Analyst June 1983 Vol. 108 pp. 701-711 701 Potentiometric Determination of Sulphite by Use of Mercury(1) Chloride - Mercury(l1) Sulphide Electrodes in Flow Injection Analysis and in Air-gap Electrodes Geoffrey B. Marshall and Derek Midgley" Central Electricity Research Laboratories Kelvin Avenue Leatherhead Surrey KT22 7SE Flow injection analysis using as the detector a solid-state ion-selective electrode with a mercury(I1) sulphide - mercury(1) chloride membrane can be used for determining sulphite or dissolved sulphur dioxide in water. At concentra-tions in the range 1.5-10mg1-1 of sulphite the method has a Nernstian response of 60 mV per decade but a t lower concentrations (down to 0.1 mg 1-l) the e.m.f. is linearly related to the sulphite concentration. Although the flow injection method is less sensitive than direct use of the electrode it avoids the problem of chloride interference and permits the determination of sulphur dioxide in the commonly used tetrachloromercurate absorbent.The only serious interference found was from sulphide although a small effect was also obtained from thiosulphate. Measurements in the range 0.1-10 mg 1-1 of sulphite had relative standard deviations for single results of no more than 2%. The method requires only two reagents (dilute nitric acid solutions) and is simple to operate. Each analysis is complete in less than 5 min. Air-gap electrodes using the same sensor had sub-Nernstian responses of very poor reproducibility and were not considered to be a practical means of determining sulphite.Keywords ; Flow injection analysis ; air-gap electrode ; ion-selective electrode ; sulphite and sulphur dioxide determination ; mercury(II)sulphide - mercury(I) chloride membrane electrodes Ion-selective electrodes with mercury( 11) sulphide - mercury( I) chloride membranes are known to respond to sulphite in solution1 according to the reaction Hg,CI,,, + 2S03,- -+ Hg(SO,),,- + Hgo + 2C1- . . - - (1) Reaction (1) occurs at the surface of the electrode which is sensitive to chloride ions; hence the electrode's e.m.f. is determined by the amount of chloride released and therefore by the original concentration of sulphite. The use of an electrode for determining sulphite directly in aqueous samples is limited be-cause the electrode responds primarily to chloride which is very commonly present in such samples.In order to avoid chloride interference we have separated the electrode from the sample by a diffusive barrier across which the sulphur dioxide but not chloride or other ions, can pass. We have used two techniques the air-gap electrode2 and flow injection analysis In the air-gap electrode acidification of the sample releases sulphur dioxide which diffuses across a small air space and is absorbed in a film of chloride-free collecting solution on the surface of the membrane. In flow injection analysis small volumes of sulphite solution are injected into a continuously flowing carrier stream of nitric acid which passes on one side of a gas-permeable PTFE membrane. Sulphur dioxide volatilised by the low pH of the carrier stream diffuses across the PTFE membrane into a continuously flowing absorbent stream of less concentrated nitric acid.This absorbent stream then flows over the surface of a mercury( I) chloride - mercury( 11) sulphide membrane electrode. (FIA) .3 * To whom correspondence should be addressed 702 MARSHALL AND MIDGLEY POTENTIOMETRIC DETERMINATION Analyst VoZ. 108 Theoretical Mercury( I) chloride is sparingly soluble and for the reversible reaction Hg2C12+Hg22+ + 2C1- - ’ (2) K = [Hga2+][C1-I2 = 10-17-88 The dissolved mercury(1) ions disproportionate to mercury(0) and mercury( 11) as follows : Hg22++Hg0 + Hg2+ - * (3) Kd = [Hg”] [Hg2+]/[Hg2”] = 10-Sulphite forms a strong complex with mercury(I1) ions: Hg2+ + 2S03”+Hg(S0,),” * (4) /3 = [Hg(SO3),2-]/[Hg2+] [S032-]2 = The introduction of sulphite into a solution in which the mercury(1) chloride electrode is immersed reduces the concentration of free mercury(I1) ions and therefore forces equilibrium (3) and then equilibrium (2) to the right.The result is that two chloride ions dissolve from the electrode for each two sulphite ions added according to the over-all reaction (1). As the electrode responds to chloride ions its e.m.f. is a measure of the sulphite ion concentration, provided that reaction (1) is the only source of