ANALYST FEBRUARY 1986 VOL. 111 151 Preparation of a Chloride-selective Electrode Based on Mercury(1) Chloride - Mercury(l1) Sulphide on an Electrically Conductive Epoxy support J. L. F. C. Lima and A. A. S. C. Machado* Chemistry Department Faculty of Science University of Oporto 4000 Oporto Portugal A chloride-selective electrode based on mercury salts on an electrically conductive epoxy support was prepared and tested. A 1 + 1 molar mixture of mercury(1) chloride and mercury(l1) sulphide was used as the sensor. Response characteristics (linear response range limit of detection slope stability and time of response pH range and potentiometric selectivity coefficients with respect to bromide iodide and thiocyanate) were determined. The standards of performance of the electrode were found t o be better than those of a commercial electrode with the same type of sensor.Reactions taking place in the membrane are discussed. Keywords Chloride-selective electrode; mercury salts; conductive epoxy support We have previously reported a simple and easy to implement procedure for the construction of inexpensive “all-solid-state” ion-selective electrodes in which a layer of very finely powdered sensor is applied to a base of electrically conductive (silver loaded) epoxy r e ~ i n . l - ~ Using this procedure elec-trodes for silver(1) and sulphide,l32 halides,’J copper(II)1.4 and other divalent cations,l with silver salts as sensors were obtained. Their response performances were found to be similar to those of the respective commercial electrodes.1-4 This work has been extended to include electrodes based on mercury salts as sensors. This type of sensor was introduced by Sekerka and Lechner,5-7 who found that an electrode containing a mixture of mercury(I1) sulphide and mercury(1) chloride as a sensor showed improved response characteristics to chloride compared with electrodes with a sensor based on a mixture of silver sulphide and silver chloride .578 The lower detection limit of this type of chloride-selective electrode has made its use very convenient for the determination of low levels of chloride in water.8-13 The performance of a commer-cial version of the electrode (Graphic Controls PHI 91100) has also been recently evaluated.14 The use of mixtures of mercury(I1) sulphide and mercury(1) chloride as sensors for the self-construction of selective electrodes has presented some difficulties,lOJ5 probably owing to the high pressures required to obtain membranes with good mechanical proper-ties.10 However Marshall and Midgley 11 successfully applied this type of sensor to the graphite surface of a RfiiiCka Selectrode. The purpose of this study was to investigate the response characteristics of a chloride-selective electrode with this type of sensor obtained by our procedure for the construction of “all-solid-state” selective electrodes which does not require high pressures for the preparation of the membrane. Experimental Apparatus Potentials were measured with an Orion 811 digital pH meter (reading to kO.1 mV) and an Orion 605 manual electode switch.Graphs for the determination of response times were obtained with a Radiometer PHM 64 pH meter and a Servograph Rec 61 plotter. A Philips GAH 110 glass electrode was used for the measurement of pH. Orion 90-02-00 double-junction elec-~~ * To whom correspondence should be addressed. trodes (of silver - silver chloride type) were used as reference electrodes (inner filling solution Orion 90-00-02; outer filling solution 10% potassium nitrate). Reagents The water used in the preparation of all the standard reagent solutions was de-ionised (Elgastat B114 mixed-bed column unit) and distilled in a quartz still (Heraeus B1 18 double distillation unit). All chemicals were of analytical-reagent grade and were used without further purification.When necessary stock solutions were standardised by potentio-metric titration. Further details are given elsewhere.14 Preparation of the Electrodes Mercury(I1) sulphide (black) was prepared by precipitation, initiated by slowly mixing equal volumes of 0.1 M sodium sulphide and 0.1 M mercury(I1) nitrate solutions. The sensor was prepared by thorough grinding of a 1 + 