438 Analyst, May, 1978, Vol. 103, pp. 438-446 Solid-state Mercury( I) Chloride Electrode for Determining 0.1-1.0 pg ml-I Levels of Chloride in Boiler Water and Other High-purity Waters G. B. Marshall and D. Midgley Central Electricity Research Laboratories, Kelvin Avenue, Leatherhead, Surrey, KT22 7 S E A solid-state electrode based on mercury(1) chloride and mercury(I1) sulphide has been developed for determining chloride concentrations of 0.01-1 .O pg 1-1 in boiler water. The greater sensitivity of the electrode compared with that of silver - silver chloride electrodes enables concentrations as low- as 0.01 pg nil-1 to be determined by a simple manual technique. The total standard deviations a t chloride concentrations of 0.5, 0.1 and 0.01 pg ml-l were 0.025, 0.005 and 0.015 pgml-l, respectively.The electrode can be prepared easily in the laboratory from commercially available materials and a RdiiCka Selectrode. The only significant interference i:; from iron(II1) ions and this interference can be eliminated by adding fluoride ions to the sample. Keywovds : Chloride determina,tion ; water anaLysis ; potentiometvy ; chlovide- selectize electrode ; mercury ( I ) chloride electrode In order to avoid acid attack in boilers, the concentration of chloride in boiler waters must be kept at very low levels [in Central Electricity Generating Board (CEGB) practice a maximum of 0.2-2 p g m F , depending on the type of boiler]. The established CEGB manual potentiometric method1 and absorpt iometric methods based on mercury( 11) thio- cyanate are insufficiently sensitive for chloride concentrations much below 0.1 pg ml-l, such as occur in many boilers.The range of the potentiometric method has been extended by using a flow cell whose temperature is precisely controlled2 and that of the absorptiometric method by first concentrating the chloride by a co-precipitation te~hnique.~ In both methods the analytical procedure is fairly complicated. The existing potentiometric method uses electrodes based on silver chloride, but electrodes made from mercury( I) chloride should be more suitable for determining low concentrations of chloride because of the lower solubility of the mercury(1) salt. The difficulty of handling mercury - mercury(1) chloride electrodes has largely restricted them to use in reference electrodes, but an experimental solid-state ion-selective electrode with a membrane composed of a compressed mixture of mercury(I1) sulphide and mercury( I) chloride4 showed promise of being easy to use.We have combined the same membrane materials with a RGiiEka Selectrode5 to produce a solid-state chloride-selective electrode that can be made easily from commercially available components and is suitable for determining chloride at concentrations as low as 0.01 pg ml-l by a simple manual technique. Theoretical As with calomel electrodes, the presence of chloride in a solution disturbs the solubility product equilibrium of mercury(1) chloride [equation (l)], but there is also a simultaneous solubility equilibrium for mercury (11) su1phid.e [equation (2)]. The two solubility equilibria are linked by the disproportionation equilibriiim of mercury(1) ions [equation (3)] :MARSHALL AND MIDGLEY 439 The e.m.f.of an electrode containing mercury(1) chloride and mercury(I1) sulphide both in equilibrium with an aqueous solution of chloride ions should be given by a form of the Nernst equation, which is expressed in terms of either a chloride or sulphide response: E = EoRgaCl4 - k lOg(Cl-} . . .. .. - . (4) .. .. . . . . (4a) k 2 = E o H g S - - log{S2} where k = RTlnlO/F is the slope factor and (S2-1 is the activity of the sulphide ion dissolved from the electrode. From equations (1)-(3) (s2-} = Kggs {C1-}2/KdK,g,a, .. .. * ' (5) and hence from equations (4) and (4a) Experimental Apparatus Potentials were measured with an Orion 801 digital pH meter reading to 0.1 mV and displayed on a Servoscribe 2s chart recorder.When two electrodes were being