added DCTA to the sample solution, to improve the measuring selectivity.6 The calibration dependences obtained under these conditions are given in Fig. 3. It can be seen that the slope of the dependence is higher for the flow measurement (68 mV per concentration decade) than for the batch measurements under identical conditions (60 mV per concentration decade). The reason for this phenomenon is not only a somewhat higher sensitivity of measurement attained under flow conditions, but also the fact that a nonsteady140 t 120
cr- ,
56
PF-
Table 2. Dependence of the time constant, T,, of the detector with the fluoride ISE on the fluoride concentration and the solution pH PH 6.7 6.7 4.6 4.6 1.1 1.1
0
ii!
loo
60
w-60 a
40
::: ;I
For the detinition of Tk see Table 1; the test and the cell solutions were identical, either 0.1M HCl (pH 1. l), or 0.M NaCl +O.OSM CH,COOH with the pH adjusted between 4.6 and 6.7 by additions of NaOH.
%
-20
L
. PF-
Fig. 3. The calibration plots for fluoride obtained in flow (1, 2) and batch (3) experiments. Buffer solution A; reactivation solution of 0.M Al)+ and HNO,.
Unsegmented flow analysis with ion-selective electrodes
state signal is measured in the flow system and that the electrode response is slower at low analyte concentrations. Therefore, the values obtained at low fluoride concentrations are further removed from the respective steady-state values than those obtained at higher analyte concentrations. The hysteresis region in the flow measurements is wider at low analyte concentrations (ca. 6 mV) and it only occurs in the presence of DCTA (this confirms the earlier observations that DCTA makes the response of the fluoride ISE slowed). Typical recordings of repeated flow determinations of fluoride are given in Fig. 4. It can be seen that the baseline is usually subject to a drift and thus it is more convenient to measure the difference between the peaks corresponding to the highest and the actual analyte concentration. Hence the signal for the highest analyte concentration defines the working baseline, which is stable for several hours. The limit of determination was determined as follows: A regression straight line was constructed for a narrow range of the lowest concentrations (O-4 x 10e7 and O-4 x 10m6 for fluoride and chloride, respectively, and the value was found on this line that corresponded to triple the standard deviation of the intercept divided by the straight line slope. The limit of determination for fluoride under the above flow conditions is 1.2 pg/l. (6.3 x lo-*M), which is a value about 30 times lower than that obtained in common batch measurements.‘g An
9
I
40 mV
4 min
Fig. 4. A recording of a series of flow determinations of fluoride. Measurements were always made in teplicate (for conditions see Fig. 3 and text). cF - (~g/25 ml): 0 (l), 0.095 (2), 0.19 (3), 0.38 (4), 0.95 (5), 1.9 (a), 3.8 (7), 9.5 (8) and 19 (9).
70
1397
r
Or”’ ’ ’ ’ 56
5.2
48 pc1-
44
4.0
-10
Fig. 5. The calibration plots for chloride. Flow measurement with (1) and without (2) DCTA (for conditions see text).