1 molar mixture of mercury( 11) sulphide and mercury(1) chloride (Merck Ref. No. 4425). The procedure used previously1 for the preparation of electrodes with silver salt sensors was followed. A piece of silver-loaded commercial epoxy (EPO-TEK 410) was applied to an end of Perspex tube (0.d. 1 cm length ca. 15 cm) to constitute a layer ‘of about 0.7 cm thickness; a shielded cable was fixed to the epoxy inside the tube and after hardening (at 100 “C for 1 h) a conical cavity was drilled in the epoxy layer.A new piece of epoxy was applied to this cavity and the very thinly powdered sensor was blown against it while still fresh from a Pasteur pipette this operation being repeated several times to obtain a continuous coat of sensor on the epoxy. After hardening (at 80 “C for 4 h) the other end of the tube was closed with Perspex glue. Finally the sensor layer was polished over glass (Wilks 004-10001) and then with polishing paper (Orion 94-82-01). Procedure for the Evaluation of the Electrodes Standard techniques were used for evaluating the response characteristics of the prepared electrodes. All the measure-ments were made with the electrodes immersed in solutions kept at 25.0k0.2 “C.Except where otherwise stated the ionic strength of the solutions was adjusted to 0.1 M and the pH to 3 with potassium nitrate and nitric acid 152 310 290 270 > E 250 230 210 190 ANALYST FEBRUARY 1986 VOL. 111 -------The slope S and standard potential EO were obtained from the experimental points in the linear range of calibration by a least-squares adjustment performed by standard pro-grams of pocket calculators. R is the correlation coefficient of linear regression given by the programs which measures the goodness of fit. Results Characteristics of Electrode Response to Chloride Reproducibility of preparation In order to assess the reproducibility of the preparation procedure and the stability of response (see below) five units were simultaneously calibrated daily with standard solutions of sodium chloride in the range 4 X 10-4-10-2 M for a period of more than 2 weeks.2-4 Between calibrations the electrodes were left in de-ionised water as this was found to be a suitable conditioning medium.(When not in use the electrodes were stored dry and in the dark and before their re-use were polished and conditioned.) Table 1 presents typical results of the calibrations obtained with one of the units (A). Table 2 gives the average values of the calibration parameters and their standard deviations for the five units (A-E). The results presented in Table 2 show that the procedure used for electrode construction yields units with reproducible response characteristics.Stability of response With respect to calibration graphs the electrodes retain their characteristics over several weeks (Table 1) without any need Table 1. Stability of the calibration graph* of an electrode unit?,$ E (10-3)/mv S/mV Time/d decade-' E'ImV R Calculated Read 1 1 2 2 3 3 8 8 9 9 10 10 18 18 -56.7 -57.0 -56.3 -56.3 -55.8 -54.9 -57.4 -57.5 -57.4 -57.5 -57.9 -57.3 -57.5 -57.6 40.0 39.6 40.2 39.8 40.2 40.0 36.8 36.8 36.6 36.9 35.4 36.0 35.2 34.5 0.9999 0.9999 0.9999 0.9999 0.9999 0.9999 0.9999 0.9999 0.9999 0.9999 0.9999 0.9999 0.9999 0.9999 210.0 209.5 209.2 208.7 207.4 204.6 209.2 209.2 208.8 209.5 209.0 208.0 207.7 207.4 210.0 209.3 208.9 208.3 207.3 204.3 209.0 209.2 208.9 209.5 208.9 208.1 207.4 207.5 * In the range 4 x 10-4-10-2 M chloride.t Unit A in Table 2. $ Symbols S = slope; E" = standard potential (vs. S.C.E.); = response M chloride; Calculated from calibration graph; Read direct R = correlation coefficient of linear regression; E to reading. Table 2. Average calibration parameters of several electrode units* ,? S/ Unit mV decade-' EOImV A 56.9 (0.8) 38 (2) B 57.8 (0.8) 33 (2) C 57.0 (1.0) 36 (2) D 56.0 (2.0) 37 (4) E 57.0 (2.0) 37 (4) * Averages of 14 calibrations over 18 d with standard deviations in t For symbols see Table 1. parentheses. for restoring the membrane by polishing provided that they are kept in water between measurements.The reproducibility of repeated calibrations obtained during the same day was generally found to better than 0.5 mV decade-' for the slope and 