used simul- taneously they were switched in turn through the pH meter by an Orion 855 automatic electrode switch. The reference electrodes were of the mercury - mercury(1) sulphate type with a ground- glass sleeve liquid junction and 1 niol 1-1 sodium sulphnte filling solution (Electronic Instru- ments Ltd., Type 1380-230). Pvepavntioqz of clzlovide clectvodes A sensitising mixture was prepared by mixing red mercury(I1) sulphide (BDH, Optran grade) and mercury(1) chloride in equimolar amounts and grinding them in an agate mortar until the mixture was uniformly pink. The mercury(I1) sulphide had been washed for 15 min with each of two portions of carbon disulphide, washed with acetone and air dried before use.Because of their toxicity these materials should be handled with care. Radiometer F3012 Universal Selectrodes were impregnated with the above mixture by rubbing the mixture into the exposed graphite surface with a glass rod? Reagents Water. Town mains water was distilled in a stainless-steel still (Manesty Machines Ltd., Liverpool), and the distillate passed through a twin-column mixed-bed de-ionisation unit (Elga Products Ltd., Model BlO6jZ). The conductivity of this water was less than 0.1 pS cm-l a t 20 "C as i t left the unit and was found by a modification of the method of Rodabaugh and Upperman3 to have a chloride content of about 0.7 pg 1-l. This water was used in the preparation of all standard or reagent solutions.A stock solution (1 000 pg ml-l of chloride) was prepared by dissolving 1.649 g of sodium chloride in water and making up to 1 1 in a calibrated flask. Further standard solutions were prepared by successive dilution of this solution. A solution was prepared by dilution of 6.25 ml of concentrated nitric acid (BDH, Aristar grade) to 11. Commercially available mercury( 11) sulphide (BDH, Optran grade) was used for most experiments, but a batch of material was also prepared in the laboratory as follows. Stadard chloride solutioi~s. Nitric acid, 0.1 mol 1-l. Mercury(I1) sul@hide.440 MARSHALL AND MIDGLEY : SOLID-STATE MERCURY(I) CHLORIDE Analyst, VOZ. 103 Sodium sulphide nonahydrate (AnalaR grade) (12 g) was dissolved in about 70 ml of water and the solution filtered through a 0.45-pm Millipore filter.The filtrate was added to a solution of 13.7 g of mercury(I1) nitrate (AnalaR grade) in about 100 ml of water. The precipitate was filtered off, washed with de-ionised water and dried. The material prepared by precipitation was black 13-cinnabar but the commercial product was the red a-cinnabar form. Analytical Procedure E.m.f. measurements were made with the electrodes immersed in stirred 50-ml portions of standard or sample solution to which 5 ml of 0.1 mol 1-1 nitric acid had been added. Between each measurement the electrodes were immersed in a stirred rinsing solution of 0.01 moll-1 nitric acid for about 2 min until the e.m.f. was less negative than that recorded with any of the solutions being analysed (about -30 m'V for solutions below 0.1 pg ml-1 of chloride).The analytical measurement therefore was always obtained from an electrode responding to an increase in concentration, which procedure gave the best over-all response time, and, in addition, contamination of one sample or standard solution by a more concentrated predecessor was prevented. Concentrations of chloride were obtained from a calibration graph prepared by making measurements, as above, with standard solutions. Before the start of each batch of analyses the ground-glass sleeve of the reference electrode was flushed by allowing a few drops of the internal filling solution to flow out. If this precaution was neglected, the e.m.f. tended to drift, as also occurred after a time when reference electrodes with ceramic frit junctions were used. Results Electrode Characteristics Nature of the membrane material Electrodes were sensitised with equimolar mixtures of mercury( I) chloride and either the red or black form of mercury(I1) sulphide.The