analogous situation is with the determination of chloride, where the limit of determination equals 8 pg/l. (2.3 x 10e7), about 14 times lower than in common batch measurements.‘g In measurements with the chloride ISE, the optimal pH range lies between 2.5 and 4.5, as expected and thus a solution of buffer B was selected for the measurement. The electrode can be reactivated in the blank sample solution, but somewhat better results are obtained when this solution is made 0.02M in DCTA. The slope of the calibration plot (Fig. 5) then slightly increases (from 61.4 mV per concentration decade in the absence of DCTA to 62.7 mV in its presence) and the linear correlation coefficient is somewhat improved (from 0.9996 to 0.9999). The cleaning effect of DCTA probably depends on the complexation reaction with the Ag+ ions on the surface of the membrane. Moreover, DCTA improves the selectivity of determination, similar to the determination of fluoride. Rather high concentrations of the reactivation solutions were used to prevent rapid loss of their function on dilution with the samples. Under the given experimental conditions, at least 150 measurements could be carried out without significant deterioration in the measuring reliability. The reactivation solution was then replaced. The results of the application of the flowthrough method to determination of fluoride and chloride in steel corrosion products and in a reference sample of an electric-arc furnace fly
JrRl LEXAand KAROL!h~fK
1398
Table 3. Determination of fluoride and chloride in the steel corrosion products and in a reference sample of the electric-arc furnace fly dust (for procedures see Experimental)
Decomposition procedure
Sample Corrosion products I Corrosion products II Fly dust
without SiO, without SiO, without SiO, with SiOr _
Fluoride content fi) Confidence interval 0.017 f 0.003 0.053 f 0.005 0.26 f 0.04 0.24 & 0.04
dust are given in Table 3 (for the procedures see Experimental). The determination of chloride was compared with the results obtained by a turbidimetric method20 employing a surfactant for stabilization of the turbidity and the agreement was satisfactory (see Table 3), except for the procedure in which Si02 was added to the fusion mixture. The turbidimetric determination is, of course, much more sensitive to colouration of the test solution and to cloudiness, which is caused in this case by a high content of hydrated silicon dioxide. The results indicate that the method yields good results and permits about 90 individual measurements per hour. There was no suitable reference method available for the determination of fluoride. REFERENCES
1. J. Rt%ieka and E. H. Hansen, Flow Injection Analysis, 2nd Ed., Wiley, New York, 1988. 2. M. Val&cel and M. D. Luque de Castro, Flow Injection Analysis. Principles and Applications, Honvood, Chichester, 1987. 3. K. Stulik and V. Paclkovi, Electroanalytical Measurements in Flowing Liquids, Horwood, Chichester 1987.
Chloride content (%I Confidence n interval
n
3 3 4 4
3 4 3 3
0.21 * 1.29 f 1.35 + 1.39 f
0.04 0.04 0.17 0.17
Turbidimetry 0.28 f 1.3 f 1.5 f 0.43 f
0.05 0.03 0.08 0.10
4. J. Koryta and K. Stulik, Ion Selective Electrodes, 2nd Ed., Cambridge University Press, 1983. 5. P. PetPk and K. Stulik, Anal. Chim. Acta, 1986, 185, 171. 6. J. Vesely, D. Weiss and K. Stulik, Analysis with IonSelective Electrodes, Honvood, Chichester, 1978. G. Horvai, K. T6th, J. Fekete and E. Pungor, Euroanalysis IV, Helsinki, 1981. 8. H. Gunasingham and B. Fleet, Anal. Chem., 1983, 55, 1409. 9. H. Gunasingham, B. T. Tay, K. P. Ang and L. L. Koh, J. Chromatog., 1984, 285, 103. 10. H. Gunasingham, Anal. Chim. Acta, 1984, 159, 139. 11. Z. Niegeisz, L. Szticz, J. Fekete, G. Horvai, K. T6th and E. Pungor, J. Chromatog, 1984, 316,451. 12. W. Matuszewski, M. Trojanowicz and W. Frenzel, Z. Anal. Chem., 1988, 332, 148. 13. L. Ilcheva, M. Trojanowicz and T. Krawczynski vel Krawczyk, ibid., 1987, 328, 27. 14. M. Trojanowicz and W. Frenzel, ibid., 1987, 328, 653. 15. W. Frenzel, Analyst, 1988, 113, 1039. 16. M. Trojanowicz and M. E. Mayerhoff, Anal. Chem., 1989, 61, 787. 17. Idem, Anal. Chim. Acta, 1989, 222, 95. 18. J. Langmaier, K. Stulik and R. Kalvoda, ibid., 1983, 148, 19. 19. J. Lexa and K. Stulik, Chem. L&y, 1988, 82, 1287. 20. J. Lexa and J. Ligka, Czech Pat. Appl. PV 4021-88, 1988: A0 270 057.