chloride ions apart from the solubility of mercury(1) chloride itself. The e.m.f. follows the usual Nernstian response : E = E” -Kl0g[SO32-] . . (5) where E” is the standard potential and k is the slope factor theoretically equal to RTln(lO)/F where R is the gas constant T is the temperature (K) and F is Faraday’s constant.It may be noted that the slope factor has the value expected for the singly charged chloride ion (-60 mV per ten-fold change in concentration) and not the 30 mV per ten-fold change in concentration that would be obtained with an electrode responding directly to a doubly charged ion such as sulphite. For operation of the electrode at low chloride (and therefore sulphite) concentrations the solution should be below pH 5.4 At pH 34 as used in this work most of the sulphite would be present as the hydrogen sulphite ion and the conditional stability constants /3* would be lOl6 and 10l8 at pH 3 and 4 respectively: when TBoz is the total concentration of dissolved sulphur dioxide.These constants are large enough to ensure that reaction (1) goes to the right and that equation (5) can be rewritten as where E* represents the apparent standard potential in the conditions concerned. Experimental Appa ratus Sensing electrodes The mercury(1) chloride electrodes used were SL-01 (Ionel Electrodes Ontario) and PHI 91100 (Graphic Controls London). The reference electrodes for the air-gap electrodes were 1370-230 mercury - mercury(1) sulphate electrodes with 1 mol 1-1 sodium sulphate filling solutions (Electronic Instruments Ltd., Chertsey). Those used in flow injection analysis were Radiometer K.601 mercury - mercury(1) sulphate electrodes with saturated potassium sulphate filling solutions June 1983 OF SULPHITE BY FLOW INJECTION ANALYSIS 703 E m .f. measurements Air-gap electrode e.m.f.s were measured with a Beckman 4500 digital pH meter reading to 0.1 mV and the time course of the e.m.f. was followed on a chart recorder. For flow injection analysis the e.m.f.s were read as peak heights from the chart recorder. Hole for brass swivel Injection port (centred over sample cup) Stud location holes 5-mm thick Perspex lid Reference electrode holder /---- l-mm step 5 mm mm Location stud Electrode holder 5 mm diameter hole (for sample cup) Channel cut to fit ceramic junction 10 mm diameter hole x 20 mm dee Diameter appropriate to electrode (lone1 1.0 cm) graphic contro 0.95 cm l+- 70 mmd to reference electrode Fig. 1. Inverted air-gap electrode. A ir-gap electrodes Air-gap electrodes were constructed by mounting the sensing and reference electrodes in suitably machined PTFE blocks.The first apparatus was similar to that used by RBiiEka and Hansen2 and because of the upright position of the sensing electrode it was difficult to achieve a thin stable film on the surface unless a high concentration of non-ionic detergent was added to the solution. With this design the sample was acidified before the electrodes were in place, which might have led to a variable loss of sulphur dioxide. This is subsequently described as the upright air-gap electrode. A modified apparatus was devised (Fig. 1) that held the electrode in an inverted position. With this arrangement the solution on the surface did not have to contain detergent unless very small volumes (<25 p1) of solution were used.It also had the advantage that acidification of the sample took place only when the apparatus was closed so that losses of sulphur dioxide were minimised. This is subsequently referred to as the inverted air-gap electrode. membrane flow-through cell valve Lay-out of FIA system. Fig. 2 704 MARSHALL AND MIDGLEY POTENTIOMETRIC DETERMINATION Analyst VoZ. 108 FZow injection apparatus The flow system (Fig. 2) including the injection valve the gas transfer block and the elec-trode holder was part of the Bifok FIA-05 apparatus (EDT Research London). The flow streams were pumped by a Gilson Minipuls 2 peristaltic pump. Reagents A stock solution (1 000 mg 1-1 of S032-) was prepared by dissolving 1.575 g of anhydrous sodium sulphite (BDH Chemicals AnalaR grade) in 500 ml of de-ionised water containing about 1 ml of glycerol.The volume was made up to 1 1 with de-ionised water. A working solution (10 mg 1-1 of SOS2-) was prepared by dilution of 5 ml of stock solution to 500 ml. About 1 ml of glycerol was added before the volume was made up to the mark. Further