1 mV for the standard potential (Table 1). A decrease in slope was normally found after a month or more of use but the response characteristics could be restored by polishing the membrane with Orion paper followed by conditioning in water. This treatment was found to be very effective even when the membrane had been subjected to strong interferences. Under normal use the electrodes do not require frequent polishing and consequently although the membrane thick-ness is small (less than 1 mm) they have high durability. Several units have been used for the determination of chloride in high-purity waters for almost two years and still show good response characteristics.Such durability is a definite advan-tage over electrodes with the same type of sensors in pressed membranes which as mentioned by Tacussel and Fomboml5 and confirmed by ourselves in a previous evaluation of two units of the Graphic Controls electrode,14 show frequent periods of irregular response and require repeated polishing, which wears out the membrane. Lower limit of linear response and limit of detection Typical calibrations by the standard additions technique for the determination of these parameters are shown in Fig. 1. Values of ca. 10-5 and ca. 5 x 10-6 M for the lower limit of linear response and the limit of detection ,I6 respectively were obtained both being similar to the corresponding values found for a commercial electrode with the same type of sensor.14 Although less intense than for the commercial electrode a "memory effect" was found; for example if the electrode had been immersed in a 10-1 M chloride solution before calibration values of ca.10-4 M were found for the lower limit of linear response. The re-establishment of the normal value above for this parameter requires polishing followed by immersion in water for at least 30 min. The value for the lower limit of linear response falls within the range of values for the parameter reported in the literature (2 X 1W6-2 X 10-5 ~8,11,14,15) and extension of the linear range is characteristic of the mercury salt-based chloride electrodes compared with those based on silver salts.A 100-fold increase in linear response range has been reported5 for pressed membrane electrodes but previously with a commercial electrode with this type of membrane an approxi-mately 20-fold increase was found14 as in this work. An interesting feature shown by the calibration (Fig. 1) is the small variation of potential below the limit of linear 1.00 x 10-6 1.00 x 10-5 1.00 x 10-4 1.00 x 10-3 acl-Fig. 1. Typical calibration graph for the electrod ANALYST FEBRUARY 1986 VOL. 111 190 > E Li 220 153 . 2.32 x ' I 0 - 3 ~ 1.03 x 10-4M 4.54 x 10-4 M t-l 12.19 x 1 0 - 4 ~ I min 250 I 1.00 x 10-4M 280 ' Time -Fig. 2. Typical recorder output of dynamic response time determina-tions for varying concentrations of chloride response only about 20 mV in the decade from 10-6 to 10-5 M.A similar situation was found by us for the Graphic Controls electrode.14 In both situations the potential varia-tion is smaller than that shown by the electrode prepared by Marshall and Midgley11J2 by coating a Rfiiitka Selectrode with the same type of sensor. As a consequence in this instance the practical limit of detection is close to the lower limit of linear detection and the usefulness of the electrode below the latter limit is limited unless perhaps rigorously controlled conditions12 are applied. Slope The slope of ca. 57 mV decade-1 for the units studied (Table 2) was slightly lower than the theoretical value. Although Sekerka and Lechner5.8 reported a Nernstian response Tseng and Gutknechtl7 were unable to obtain the same slope for the electrodes prepared by a similar procedure and obtained slopes of 50-51 mV decade-1.However other worker~1~J1J5 have reported slopes in the range 57-59 mV decade-l for electrodes with the same type of sensor constructed by different procedures as was found by us in this and in previous work. 14 Standard potential Correcting the average value of this parameter for the units included in Table 2 +36 mV by +242 mV to express S.C.E. against N.H.E. and -7 mV ( S x logfcl-,fcl- = +0.755 at Z = 0.1 MIS) to compensate for the ionic