Copyright 0
0039-9140/91 $5.00 + 0.00 1991 Pergamon Press plc
UNSEGMENTED FLOW ANALYSIS WITH ION-SELECTIVE ELECTRODES BY USE OF A LARGE-VOLUME WALL-JET CELL CONTINUOUS ELECTRODE REACTIVATION IN THE DETERMINATION OF FLUORIDE AND CHLORIDE J&f LEXA Chemical Laboratories, Central Research Institute, Skoda Works, 316 00 P&en, Czechoslovakia
Department of Analytical Chemistry, Charles University, Albertov 2030, 128 40 Prague 2, Czechoslovakia (Received 30 July 1990. Revised 3 June 1991. Accepted 11 Jzme 1991)
Summary-A simple, large-volume wall-jet cell was designed for unsegmented flow analysis. The working electrode is immersed in a solution that reactivates the electrode surface. During the sample measurement, the working electrode is screened from the reactivation solution by the streaming sample solution; between the individual samples, air is pumped through the jet and agitates the reactivating solution at the electrode surface. The properties of the cell were investigated with the fluoride and chloride ion-selective electrodes and the method was applied to determination of fluoride and chloride in steel corrosion products and in a reference sample of the fly dust from electric-arc furnaces. The method permits about 90 measurements per hour, the results are reproducible and the limits of determination (6.3 x lo-* and 2.3 x lo-‘M for fluoride and chloride, respectively) are substantially lower than the values commonly obtained in batch experiments. At least 1.50measurements can be carried out without significant changes in the reliability of determination and the reactivation solution can then be replaced.
Various techniques of flow analysis have rapidly gained importance, as they exhibit many advantages that are thoroughly discussed in the literature.1*2 In addition to the most common spectrophotometric detection, some electrochemical methods are being progressively used more (primarily potentiometry, amperometry and coulometry), in view of their high sensitivity and selectivity that can, to a certain extent, be varied by judicious control of the experimental conditions.3 Potentiometry with ion-selective electrodes (ISE’s) is very popular because it is highly selective and simple. However, problems may arise in flow detection due to rather sluggish response of many BE’s which, to make it worse, is dependent on the analyte activity (the response time generally increases with decreasing activity). M Another difficulty, common to all the electrochemical methods except for high-frequency impedance measurements, stems from the interactions between the electrode surface and the test solution and leads to electrode passivation, memory effects and enhanced noise.3,6 TN.
3*,1*--D
On the other hand, an advantage of flow measurements with BE’s lies in the fact that the measuring sensitivity is often somewhat higher than in batch experiments6 and that many operations required for the sample pretreatment and electrode reactivation are easy to carry out. The cells for flow potentiometry are mostly low-volume thin-layer, wall-jet and tubular cells or open wall-jet systems with the working electrode placed above a solution containing a reference electrode.3 All these cells often suffer from noise and random interruptions to the electric circuit, due to air bubbles trapped in small cells or breakage of the thin film of liquid connecting the working electrode with the solution containing the reference electrode in the open systems. Moreover, if the working electrode is to be reactivated, it must either be taken out of the cell, or a different solution must be pumped through the cell, which is awkward and time consuming. A large-volume wall-jet cell, in which the test solution is fed from a jet onto the working electrode surface and the whole system is placed
1393
JIIU LEXAand KARELSTIJL~K
1394
in a large vessel containing a suitable solution, was first described by Horvai et al.’ and later studied in detail by Gunasingham et aL8-” The detector has primarily been used for amperometric detection in HPLCal’ and recently also for flow-injection potentiometric stripping analysis,” flow-injection potentiometry13-l5 and potentiometric detection in ion chromatography.16s’7 It has been shown”” that the liquid beam coming from the jet remains intact even at lo-mm distances of the jet orifice to the electrode surface, that the effective working volume of_the cell is small (less than the volume of the hydrodynamic boundary layer at the electrode surface) and that the impinging test solution shields the electrode from the effects of the liquid in the cell, so that it is, e.g., possible to perform amperometric reductive measurements without removal of oxygen from the solution in the cell. The present paper describes a large-volume wall-jet cell of a very simple design and its application to unsegmented flow ISE potentiometry. The favourable properties of the cell are utilized for continuous reactivation of the working electrode surface and the method is applied to determination of fluoride and chloride in the products of steel corrosion and in a reference sample of fly dust from electric-arc furnaces. EXPERIMENTAL
Apparatus
The potentiometric measurements were performed with an ION-85 instrument and a REC80 chart recorder with a REA-120 module (all from Radiometer, Copenhagen, Denmark). The fluoride and Chloride ISE’s and the saturated silver/silver chloride reference electrode were Crytur types 09-27, 17-27 and RAE-l 12, reobtained from Monokrystaly, spectively, Turnov, Czechoslovakia. In order to suppress the noise caused by the high impedance of the chloride ISE, an impedance convertor based on a WSH-220 operational amplifier (Tesla, LanBkroun, Czechoslovakia) was placed inside the electrode body.” The test solutions were propelled by a Type 3 15 peristaltic pump (Zalimp, Warsaw, Poland). In the measurement of the detector time constant dependence on the solution flow-rate, an ABU-80 autoburette (Radiometer, Copenhagen, Denmark) was used for pumping.