solutions were prepared as required by dilution of the working solution. Some solutions also contained 0.1 mol 1-1 of sodium tetrachloromercurate preservative [11.7 g 1-1 of sodium chloride + 27.2 g 1-1 of mercury(II)chloride]. The stock solution (1 mol 1-1) was prepared by diluting 31.5 ml of concentrated nitric acid (BDH Chemicals Aristar grade) to 500 ml. The carrier solution (0.1 moll-l) and the absorbent solutions and 10-4 moll-l) for flow injection analysis were prepared by successive dilution of the stock solution.Interfered soZzdions. Solutions containing 10-3 moll-1 of each of the interferents were prepared from BDH Chemicals AnalaR materials sodium acetate formic acid sodium carbonate sodium sulphide and sodium thiosulphate. Sodium sulphide was also tested at the and mol 1-1 levels. mol 1-1 nitric acid from dilution of the 1 mol 1-1 stock. Other components were sometimes added a non-ionic detergent (ICI Lissapol NX or BDH Nonidet P40) and mol 1-1 mercury(1) nitrate in Procedure Flow injection analysis The carrier sample and absorbent streams were pumped continuously at rates of 1 1 and 0.23 ml min-l respectively. When a steady base line had been reached a 30-pl slug of sample was injected into the carrier stream.The output of the pH meter acting as a unity gain amplifier was registered on the 100-mV scale (0-10 mg 1-1 sulphite solutions) or the 20-mV scale (0-1.5 mg 1-1 sulphite) of the chart recorder. When the base line was reached again, another slug of sample could be injected. The minimum time between samples was about 4 min. It is important to adjust the electrode holder properly if noise is to be minimised. The inlet should be positioned at the top edge of the indicator electrode so that the sample flows evenly over the whole sensing surface. The outlet should be positioned so that the level of spent sample in the sump of the holder does not reach the sensing surface of the indicator electrode yet remains in continuous contact with the solution flowing over the surface.The formation of large drops at the bottom of the indicator electrode should also be avoided. A ir-gap electrodes A volume of film solution (5400 pl) was applied to the surface of the sensing electrode by means of a syringe pipette care being taken to see that the film also covered the end of the frit leading to the reference electrode. The electrode was allowed to stand until a reading of -30 to -50 mV was obtained i.e. similar to the reading in a bulk solution free of chloride and sulphi t e . A 1-ml portion of sulphite solution was added to the sample container containing a magnetic stirrer bar but with the stirrer off. The procedure then varied according to the apparatus used. Sulphite solutions. Nitric acid solutions.Film solutions. Film solutions for the air-gap electrode contained mol 1-l nitric acid. The amounts added are described in the text as they occur. U$wight air-gap electrode When the electrode was reading in the range -30 to -50 mV 0.3 ml of 1 mol 1-1 nitric acid was added to the sample container the electrode holder was quickly placed on top of the sample container and the stirrer was started. The subsequent change in e.m.f. was recorded Jme 1983 O F SULPHITE BY FLOW INJECTION ANALYSIS 705 Inverted air-gap electrode The lid was placed on the electrode holder and when the electrode was reading in the range -30 to -50 mV 0.2 ml of 1 mol 1-1 nitric acid was injected into the sample cup by a self-filling syringe pipette. The stirrer was started and the subsequent change in e.m.f.was recorded. Results Preliminary Experiments Before use in FIA or air-gap electrodes the mercury(1) chloride electrodes were tested for their response in bulk solutions of sulphite ion. A 1000 mg 1-1 sulphite solution was added from an Agla micrometer syringe to 50 ml of 10-3 mol 1-1 nitric acid; the steady e.m.f. was recorded after each addition. Fig. 3 shows that the calibrations obtained with electrodes from the two manufacturers were very similar having Nernstian responses of 55-56 mV per ten-fold increase in concentration above about 1 mg 1-1 of sulphite. Deviations above the line in the region 0.2-1.0 mg l-l are typical of a loss of sulphite by oxidation or volatilisa-tion. Deviations below the line (<0.2 mg 1-l) are caused by chloride dissolved from the mercury(1) chloride in the electrode itself.The chloride calibration (moll-1) of the Ionel electrode is also shown; as predicted it coincides with the sulphite calibration. The response to each increment in sulphite concentration was complete in 1-2 min and the calibrations were reproducible to &2 mV. 0 E ui -100 Su I p hite concent rat ion/mg I - ’ 0.05 0.1 0.5 1.0 5.0 - 1 I I 1 1 10-7 10-6 10-5 I O - ~ Sulphite concentration/mol I- ’ Fig. 3. Calibration in bulk solution. 