electrodes with the red form had Nernstian responses (57-59 mV per decade) at concentrations above 1 pg ml-l, while those made with the black form had calibration slopes 1-2 mV per decade lower and were also much slower to respond. When the red mercury(I1) sulphide was washed with carbon disulphide, the sensitivity and response time of the electrodes were further improved. Electrodes prepared with two batches of red mercury(I1) sulphide had almost identical characteristics. Because of its poorer performance, no further tests were made using the black mercury(I1) sulphide. Electrodes prepared with mercury(1) chloride only were still sensitive to chloride, but the e.m.f.difference was only 45 mV between 0.1 and 1 pg ml-l solutions (cf., Table 11) and about 30 min were required for a steady potential to be established; for these reasons no detailed work was done with electrodes of this type. Variations between electrodes The amount of sensitising mixture in the membrane and the pressure exerted in forming it on the end of the electrode did not affect the performance of the electrode. The four Radiometer F3012 electrodes used during the tests gave similar results; the potentials of electrodes in the same solution rarely differed by more than 2 mV, regardless of the concentra- tion of the solution.Removal of half of the membrane and exposure of the graphite substrate had no deleterious effect on the electrode and. did not change the standard potential or the calibration slope. Conditioning of electrodes pared with the same electrode after use for 1 d. chloride solution was effective in conditioning the electrodes. Freshly impregnated electrodes showed a sluggish response and a lower sensitivity com- Immersion overnight in a 10 pg ml-1M a y , 1978 ELECTRODE FOR DETERMINING CHLORIDE IN BOILER WATER 441 Optimum PH f o r chlovide determinations At pH values of 3 or above, the formation of hydroxo complexes of mercury reduces the sensitivity of the electrode at the lower end of the concentration range ((10 pg ml-l). The addition of nitric acid to the sample reduced this interference (Table I) and it was found that there was little practical difference in the sensitivity over the range 10-3-10-2 mol 1-1 of nitric acid.Once enough acid has been added to overcome the hydroxide interference, the shift in the potential with increasing acid concentration at constant chloride concentra- tion is a more likely source of error than small variations in sensitivity; a concentration of mol 1-1 of nitric acid was, therefore, adopted for the analytical procedure, as the greater buffer capacity was considered advantageous. TABLE I EFFECT OF ACIDITY ON ELECTRODE RESPONSE (ELECTRODE E.M.F. IN MILLIVOLTS) Nitric acid Chloride concentration/g ml-I concentration1 I A I moll-1 0" 0.1 1.0 10 100 - 34.5 -83.5 - - 0 - 67 - - 94 - 141 - 204 10-3 - 13 10-2 - 24 -47.5 - 96 - 155 -213 * De-ionised water, no added chloride.E f e c t o f stirring with 5 min in a stirred solution. equilibrium e.m.f. Without stirring, the electrode took 20-30 min to reach an equilibrium e.m.f,, compared The rate of stirring had little effect (<2 mV) on the E f e c t of light The potential decreased (indicating an apparently higher chloride concentration) when the electrode was exposed to more intense light and increased when it was shaded. The effect was reversible in all instances. Switching off the laboratory fluorescent lights on a dull day caused a 2-mV change in potential within 1 min and similar changes could be produced by variations in the intensity of sunlight. For the main body of the tests the electrodes were fixed in an opaque holder which fitted on top of black-painted 100-ml beakers, thus excluding virtually all light from the membrane of the electrode.Resistance The resistance of the cell formed by the chloride electrode and the reference electrode immersed in 50 ml of 0.1 pg ml-l chloride solution to which 5 ml of 0.1 mol 1-1 nitric acid had been