The sample solutions were aspirated from polypropylene beakers and the sample size was controlled by controlling the aspiration time. The solutions were aspirated through a PTFE tubing (1 mm in internal diameter) and led to the detection cell by silicone rubber tubing (40 cm long and 2 mm in internal diameter). Unless stated otherwise, the flow-rate was 3 ml/min. The detection cell is schematically depicted in Fig. 1. The cell consists of a cylindrical glass vessel (1) with an opening for the working electrode (2) and reference (3) electrode and for the jet (4) which are fixed in place by polyethylene O-rings. The horizontal position of the ISE and the jet prevents gas bubbles from adhering to the ISE surface. The polyethylene jet is 0.3 mm in internal diameter and its distance from the ISE can be varied. The vessel is provided with a solution outlet stop-cock (5) and is placed in the neck of a waste bottle (7). Tube (6) maintains a constant solution level (a solution volume of 50 ml). Reagents
Analytical-grade chemicals were employed as received from Lachema, Bmo, Czechoslovakia. Stock solutions of the analytes, O.lM, were prepared by dissolving solid sodium fluoride and sodium chloride in water, and used as primary standards. Buffer solution A [0.5M in sodium nitrate, acetic acid and sodium acetate and O.lM in diaminocyclohexanetetraacetic acid (DCTA)], pH 4.3, was prepared as follows. A 36-g amount of DCTA was dissolved in ca. 500 ml of an aqueous solution of 15 g of sodium hydroxide,
Fig. 1. The large-volume wall-jet cell (for description see test).
Unsegmentedflow analysiswith ion-selectiveelectrodes then 43 g of sodium nitrate and 30 ml of acetic acid (p = 1.060) were added and the mixture was made up with water to 1000 ml. Buffer solution B (pH 4.3) was 0.5M in sodium nitrate, acetic acid and sodium acetate, and was prepared by dissolving 6.0 g of sodium hydroxide in water, adding 30 ml of acetic acid (p = 1.060) and 43 g of sodium nitrate, and diluting with water to 1000 ml. Solutions A and B were filtered before use. Buffer solutions A and B were mixed with the sample solutions at a volume ratio of 1: 5. A reactivation solution, 0.M in A13+ and nitric acid, was prepared by dissolving 3.3 g of Al,(SO,),. 18H,O in 100 ml of O.lM nitric acid. General measuring procedure
The detection cell is prepared for measurement by directing the jet to the centre of the working electrode and adjusting its distance from the electrode. The jet is placed excentritally in a plastic stopper, so that its proper position is attained by turning the stopper, while pumping water onto the electrode surface through the jet in the empty vessel. The vessel is then filled with 50 ml of an appropriate reactivation solution and the ISE potential is allowed to stabilize. The test solution is pumped for 5-10 set, thus obtaining sample solution volumes from 0.25 to 0.5 ml at a flow-rate of 3 ml/min. When a sample is not aspirated, air is pumped through the jet and the reactivation solution is thus stirred at the electrode, hastening the reactivation process. Procedure for determination of fluoride and chloride in the steel corrosion products and in the j?y dust from electric-arc furnaces
A sample of lo-25 mg is fused in a platinum crucible with 0.5 g of sodium carbonate; 20 mg of silica may be added to assist the sample decomposition. The melt is extracted with 5 ml of cold water in an ultrasonic bath, the extract acidified with 2 ml of 5M nitric acid, degassed in an ultrasonic bath and diluted with water to 10 ml. A part of this solution is measured in a polyethylene beaker and diluted with water to 20 ml. A volume of 5 ml of buffer A (determination of fluoride) or buf?‘er B (determination of chloride) is added to the beaker, the mixture is stirred and the determination is carried out according to the above measuring procedure. The vessel contains the appropriate
1395