0 Graphic controls. Ionel 0 SOae-; and x C1-. Flow Injection Analysis Carrier solution The 1.0 mol 1-1 nitric acid was tried first but at this high concentration nitric acid diffused through the membrane and lowered the pH of the absorbent (10-4 mol 1-1 nitric acid absor-bent emerged at pH 3.5).Even a slug of de-ionised water was sufficient to give a signal, because interrupting the flow of carrier caused an increase in the pH at the electrode’s surface and hence a negative shift in e.m.f.4 With 0.1 mol 1-1 nitric acid carrier the problem did not occur and this concentration was used for all subsequent tests. Absorbent solution The peak heights obtained for 10 mg 1-1 sulphite in 10-4 and 10-3 mol 1-1 nitric acid absorb-ents were 67.6 & 0.5 and 68.3 & 1.5 mV respectively. The difference was negligible any increase in the efficiency of absorption of sulphur dioxide and formation of the mercury(I1) -sulphite complex at the higher pH presumably being balanced by the effect of the small but significant increase in the formation of hydroxo - mercury complexes 706 MARSHALL AND MIDGLEY POTENTIOMETRIC DETERMINATION Analyst VoZ.108 \ + E ui H 10 min + Time Fig. 4. FIA responses to 1-10 mg 1-1 of sulphite. 1.5 mg I-' I mg I-' -4 I C- Time Fig. 5. of sulphite. FIA responses to 0-1.5 mg 1-' Sensitivity The responses for 1-10 mg 1-1 of sulphite are shown in Fig. 4 and for 0-1.5 mg 1-1 of sulphite (on a five-fold expanded scale) in Fig. 5. When the peak heights are plotted against the logarithm of the concentration (Fig. 6) it can be seen that a linear response is obtained in the region 10-1.5 mg 1-l. At 60.5 mV per decade change in concentration the sensitivity is almost Nernstian at 25 "C. At concentrations below 1.5 mg 1-1 the response increasingly deviates from Nernstian linearity but plotting peak heights directly against concentration in this region gives a linear calibration with a correlation coefficient of more than 0.99 (Fig.7). The limit of Nernstian response for an electrode directly immersed in sulphite solution was about 0.1 mg 1-1 (Fig. 3) implying that flow injection analysis is less than 10% efficient in its use of the sulphite available in the sample solution. 20 I 1 0.1 0.5 2 10 Sulphite concentration/mg I-' 0 0.5 1.0 1.5 Sulphite concentration/mg I- ' Fig. 6. FIA calibration at concentrations up to 10 mg 1-' of sulphite 0. results from normal scale (Fig. 4) ; x results from expanded scale (Fig. 5). Fig. 7. FIA calibration at concentrations up to 1.6 mg 1-l sulphite.Precision Within-batch standard deviations were determined from four successive injections of each standard solution. For measurements on the 100-mV scale of the recorder the standard deviations for a single result were 0.5 mV for 2 5 and 10 mg 1-1 solutions and 0.3 mV a June 1983 OF SULPHITE BY FLOW INJECTION ANALYSIS 707 1 mg 1-l; these correspond to relative standard deviations in concentrations of less than 2%. For measurements made on the 20-mV recorder scale the standard deviations for single results were 0.1-0.2 mV for solutions in the range 0.1-1.5 mg l-l corresponding to relative standard deviations in concentration of 1 4 % . Interferences In order to interfere a substance must be capable both of crossing the PTFE membrane in the FIA apparatus and of reacting at the electrode’s surface either by forming a less soluble precipitate than mercury( I) chloride or by forming complexes with mercury(1) or mercury(I1) ions.In order to cross the membrane the potential interferent must exist in a volatile non-ionic form in the acidified sample stream; halides therefore do not reach the electrode and so cannot interfere. A negative interference could occur if a substance hindered the diffusion of sulphur dioxide across the membrane because it formed complexes with sulphite that were either thermodynamically or kinetically stable in the acid carrier. Solutions of mol 1-1 sodium acetate formic acid and sodium carbonate produced peaks of less