added was 30 kQ, as measured with an Avometer. This resistance is small enough to allow the cell potentials to be measured with many types of digital voltmeter; potentials measured with a Solartron LM 1420.2 digital voltmeter were within 0.1 mV of those measured on a pH meter with a high input impedance. Observed and Predicted e.m. f.s The e.m.f. observed when the electrode is immersed in a 1 pg ml-1 chloride solution at 25 "C and pH 2.5 (obtained by addition of 5 ml of nitric acid per 50 ml of solution) can be compared with the values predicted by equations (4) and (4a).The standard potentials6 are E& = - 750 mV and E&ZClo = 268 mV and the chloride ion activity is 2.32 x mol 1-1 (allowing for dilution and using activity coefficients calculated from the Davies equation') ; hence E = E&,a, - 59.16 log{Cl-} = 542 mV From equation (5) and the equilibrium constants6 for equations (1)-(3), and hence {SZ-) = 1.2 x 10-43 mol 1-1 E = E& - 29.58 log(S2-} = 520 mV442 MARSHALL AND MIDGLEY : SOLID-STATE MERCURY(1) CHLORIDE Analyst, VOZ. 103 The observed values, after correction for the :potential of the reference electrode, were in the range 530-537 mV.The agreement between the observed and predicted e.m.f.s is acceptable, particularly in view of the uncertainty in the value of K,,,. Performance Tests The electrodes used for the following tests were made with red mercury(I1) sulphide that had been washed with carbon disulphide and were immersed in stirred chloride solutions containing 0.1 moll-1 nitric acid (5 ml per 60 ml) in darkened glass beakers. Electrodes were conditioned overnight in 10 pg ml-l chloride solution before being used for the first time. Unless otherwise stated, the solutions were brought to a temperature of 25 "C in a water-bath before being analysed. Concentration range The response of the electrode is Nernstian from at least 1 000 down to about 0.35 pg ml-1. Below that level the calibration graph becomes increasingly curved.Fig. 1 shows the curved portion of a typical calibration graph. The slope of the linear portion (1-10 p g ml-l) was -57.8 mV per decade increase in concentration, with a standard deviation of 0.25 mV per decade increase in concentration (estimated from the least-squares fit of the points to a straight line). It is possible to use the non-linear part of the calibration graph to measure chloride concentrations down to 0.01 pg ml-l and the sensitivity can be increased by working at reduced temperatures. C h I or i d e con ce n t r a t i o n / p a m I -- ' 1.O 0.1 @.Of 0.001 - 1 I 10-5 10--6 10--7 C h I or i d e con ce n t r a t i on / m o 1 I -- Fig. 1. Calibration of chloride ion- selective electrode.Precision Over a period of 3 d, five batches of five standard solutions each were analysed in duplicate. The e.m.f. values were normalised with respect to the mean e.m.f. reading for the 1 pg ml-2 standard in each batch and the within-batch, between-batch and total standard deviations were calculated. The results for one electrode are shown in Table 11. A second electrode immersed in the same solutions at the same time gave almost identical results, in which neither the mean values of the normalised e.rx1.f.s nor the standard deviations were signifi- cantly different, at the 5% level, from those in Table I1 (t- and F-tests, respectively). Table I11 shows the recorded e.m.f. values for the two electrodes in batches of 1.0 pg ml-1 chloride solution. The correlation coefficient between the two sets of e.m.f.s was 0.97, showing the high degree of co-variance between electrodes immersed in the same solution at the same time.It is inferred that the changes in e.m.f. are caused less by the variability of the chloride electrodes than by factors that could affect both simultaneously, e.g., small changes in temperature, pH or liquid junction potential.May, 1978 ELECTRODE FOR DETERMINING CHLORIDE IN BOILER WATER TABLE I1 443 PRECISION OF MEASUREMENTS OF CHLORIDE CONCENTRATIONS Standard deviationt