reactivation solution, i.e., 0.M A13++ O.lM HNO, (fluoride) or O.lM NaNO,+O.lM CH, COOH + 0.1M CH, COONa + 0.02M DCTA, pH 4.3 (chloride). The samples are pumped for 10 see at a flow-rate of 3 ml/min; the reactivation interval is 30 sec. RESULTS AND DISCUSSION So far, when the activity of a measuring ISE in a flow cell changed, the electrode was either removed from the cell and reactivated by a suitable mechanical or chemical procedure, or a chemical reactivation was carried out directly in the cell, by passage of a reactivating solution. In both cases, the sequence of analytical measurements had to be interrupted. However, the favourable properties of the large-volume walljet cell, primarily the fact that the solution impinging on the electrode efficiently separates the electrode surface from the bulk solution in the cell, make it possible to continuously reactivate the electrode by placing the reactivating solution in the cell. During passage of a sample solution, the reactivating solution has no access to the electrode surface; between samples, air is pumped through the jet and stirs the reactivating solution at the electrode. The most efficient way of reactivating solidmembrane ISE’s is removal of a very thin layer of the electrode surface, either by mechanical polishing or chemical etching. In this work, complexation reactions are used for the purpose-an acidic solution of aluminium ions for the fluoride ISE and a weakly acidic solution of DCTA for the chloride ISE. The action of these solutions is efficient but gentle and no mechanical damage is done to the electrode surface.
Basic properties of the cell
The results are in agreement with those published earlier.p” The optimal distance of the jet orifice from the electrode surfaces lies within an interval of 2-12 mm. The higher the flow-rate, the greater is the optimal distance. At distances shorter than 2 mm, the noise sharply increases; at distances longer than ca. 12 mm, the beam of the liquid coming from the jet ceases to be intact. For the flow-rate used in further experiments, 3 ml/r&, a distance of 3-6 mm was found optimal and was maintained within this interval throughout. At this or greater distance, there was also no danger of air bubbles being trapped between the jet and the electrode.
Jti LEXAand KAREL~JL~K
1396
Table 1. Dependence of the time constant, T,., of the detector with the chloride ISE on the flow-rate of the test solution
PH
T k, see
Flow rate, mllmin
1 2 3 4 6 8
I.1
0
33
4.60 1.93 1.14 0.73 0.55 0.40
The T, value is the time required for the signal to increase to 63.2% of the maximum value on a step change in the chloride concentration; the test solution, 0.M HNO, + 2.5 x 10-j or 2.5 x lo-‘M Cl-; the cell solution, O.lM HNO,.
As can be expected, the response time decreases with increasing flow-rate (for the dependence with the chloride ISE see Table 1). From the points of view of electrode response and sample consumption, a flow-rate of 3 ml/mm is a useful compromise. Determination of fluoride and chloride
Whereas the response rate of the chloride ISE is independent of the solution pH from 1.l to 4.6 and of the chloride concentration within the range of 2.5 x 10e4-2.5 x lo-‘M Cl-, the response rate of the fluoride ISE increases with increasing concentration of fluoride and decreasing pH (see Table 2). Moreover, the linear dynamic range increases with decreasing pH (see Fig. 2). However, measurement in strongly acidic solutions is inconvenient because the potential stabilization is slow and the selectivity of measurement is poorer. Therefore, it is better to work at a pH between 4 and 5 (we further used
6.0
-20
T k,
pgl25 ml 1.9 19 1.9 19 1.9 19
see 4.8 1.5
6.2
4.6 \-
L
Fig. 2. The effect of the test solution pH on the fluoride calibration plot. The test and the cell solution were identical: O.lM sulphosalicyclic acid + 0.02M EDTA with the pH adjusted with an NaOH solution, or a O.lM HCl @H 1.1).