strength a value of +271 mV vs. N.H.E. was obtained as the standard potential. This value is close to +268 mV the standard potential of the Hg2C12 - Hg system,19 suggesting that there is metallic mer-cury in the membrane to fix aHg at 120921 (see Discussion).Response time As shown in Fig. 2 the dynamic response time of the electrode for increases in chloride concentrations in the range 10-4-10-2 M is less than 1 min. Sekerka and Lechner5.8 quoted shorter response times for their electrode than for chloride-selective electrodes with silver salts as sensors but the present electrode and the corresponding one of the latter type3 showed similar response times. The electrode prepared by Marshall and Midgleyll by application of the sensor on a Rfiiitka Selectrode was also found to be slower than the pressed membrane electrode of Sekerka and Lechner.5~8 Our previous evaluation14 of the Graphic Controls commercial electrode did not show any improvement of response times over electrodes with silver salts as sensors.When the electrode is exposed to a decrease in chloride concentration response times are much longer especially if the initial value of concentration is high. This "memory effect" 340 300 260 220 > 5 180 UI 140 100 60 20 0 2.00 4.00 6.00 8.00 10.00 12.00 PH Fig. 3. Variation of response potential with pH for various concen-trations of chloride A 1.00 x 10-5 M; B 1.00 X 10-4 M; C 1.00 X M; D 1.00 x 10-2 M; and E 1.00 x lo-' M (cf. discussion under Lower limit of linear response and limit of detection) is much more pronounced for the electrode reported here than for the electrode based on silver salts. This has also been found for the Graphic Controls electrode,l4 based on a pressed membrane sensor and also for the electrode of Marshall and Midgley.11 As the degree of compactness of the sensor in the membrane is different for these three electrodes the effect cannot be ascribed only to surface irregularities.A possible explanation is a tendency for adsorption of chloride at the membrane surface which may be related to the high values of the stability constants of [HgCl,](n-2)- complexes. Indeed such values are much higher than those of the corresponding [AgCl]("-l)- species.22 It was observed that when this electrode was immersed in water for conditioning its potential suffered a quick change until a stable value was reached which may be due to the washing out of chloride from the surface of the membrane.Effect of pH on the Response The influence of the pH on the response of the electrode at fixed chloride concentrations between 10-1 and loe5 M (in solutions with an ionic strength adjusted to 0.2 M with potassium nitrate) is presented in Fig. 3. The potential is independent of pH from a lower limit of pH between 1.5 and 2 and an upper limit that depends much more markedly on the chloride concentration than for the hydroxide interference in other solid membrane electrodes e.g. for chloride-selective electrodes with silver salts as sensors.3 The marked variation of the upper limit of the operative response plateau can be understood when the value of KCI,OH = aC1-/aOH- is calculated from the solubility products of Hg2C12 and Hg20,14 the large value obtained (ca.3 x 102) explaining the observed variation. This feature of the chloride-selective electrode with an Hg2C12 - HgS sensor has not been discussed in the literature where a value of pH 6 is invariably indicated as the upper limit of operation.s.15.23 These results show that this value of pH is too high for the measurement of chloride at low concentrations (less than ca. 10-4 M) and support the pro-cedure established by Marshall and Midgleyl' where the hydrogen ion concentration is fixed at a constant value of M 154 ANALYST FEBRUARY 1986 VOL. 111 Table 3. Potentiometric selectivity coefficients gg:x This work* at chloride concentration Reference X 10-3 M 10-4 M Calculatedt 5 15 14$ 23 - SCN- 70 k 12 7 k 1 40 2 - 25 -Br- 70 k 11 43 k 6 2 x 104 6.3 x 102 - 102 -25 6 x 102 1- 30 k 9 5 + 1 3 x 10'0 3.2 x 103 -10 - 25 3 x 103 * Values are averages of six results (duplicate determinations with three units) obtained at the chloride concentrations given.t Calculated14 by K E F