Chloride concentration/ A e.m.f./ I A \ pg ml-1 mV* Within-batch Between-batch Total 1.0 0.0 0.63 - - (0.03) 0.5 17.6 0.93 0 0.93 (0.02 5) (0) (0.025) 0.1 51.1 0.83 NSS 0.91 (0.004) (0.0 04) (0.00 5) (0.005) 0.05 59.3 0.74 NS 0.75 0.01 64.6 2.04 0 2.04 (0.0 15) (0) (0.0 15) * E.m.f.normalised with respect to 1 p g ml-l solution, e.g., AOel = EOal - El.o. t Figures in parentheses are standard deviations in concentration units ( p g ml-l). NS = non-significant a t the 5% level. TABLE I11 VARIABILITY OF E.M.F.S OF TWO ELECTRODES IN 1.0 pg ml-l CHLORIDE SOLUTION E.m.f./mV 1 2 3 4 6 r L \ Electrode ,-A-3 r - A - l ,-A-~ c p A - \ number A* B* A B A B A B A B 23 - 106.0 - 106.3 - 103.3 - 103.8 - 103.2 - 105.2 - 102.3 - 101.4 - 100.6 -100.9 24 -102.2 -103.4 -101.4 -101.8 -101.0 -102.3 -99.9 -99.3 -98.3 -98.9 * A and B are the first and second 1 p g ml-1 solutions, respectively, in each batch. Accuracy The accuracy of the analyses using the electrode was tested both by analysing samples of boiler water from five power stations and comparing the results with those obtained by the mercury(I1) thiocyanate absorptiometric method,s and by spiking the samples with 0.1 pg ml-l of chloride and measuring the recovery.The results are shown in Table IV. TABLE IV ANALYSIS OF BOILER WATERS Chloride contentlpg ml-l I Potentiometry Station Location A Boiler 1(A) Boiler 1 (B) B Unit 4(A) Unit 4(B) C Unit 1 Unit 2 D Unit 1 Unit 3 E Boiler 1 Boiler 2 Absorptiometry : sample concentration 0.17 0.18 0.08 0.09 0.11 0.07 0 01 0.01 0.01 0.03 r Sample concentration 0.17 0.17, 0.10 0.10 0.08 0.06 0.01 0.01 0.01 0.01 1 Recovery of 0.1 p g ml-l spike 0.11 0.11 0.11 0.10 0.10 0.11 0.10 0.10 0.10 0.10444 MARSHALL AND MIDGLEY: SOLID-STATE MERCURY(1) CHLORIDE Analyst, VOZ. 103 Response time The response for a change from de-ionised water to 0.1 pg ml-l chloride solution or from 0.1 to 1.0 pg ml-l chloride solution was complete in 5 min, but changes in the reverse direction took longer (15-20 min).A better over-all time (approximately 5 min) for the analytical procedure was obtained by immersing the electrode in stirred 0.01 moll-1 nitric acid solution for about 2 min before the next solution was analysed. Interferences Substances that could occur in power station waters were tested for their interference effects by observing the change in e.m.f. of the electrode when 100-pl portions of concentrated standard solutions of the interferents were injected into a mixture of 50 ml of 0.1 pg ml-1 chloride solution and 5 ml of 0.1 mol 1-1 nitric acid.Dilution of the chloride solution on addition of the interferent solution was calculated to cause a change of less than 0.05mV, which would not have been detectable on the pH meter. The concentrations of interferents tested were generally much higher than those expected in power station waters. The following (separately) caused an interference no greater than 0.001 pg ml-l in the determina- tion of 0.1 pg ml-l of chloride: 2 pg ml-l of 200 pg ml-l of carbon dioxide, 0.43 pgml-1 of silicon dioxide, 120 pg ml-l of CHJOO-, 20 pgml-l of PO:-, 12.6 pgml-l of Cu2+, 20 pg ml-l of Ni2+, 5 pg ml-l of Ca2+ + 5 pg ml-l of Mg2+, 2 pg ml-l of Zn2+, 10 pg ml-1 of Cr3+, 0.1 pg ml-l of Fe2+, 0.02 pgml-l of Fe3+, 10 pgml--l of ammonia, 1 pg ml-1 of hydrazine, 2 pg ml-l of cyclohexylamine or 1100 pg ml-l of morpholine.caused a bias of +0.03 pg ml-l in the determination of 0.1 pg ml-l of chloride and 0.1 pg ml-l of Fe3+ caused a bias of -0,034 pg ml-l, which was eliminated by the addition of fluoride. Anions that form insoluble mercury( I) salts will interfere if present in such concentrations that they can displace chloride from the mei-cury(1) chloride in the membrane. The most strongly interfering ions of this type are sulphide, cyanide, iodide and bromide, but these are not normally present in power station waters. The interference effects of some ions can be eliminated, as in the proposed analyticad procedure, by acidifying the