than 0.3 mV with the recorder set to 20 mV full scale i.e. less than the blank readings shown in Fig.6. Sodium thiosulphate moll-1) produced a peak of 0.6 mV equivalent to 3.5 x 10-7 mol 1-1 (0.04 mg 1-l) sulphite. The only serious interference was sulphide as this can diffuse across the PTFE membrane as hydrogen sulphide and displace chloride from mercury(1) chloride as follows : A concentration of mol 1-1 of sulphide produced a peak of 2.5 mV equivalent to about 1.8 x loA6 mol 1-1 (0.2 mg 1-l) sulphite and 10-4 moll-1 sulphide gave a peak of 52 mV which is equivalent to 5 x mol 1-1 of sulphide produced a very large off-scale peak and the electrode’s response was subsequently very noisy. On inspection the surface of the membrane seemed to be tarnished and after it was polished the performance was restored. In the determination of sulphur dioxide in air the sulphur dioxide is often collected in a preservative solution of sodium tetrachloromercurate.The possible interferences of this solution were investigated in case the complexes formed by the preservative HgSO,Clnn- with n = 1 2 or 3 could not be decomposed by the acid. Sulphite solutions of concentration 10 mg 1-1 with and without tetrachloromercurate gave peaks of 65.4 & 1.1 mV and 64.6 & 1.0 mV respectively and hence no significant interference could be detected. This also con-firms that chloride does not interfere. mol 1-1 (5.6 mg 1-l) sulphite. A concentration of Response time As shown in Figs. 4 and 5 the peak was reached in less than 1 min after breakthrough but the wash-out was slower corresponding to the usual differences in rate of response for increases and decreases in c~ncentration.~~~ Sharper but smaller peaks were obtained by increasing the rates of flow of the carrier and absorbent streams.Air-gap electrodes Sensitivity All the variations of physical configuration and film composition in the air-gap electrode gave a fairly large response to sulphite (Fig. 8) but in most instances the response was unstable. Even where stable responses were reached in an acceptably short time (< 10 min) the repra-ducibility was poor and the sensitivity was less than the theoretical 60 mV per decade. The response took the shape of the two curves on the left of Fig. 8 when the film of solution on the sensing electrode had contracted so that part of the electroactive surface was directly exposed to the atmosphere inside the air-gap electrode; in the worst instances no plateau was observed.A more stable film was obtained if the proportion of detergent was increased but this had the effect of increasing the blank reading. Fig. 9 shows calibrations obtained from two successive applications of the same film solution (1 drop of Lissapol in 500 ml) to the same Tone1 electrode. The sensitivities of 35 mV per decad 708 MARSHALL AND MIDGLEY POTENTIOMETRIC DETERMINATION Analyst VoZ. 108 (A) and 40 mV per decade (B) are typical of the range found for all the variations of the air-gap electrode tested but lower than found for the response in bulk aqueous solutions of sulphite, even though the e.m.f.s are in the same range (compare Figs. 3 and 9). Time ___+ f A / Contracting Stable films films Fig.8. Response curves of the air-gap electrode. Curves are labelled with the sulphite concentration (mg 1-l). F indi-cates the renewal of the film solution. -130 > y: E E ui -150 -170 -190 1 2 5 10 Su I p hi te concent rat ion/mg I- ’ Fig. 9. Calibration of two suc-cessive assemblies (A and B) of an air-gap electrode. T shows the theoretical slope arbitrarily placed on the e.m.f. axis. Reproducibility Fig. 9 shows how much successive assemblies of an air-gap electrode can differ in the e.m.f. produced in response to identical samples. The differences between assemblies were much larger than were obtained for repeated presentation of identical sulphite solutions to the same assembly. In later tests with a 50% m/m Nonidet film solution the standard deviations of the e.m.f.s for 10 and 5 mg 1-1 solutions were 7.1 and 8.6 mV respectively; with a sensitivity of 40 mV per decade these standard deviations are equivalent to about 50% in concentration.Resportse time With film volumes of 25 pl or less equilibrium was reached in less than 10 min (Table I), i.e. the time taken for sulphur dioxide to be distributed between the sample and the film was longer than the response time of 1-2 min for the electrode itself. The acidified sample had to be stirred; with quiescent solutions the response was very slow presumably being limited by the diffusion of sulphur dioxide in water. TABLE I EFFECT OF FILM VOLUME Volume of filmlpl 100 60 25 E.m.f. for 10 mg 1-l SOa2-/mV .. . . -152 -157 -167 Time to equilibriumlmin . . 