Kso(Hg2C12) - Kso(Hg2X2) using values of K, given in reference 27. $. Obtained at a M chloride concentration. A comparison of Fig. 3 with similar results for the Graphic Controls electrode14 shows that this electrode is more sensitive to pH changes below 2 which may be explained by the sensor being less compact when on the epoxy support than in a pressed membrane where solubilisation is more difficult. Interferences For this type of electrode there are discrepancies between literature values5J4J5J3 for the potentiometric selectivity coefficients relative to interferences of anions whose mer-cury(1) salts are more insoluble than mercury(1) chloride (namely bromide iodide and thiocyanate) as well as anomal-ies in their relative values found by some workers.14J5 In this study the coefficients were determined by the mixed solution method (with the chloride concentration fixed at 10-3 and 10-4 M without adjustment of the ionic strength but with the pH fixed at ca.4 M with nitric acid). The results of replicate determinations showed a certain degree of variability owing to the lack of reproducibility of the straight segments corre-sponding to response to interferences even when the experimental conditions (rate of interferent addition criterion of readings etc.) were kept constant.Therefore the experimental values presented in Table 3 which are averages of the results of six determinations should be considered only as orders of magnitude of the parameter. In Table 3 literature values are also included for comparison as well as values obtained from the solubility products which were calculated using standard procedures. The experimental values of the selectivity coefficients obtained for the electrode on a conductive epoxy support were much lower than those predicted by calculations. This result is similar to that found with electrodes with silver salts,lJ but the differences between the experimental and calculated values appear to be more accentuated in this instance. Moreover as found by Tacussel and Fombomls for the bromide and iodide interferences the relative strength of interferences found follows the order SCN- 3 Br- 3 I- which is opposite to the order predicted from calculations.In our previous study of the Graphic Controls electrode which for interferences was less detailed than this one about the same value was obtained for the selectivity coefficients with respect to the three interfer-ents.14 Tacussel and Fombomls have suggested that such anomalies have kinetic causes. Another interesting point shown by these results is that the values of the selectivity coefficients at the 10-4 M chloride level are smaller than the corresponding values at 10-3 M. The electrode seems to feel the effect of these interferences less extensively at lower concentrations and this may also be a consequence of slower response to lower concentrations of interferents.However it should be pointed out that in these experiments in contrast to observations in similar experi-ments with chloride-selective elctrodes with silver salts as sensors,3 visual inspection of the membrane did not show any strange precipitates on its surface. As the chemical behaviour of an Hg2C12 - HgS membrane seems to be extremely complex (see Discussion) an interference mechanism more complex than that accepted for electrodes with silver salts as sensors cannot be ruled out. Response to the Mercury(1) Ion Calibrations for the response to mercury(1) ion were obtained by the titration technique using lo-210-4 M mercury(1) nitrate solutions. These showed a slope slightly higher than the theoretical one 32.0 (0.5) mV decade-1 and a standard potential of 819 (2) mV vs.N.H.E. These are average values obtained from four determinations with three units the standard deviations being given in parentheses. The standard potential includes a correction of +13 mV for ionic strength compensation calculated as before with fHg2+ = 0.355 at Z = 0.1 ~ . 