solution, e.g., hydroxide, carbonate and hydrogen phosphate. Sulphate interfered at much lower con- centrations than predicted by this mechanism, probably because the solubility of mercury( I) chloride was enhanced by the formation of a soluble Hg,SO,O complex.Substances that form strong complexes with mercury(I1) ion will promote the dispro- portionation of mercury(1) ion, causing chloride to be released from the membrane and the electrode to indicate a higher chloride concentration than was present in the sample solution. Under the conditions of the analytical procedure, 16 pg ml-l of sulphite caused an inter- ference equivalent to 3 pg ml-l of chloride because of the reaction The addition of 20 pg ml-l of Hg,Cl, + 2S0,2- -+ Hg(S0,),2- + Hgo + 2C1- Cations that form strong chloro complexes will reduce the concentration of free chloride, but this interference can be overcome by adding a substance that will compete with chloride for the interferent but not for mercury(1) ions.The only important interferent of this kind likely to be present in boiler water is iron(IIl[), the effect of which is eliminated by adding fluoride ions, e.g., 0.1 p g ml-l of Fe3+ reduced the reading obtained with 0.1 pg ml-1 of chloride by 0.034 pg ml-l, but with 20 pg ml-L of fluoride present the interference effect was less than 0.01 pg ml-l of chloride. Iron(I1) and other divalent metal ions had no detectable interference effects at the concentrations tested, which are much higher than those likely to be found in boiler water. E f e c t of temperature The Nernstian sensitivity of the electrode increases with temperature, according to the factor RTlnlOIF, but at reduced temperatures,, because the solubility of mercury( I) chloride is suppressed, the linear calibration range is extended to lower concentrations.In addition, there is a shift in the standard potential of the cell formed by the chloride and reference electrode pair to more negative e.m.f. values ,as the temperature decreases. The combined effect of these factors was tested by measurements made at three temperatures (Table V).May, 1978 ELECTRODE FOR DETERMINING CHLORIDE I N BOILER WATER 445 For the analysis of solutions containing less than 1.0 pg ml-l of chloride, the shift in the standard potential is a greater source of error than the change in sensitivity. For measure- ments below about 0.05 vg. ml-l of chloride, it may be worth working at reduced temperatures to obtain greater sensitivity, in spite of the longer response times, e.g., about four times longer at 7-10 "C than at 25 "C.TABLE V EFFECT OF TEMPERATURE ON E.M.F. E.m.f. /mV Concentration of I A > chloride/pg ml-l 25 "C 15 "C 7 "C 1.0 - 87 - 108.5 - 115 0.1 - 39 -59.5 -67.5 0.01 - 25 - 32 - 35 0" -22.5 - 24 - 29 * De-ionised water, no added chloride. Lifetime of the electrode Each application of the sensitising mixture gave a membrane with a lifetime of at least 2 months. When the response of the electrode became sluggish the membrane was shaved off the end of the electrode with the tool provided and the exposed tip re-impregnated with more of the sensitising mixture. The RfiiiEka Selectrode should last for years, its lifetime being limited only by the loss of material each time an old membrane is shaved off.The sensitising mixture can be stored for at least 6 months without ill-effect. Table VI shows the potentials recorded during the life of a typical electrode. As the measurements were made at ambient temperature, the potentials could not be expected to be constant over the period of trial, but no trend in the calibration slope (exemplified by the e.m.f. difference in Table VI) or the standard potential could be discerned. TABLE VI ELECTRODE RESPONSE AS A FUNCTION OF TIME The electrode was prepared on May 25th, 1976. Date of test 26.5.76 10.6.76* 14.6.76 21.6.76 20.7.76 23.8.76 1.9.76 E.m. f .ImV Chloride content Chloride content 1 pgml-l 0.1 pg ml-l f J. \ - 147.2 - 96.5 - 104.2 -53.9 - 103.4 -53.2 - 101.9 -50.8 - 