40 20 10 Efect of fdm volume The smaller the volume of the film applied to the sensing electrode surface the quicker the response of the electrode and the more negative the e.m.f. taken up (indicating an apparently higher sulphite concentration) ; Table I shows results for the Ionel electrode with films of 1% Nonidet + 10-5moll-1 of Hg,(NO& The sensitivity over the range 1-10mg1-1 varied between 30 and 45 mV per decade but could not be correlated with the film volume. The changes observed would (qualitatively) be expected for equilibrium distributions between a fixed source volume and a variable receiver volume but the film volume was observed to affect the e.m.f. even when no gas transfer occurred (in tests with chloride solution).With applications of 100-p1 volumes of 0.1 mg 1-1 Cl- solution to the electrode surface a steady reading could not be obtained the e.m.f. reaching a maximum of about -70 mV in 2 min before becoming more negative. With 200- and 400-pl applications steady e.m.f.s of abou June 1983 OF SULPHITE BY FLOW INJECTION ANALYSIS 709 -63 mV and -52 mV were obtained. The sensitivity of the electrode to 200-4 films of chloride solution was less than that in bulk solution the difference in e.m.f. being 41-43 mV compared with a range of 50-54 mV over the range 0.1-1.0 mg 1-1 of C1-. The results with chloride solutions show that using small volumes of solution is in itself sufficient to reduce the sensitivity of the electrodes by as much as 50% which closely parallels the difference between the sulphite response in solution and with air-gap electrodes.Detergent content of Elm solzttion In order to be able to achieve a stable covering of the electrode surface with a small (5-25 pl) volume of film solution a high detergent content was required (up to 50% m/m). As the pro-portion of detergent in the film solution increased the “blank” electrode potential i e . before absorption of any sulphur dioxide became more negative as in Table 11. The main cause of this shift was inferred to be a chloride impurity in the detergent; an e.m.f. of -60 mV in bulk solution is normally indicative of 0.1-0.2 mg 1-1 of chloride. Adding mercury(1) nitrate to the detergent solution should precipitate the chloride so that when an exactly equivalent amount is added the resultant solution should contain the same amount of chloride as a pure aqueous solution.The addition of 67 pl of mol 1-1 of Hg,(NO,) per gram of Nonidet was required to produce the same e.m.f. (-40 mV) in solutions containing 0 and 50% m/m Nonidet indicat-ing that the Nonidet contained about 50 pg g-l of chloride. This result is subject to error because the Nonidet may influence the e.m.f. not only through chloride contamination but by changing the dielectric constant and viscosity of the solution which would affect the activity coefficients and liquid-junction potential respectively. TABLE I1 EFFECT OF NONIDET CONCENTRATION Nonidet % / V / V . . . . 0 0.1 1 10 20 50* 50**t Blank e.m.f./mV . . -40 -43 -50 -60 -75 -150 -40 Sensitivity/mV per decade .. . 30-40 30 - 28 30-40 - 45 * Nonidet % mlm. t Nonidet including Hg,(NOJ,. Design of the air-gap electrode The design with the inverted-sensing electrode was easier to use as regards acidification of the sulphite solution and application of the film solution. It also enabled tests to be carried out with relatively large (>50 p1) volumes of film solutions. The inverted arrangement gave no better sensitivity or reproducibility than the original design. The inverted design had the advantage that the surface could be inspected without dismantling the assembly so that a slow response was not confused with the drift associated with exposure of the sensing surface (Fig. 7). No significant difference was observed between the performance of Ionel and Graphic Controls sensors in the air-gap electrode.Discussion Flow Injection Analysis Sensitivity The loss of sensitivity in flow-injection analysis compared with the equilibrium performance of the detector is not unexpected as the sulphite is diluted