1 8 The value of the standard potential is similar to that found previously14 for the Graphic Controls electrode (813 mV) both being greater than the standard potential of the Hg2+ - Hg system (792 mV).19 The difference between the electrode standard potential and the standard potential of the relevant couple is larger in this instance than when the response to chloride is considered. Discussion Our previous work on ion-selective electrodes based on mixtures of silver salts of divalent metal sulphides and silver(1) sulphide on a conductive epoxy support has shown that this construction procedure yields electrodes with response characteristics similar to those of the corresponding commer-cial electrodes.1-4 There is indirect evidence that in these electrodes the metallic silver in the epoxy support does not contact the solution.374J4 This is also suggested by observation of the surface of the membranes by scanning electron microscopy which shows that in the small resin areas (ca.10-3-10-4 mm2) exposed between sensor microcrystals the silver signal is very weak (less than 1% of the signal of microcrystals and probably having this origin) .25 In this situation the operation of the constructed electrode was found to be less troublesome than for the commercial electrode evaluated previously,l4 even though the numerical values of the characteristic parameters of electrode response are very similar.Less frequent polishing for maintaining electrode performance was required and sudden outbreaks of bad behaviour of the electrode were not observed as for the Graphic Controls electrode. 14 Other problems of the chloride and other selective electrode with pressed membranes based on mercury salts e.g. irreproducibility of response15 or troublesome operation,26 have been discussed in the litera-ture. 1 0 ~ 5 ~ 7 ~ 2 6 The proposed method of construction minimises such problems. The values of the standard potentials of this electrode in response to chloride (+271 mV) and mercury(1) ions (+819 mV) are respectively close to the values for the Hg2C12 - Hg and Hg2+ - Hg couples and yield a value of 2.8 X 10-19 for the solubility product of Hg2C12 in agreement with literature values,27 e.g.1.3 x 10-18 ~ 3 . These data show that the electrode responds to chloride as a second kind electrode ANALYST FEBRUARY 1986 VOL. 111 According to Koebel20 and Buck and Shepard,21 this requires the occurrence of free mercury in the membrane to fix the value of the activity of the metal equal to unity. The presence of the free metal in the sensor is understood if the value of the equilibrium constant for the disproportiona-tion of Hg2C12 (s = solid sol = solution) is considered.It can be calculated from the solubility products of Hg2C12 (1.3 X 10-18) and HgS (black) (1.6 x lO-52)27 and the equilibrium constant of the disproportionation (Kdisp) by the expression &isp can be calculated from the standard potentials of the couples Hg2+ - Hg22+ (+907 mV) and Hg22+ - Hgo (+792 mV),19 to be &iSp = 10-1.95. A value of ca. 1032 is obtained for K. This value is so large that the reaction can occur even for large although reasonable values of chloride concentration in solution. Sulphide provided by the intrinsic solubility of the mercury(I1) sulphide is used up i.e. the reaction (1) occurs instead of the dissolution of this salt. Alternatively if the intrinsic solubility of Hg2CI2 is con-sidered [or when the electrode is immersed in a solution of mercury(1) ion] precipitation of free mercury at the mem-brane surface is explained by the reaction Hg2C12(~) + S’-(Sol) = Hg(1) +HgS(s) + 2C1- (sol) (1) Hg22+ (sol) = Hg(1) + Hg2+ (sol) .. (2) K = [CI-l2/[S2-] = &is, X Kso(Hg~C1~)IKso(HgS) (3) Hg22+(~01) + S2-(sol) = Hg(1) + HgS(s) . . (4) of which the equilibrium constant K’ = l/([Hg22+] X [S2-]) = Kdisp/Ks,(HgS) . . ( 5 ) is also very large (K‘ = 1050). These calculations show that the disproportionation of mercury(1) ion in the membrane is spontaneous in the thermodynamic sense and is expected to occur indefinitely with conversion of Hg2C12 into HgS (and free mercury) with the release of chloride into the solution. It is interesting to observe that Hulanicki et al.,26 in a paper discussing the construction of a bromide-selective electrode based on mer-cury salts where reactions similar to those above are involved,28 reported the appearance of small droplets of metallic mercury on the walls of the die where a membrane consisting of