107.7 -57.5 - 108.3 -55.7 - 105.1 - 54.4 E.m.f.difference/ mV 50.7 50.3 60.2 51.1 50.2 52.6 50.7 * Reference electrode changed. Discussion Lechner and Sekerka4 made electrodes with pelleted membranes of the same materials as we have used and obtained generally similar results, although their response times were shorter. The precision of analyses with their electrodeg was poorer than that reported here, e.g., at 0.1 pg ml-l they obtained a relative standard deviation of 14% compared with 4% in this work. The performance of the pelleted membranes depended on the temperature and duration of compression and at best 3 h at 150 "C were required for the preparation of a successful electrode, whereas variations in preparing the impregnated graphite electrode had no appreciable effect on the performance.The solid-state chloride-selective electrode described in this paper is suitable for the analysis of power station waters containing 0.01-1.0 pg ml-l of chloride, using a simple446 MARSHALL AND MIDGLEY manual technique. Higher concentrations can also be determined, if required. At 0.1 pg ml-1 of chloride, better precision can be obtained with this electrode than with the silver - silver chloride type1 (0.004 pg ml-l compared with 0.04 pg ml-l for the total standard deviation), although at 1 pgml-1 the two electrodes are equally precise. At concentrations below 1 pg ml-l the mercury(1) chloride electrode is much more sensitive than the silver chloride types, e.g., the difference between the e.m.f.s observed in 1.0 and 0.1 pg ml-1 chloride solutions is 51 mV for the mercury(1) chloride electrode and 19 mV for silver chloride electrodes and between 0.1 and 0.01 pg ml-l the corresponding differences are 13.5 and 2 mV, respectively.The mercury(1) chloride electrode is slower to reach equilibrium than the silver chloride type, taking 5 min to reach a steady e.m.f. where the latter requires less than 1 min. The mercury(1) chloride electrode is similar in precision at 1 pg ml-1 of chloride to the mercury(I1) thiocyanate absorptiometric method,1° but its precision is better at lower concentrations. Power station waters are unlikely to contain sufficient concentra- tions of interfering substances to bias the results of analyses and generally good agreement was obtained when samples of boiler water from several power stations were analysed using the electrode and the mercury( 11) thiocyanat e absorptiometric method. Good recoveries of spikes were obtained from the same samples and there was no evidence of an effect from iron(II1) ions, the most likely source of interference. The sensitivity of the electrode could make it suitable for monitoring very low concentra- tions of chloride ((0.01 pg ml-l), especially if used in a flow-cell at a reduced and carefully controlled temperature, as has been done with silver chloride electrodes.2 Further work is in hand to assess the performance of the electirode in continuously flowing solutions at low temperatures. This work was carried out at the Central Electricity Research Laboratories and is published by permission of the Central Electricity Generating Board. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. References Torrance, K., Analyst, 1974, 99, 203. Tomlinson, K., and Torrance, K., Analyst, 1977, 102, 1. Rodabaugh, R. D., and Upperman, G. T., Analytica Chim. Acta, 1972, 60, 434. Lechner, J. F., and Sekerka, I., J . Electroanalyt Chem. Interfacial Electrochem., 1974, 57, 317. Hansen, E. H., Lamm, C. G., and RbiiCka, J., .AnaZytica Chim. Acta, 1972, 59, 403. Latimer, W. M., “Oxidation Potentials,” Second Edition, Prentice-Hall, Englewood Cliffs, N. J . , Davies, C. W., “Ion Association,” Butterworths, London, 1962. Florence, T. M., and Farrar, Y . J., Analytica Chim. Acta, 1971, 54, 373. Sekerka, I., Lechner, J. F., and Wales, R., Wat. Res.. 1975, 9, 663. Webber, H. M., Wheeler, E. A., and Wilson, A. L., unpublished work. 1952. Received November 16th, 1977 Accepted December 28th, 1977