a pH of 4.3), and add aluminium(II1) ions to the cell solution to enhance the electrode reactivation (O.lM HN03 + O.lM Al’+). We further added DCTA to the sample solution, to improve the measuring selectivity.6 The calibration dependences obtained under these conditions are given in Fig. 3. It can be seen that the slope of the dependence is higher for the flow measurement (68 mV per concentration decade) than for the batch measurements under identical conditions (60 mV per concentration decade). The reason for this phenomenon is not only a somewhat higher sensitivity of measurement attained under flow conditions, but also the fact that a nonsteady140 t 120
cr- ,
56
PF-
Table 2. Dependence of the time constant, T,, of the detector with the fluoride ISE on the fluoride concentration and the solution pH PH 6.7 6.7 4.6 4.6 1.1 1.1
0
ii!
loo
60
w-60 a
40
::: ;I
For the detinition of Tk see Table 1; the test and the cell solutions were identical, either 0.1M HCl (pH 1. l), or 0.M NaCl +O.OSM CH,COOH with the pH adjusted between 4.6 and 6.7 by additions of NaOH.
%
-20
L
. PF-
Fig. 3. The calibration plots for fluoride obtained in flow (1, 2) and batch (3) experiments. Buffer solution A; reactivation solution of 0.M Al)+ and HNO,.
Unsegmented flow analysis with ion-selective electrodes
state signal is measured in the flow system and that the electrode response is slower at low analyte concentrations. Therefore, the values obtained at low fluoride concentrations are further removed from the respective steady-state values than those obtained at higher analyte concentrations. The hysteresis region in the flow measurements is wider at low analyte concentrations (ca. 6 mV) and it only occurs in the presence of DCTA (this confirms the earlier observations that DCTA makes the response of the fluoride ISE slowed). Typical recordings of repeated flow determinations of fluoride are given in Fig. 4. It can be seen that the baseline is usually subject to a drift and thus it is more convenient to measure the difference between the peaks corresponding to the highest and the actual analyte concentration. Hence the signal for the highest analyte concentration defines the working baseline, which is stable for several hours. The limit of determination was determined as follows: A regression straight line was constructed for a narrow range of the lowest concentrations (O-4 x 10e7 and O-4 x 10m6 for fluoride and chloride, respectively, and the value was found on this line that corresponded to triple the standard deviation of the intercept divided by the straight line slope. The limit of determination for fluoride under the above flow conditions is 1.2 pg/l. (6.3 x lo-*M), which is a value about 30 times lower than that obtained in common batch measurements.‘g An
9
I
40 mV
4 min
Fig. 4. A recording of a series of flow determinations of fluoride. Measurements were always made in teplicate (for conditions see Fig. 3 and text). cF - (~g/25 ml): 0 (l), 0.095 (2), 0.19 (3), 0.38 (4), 0.95 (5), 1.9 (a), 3.8 (7), 9.5 (8) and 19 (9).
70
1397
r
Or”’ ’ ’ ’ 56
5.2
48 pc1-
44
4.0
-10
Fig. 5. The calibration plots for chloride. Flow measurement with (1) and without (2) DCTA (for conditions see text).