Hg2C12 HgS and Ag2S had been pressed with thermal treatment.This thermodynamic instability may explain the problems connected with the troublesome operation observed for the chloride and other electrodes with pressed membranes based on mercury salts.1O714J5J7926 A practically stable response of such electrodes requires that a metastable equilibrium state be reached in the membrane. The results of this work and their comparison with previous14 and literature results suggest that the use of an unpressed mixture makes the attainment of such a metastable state easier than when the mixture is disturbed by pressure and heating.Conditioning of the mercury salt electrodes in water after polishing the final stage of the construction or when the membrane is regenerated has been recommended5J3J4J3 and was found to be necessary in this work in order to obtain acceptable response characteristics. The effectiveness of such conditioning may result from the removal of chloride and other soluble species formed by the disproportionation reactions (discussed above) from the membrane surface. Financial support from INIC Lisbon (C.I.Q.U.P. Line 4A) and JNICT Lisbon (Research Contract 50.78.127) is grate-155 fully acknowledged as well as helpful discussions with Dr. J. D.R. Thomas (Cardiff Wales) made possible by a Travel Grant from the Scientific Affairs Division of the North Atlantic Treaty Organisation (Grant No. 069/84). We thank Mr. A. J. T. Sousa and Mrs. M. Isabel R. G. F. Sampaio for carrying out routine work. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. References Lima J. L. F. C. and Machado A. A. S . C. in Albaiges J., Editor “Analytical Techniques in Environmental Chemistry,” Volume 2 Pergamon Press Oxford 1982 p. 419. Lima J . L. F. C. and Machado A. A. S . C. Rev. Port. Quim., 1979 21 15. Lima J. L. F. C. and Machado A. A. S . C. Rev. Port. Quim., 1979 21 153. Lima J. L. F. C. and Machado A. A. S .C. Rev. Port. Quim., 1982 24 156. Lechner J. F. and Sekerka I. J. Electroanal. Chem. 1974, 57 317. Sekerka I. and Lechner J. F. J . Electroanal. Chem. 1976, 69 339. Sekerka I. and Lechner J. F. Anal. Lett. 1976 9 1099. Sekerka I. Lechner J. F. and Wales R. Water Res. 1975,9, 663. Sekerka I. Lechner J. F. and Harrison L. J. Assoc. Off. Anal. Chem. 1977,60,625 Bailey P. Wilson J. Karpel S. and Riley M. in Pungor E., Editor “Conference on Ion Selective Electrodes Budapest, 1977,” Elsevier Amsterdam 1978 p. 201. Marshall G. B. and Midgley D. Analyst 1978 103 438. Marshall G. B. and Midgley D. Analyst 1979 104 55. Ryan,.T. E. Peterson A. J. and Subsara W. P. in Moody, G. J. Editor “International Symposium on Electroanalysis in Clinical Environmental and Pharmaceutical Chemistry,” UWIST Cardiff 1981 paper 5.Lima J. L. F. C. andMachado A. A. S . C. Rev. Port. Quim., 1982 24 61. Tacussel J. and Fombom J . J. in Pungor E . Editor, “Conference on Ion Selective Electrodes Budapest 1977,” Elsevier Amsterdam 1978 p. 567. Guilbault G. G. Editor “Recommendations for Publishing Manuscripts on Ion-Selective Electrodes,” Pure Appl. Chem., 1981 53 1907. Tseng P. K. C. and Gutknecht W. Anal. Chem. 1976 48, 1996. Kielland J. J. Am. Chem. SOC. 1937 59 1675. Lurie J . “Handbook of Analytical Chemistry,” Mir Moscow, 1975 p. 305. Koebel M. Anal. Chem. 1974,46 1559. Buck R. P. and Shepard V. R. Anal. Chem. 1974,46,2097. Sillen L. G. and Martell A. G. Editors “Stability Constants of Metal Ion Complexes,” Special Publication No. 17 Chem-ical Society London 1964 pp. 286-288 and 292-293. “Ultra-Sensitive Solid State Chloride Electrode PHI 91 100 (Instruction Manual) ,” Graphic Controls Buffalo NY USA. da Silva M. G. P. Lima J . L. F. C. and Machado, A.,A. S . C Port. Electrochim. Acta 1984 2 29. Lima J. L. F. C. Machado A. A. S. C. and SB C. M., “Abstracts of the Sixth Meeting of the Sociedade Portuguesa de Quimica,” Sociedada Portuguesa de Quimica Aveiro 1983, paper PC36. Hulanicki A. Lewandowski R. and Lewenstam A. Anal. Chim. Acta 1979 110 197. Lurie J. “Handbook of Analytical Chemistry,” Mir Moscow, 1975 p. 110 Lima J. L. F. C. and Machado A. A. S . C. Port. Electrochim. Acta submitted for publication. Paper A51269 Received July 22nd 1985 Accepted August 8th 198