analogous situation is with the determination of chloride, where the limit of determination equals 8 pg/l. (2.3 x 10e7), about 14 times lower than in common batch measurements.‘g In measurements with the chloride ISE, the optimal pH range lies between 2.5 and 4.5, as expected and thus a solution of buffer B was selected for the measurement. The electrode can be reactivated in the blank sample solution, but somewhat better results are obtained when this solution is made 0.02M in DCTA. The slope of the calibration plot (Fig. 5) then slightly increases (from 61.4 mV per concentration decade in the absence of DCTA to 62.7 mV in its presence) and the linear correlation coefficient is somewhat improved (from 0.9996 to 0.9999). The cleaning effect of DCTA probably depends on the complexation reaction with the Ag+ ions on the surface of the membrane. Moreover, DCTA improves the selectivity of determination, similar to the determination of fluoride. Rather high concentrations of the reactivation solutions were used to prevent rapid loss of their function on dilution with the samples. Under the given experimental conditions, at least 150 measurements could be carried out without significant deterioration in the measuring reliability. The reactivation solution was then replaced. The results of the application of the flowthrough method to determination of fluoride and chloride in steel corrosion products and in a reference sample of an electric-arc furnace fly
JrRl LEXAand KAROL!h~fK
1398
Table 3. Determination of fluoride and chloride in the steel corrosion products and in a reference sample of the electric-arc furnace fly dust (for procedures see Experimental)
Decomposition procedure
Sample Corrosion products I Corrosion products II Fly dust
without SiO, without SiO, without SiO, with SiOr _
Fluoride content fi) Confidence interval 0.017 f 0.003 0.053 f 0.005 0.26 f 0.04 0.24 & 0.04
dust are given in Table 3 (for the procedures see Experimental). The determination of chloride was compared with the results obtained by a turbidimetric method20 employing a surfactant for stabilization of the turbidity and the agreement was satisfactory (see Table 3), except for the procedure in which Si02 was added to the fusion mixture. The turbidimetric determination is, of course, much more sensitive to colouration of the test solution and to cloudiness, which is caused in this case by a high content of hydrated silicon dioxide. The results indicate that the method yields good results and permits about 90 individual measurements per hour. There was no suitable reference method available for the determination of fluoride. REFERENCES
1. J. Rt%ieka and E. H. Hansen, Flow Injection Analysis, 2nd Ed., Wiley, New York, 1988. 2. M. Val&cel and M. D. Luque de Castro, Flow Injection Analysis. Principles and Applications, Honvood, Chichester, 1987. 3. K. Stulik and V. Paclkovi, Electroanalytical Measurements in Flowing Liquids, Horwood, Chichester 1987.
Chloride content (%I Confidence n interval
n
3 3 4 4
3 4 3 3
0.21 * 1.29 f 1.35 + 1.39 f
0.04 0.04 0.17 0.17
Turbidimetry 0.28 f 1.3 f 1.5 f 0.43 f
0.05 0.03 0.08 0.10
4. J. Koryta and K. Stulik, Ion Selective Electrodes, 2nd Ed., Cambridge University Press, 1983. 5. P. PetPk and K. Stulik, Anal. Chim. Acta, 1986, 185, 171. 6. J. Vesely, D. Weiss and K. Stulik, Analysis with IonSelective Electrodes, Honvood, Chichester, 1978. G. Horvai, K. T6th, J. Fekete and E. Pungor, Euroanalysis IV, Helsinki, 1981. 8. H. Gunasingham and B. Fleet, Anal. Chem., 1983, 55, 1409. 9. H. Gunasingham, B. T. Tay, K. P. Ang and L. L. Koh, J. Chromatog., 1984, 285, 103. 10. H. Gunasingham, Anal. Chim. Acta, 1984, 159, 139. 11. Z. Niegeisz, L. Szticz, J. Fekete, G. Horvai, K. T6th and E. Pungor, J. Chromatog, 1984, 316,451. 12. W. Matuszewski, M. Trojanowicz and W. Frenzel, Z. Anal. Chem., 1988, 332, 148. 13. L. Ilcheva, M. Trojanowicz and T. Krawczynski vel Krawczyk, ibid., 1987, 328, 27. 14. M. Trojanowicz and W. Frenzel, ibid., 1987, 328, 653. 15. W. Frenzel, Analyst, 1988, 113, 1039. 16. M. Trojanowicz and M. E. Mayerhoff, Anal. Chem., 1989, 61, 787. 17. Idem, Anal. Chim. Acta, 1989, 222, 95. 18. J. Langmaier, K. Stulik and R. Kalvoda, ibid., 1983, 148, 19. 19. J. Lexa and K. Stulik, Chem. L&y, 1988, 82, 1287. 20. J. Lexa and J. Ligka, Czech Pat. Appl. PV 4021-88, 1988: A0 270 057.