during the gas transfer and absorp-tion stages and the actual concentration at the electrode’s surface will be smaller than in the original sample. Interference Except for hydrogen sulphide commonly occurring volatile acids do not interfere even though they can cross the FIA membrane to some extent. Because the absorbent stream is itself acidic (pH 3) most of the substances absorbed will be protonated and only those capable of reacting very strongly with the electrode material will interfere.Pre-treatment of the sample solution with lead nitrate6 or bismuth nitrate‘ should remove sulphide interference 710 MARSHALL AND MIDGLEY POTENTIOMETRIC DETERMINATION Analyst VoZ. 108 The thiosulphate ion interferes with the mercury(1) chloride electrode.8 In the 0.1 mol 1-1 nitric acid carrier about 4% of the thiosulphate should be present as the uncharged thio-sulphuric acid which may be capable of diffusing into the absorbent stream and so causing the small interference observed. Decomposition of thiosulphate to sulphite is also possible, although the solution used was freshly prepared. The strongest metal - sulphite complexes are those formed with the mercury(1) ion and as these had no effect on the size of the peaks sulphito complexes of other metal ions would not be expected to interfere either.Alkaline substances present in such concentration as to raise the pH of the carrier stream and so reduce the extent of sulphur dioxide formation would cause a negative bias but approxi-mate neutralisation of the sample before analysis would prevent this occurring. Comparison with other methods Compared with direct use of the mercury( I) chloride electrode flow injection analysis avoids the problem of the commonest interferent chloride although at the cost of raising the lower limit of Nernstian response. Any substance that interferes in flow injection analysis would also interfere in the direct method probably to a greater extent. Potentiometric gas-sensing membrane electrodes based on the principle of measuring the pH change produced on absorption of sulphur dioxide in sodium hydrogen sulphite solution, show9 similar precision and limits of Nernstian response as found for flow injection analysis with the mercury(1) chloride electrode but are more susceptible to interference by acidic gases, hydrogen sulphide excepted.The commonest method of determining sulphite is probably the colorimetric method based on the fuchsin - formaldehyde - sulphurous acid complex. The relative standard deviation of measurements by this method are 1-3% and the limit of detection is about 0.015 mg 1-1 of sulphite.1° This method has also been used in flow injection analysis with a stopped-flow stage,3 but not below 5 mg 1-l. The loss of sensitivity in the flow injection variant of the pro-cedure is not surprising in view of the 30-min development time allowed in the conventional method.Flow injection analysis is therefore much more sensitive when using a mercury(1) chloride electrode and although it is less sensitive than the conventional colorimetric procedure, it avoids the development time. Flow injection analysis with potentiometric detection is more selective than instrumental techniques commonly used for determining sulphur dioxide in air12 (e.g. coulometry conductiv-ity and flame-emission spectrometry). Mercury displacement detectionlsJ4 is very selective and much more sensitive than flow injection analysis but lacks the adaptability of a general-purpose technique. The proposed method is more sensitive than results so far reportedl5~ls for ion chromatography but these did not seem to utilise the full sensitivity available from the ion chromatograph and the comparison may be misleading.Air- gap Electrodes The air-gap electrodes examined in this work cannot make full use of the response of the mercury(1) chloride electrode to sulphite ion. The critical factor is that in order to achieve an acceptably short response time (<lo min) the volume of film solution on the electrode’s membrane surface must be small (<25 pl). The use of such small volumes necessitates the inclusion of detergent to stabilise the film on the membrane but the chloride content of the detergents tested was high enough to interfere with the sulphite measurements ; adding mercury( I) nitrate to precipitate the chloride removed this interference.The sensitivity of the mercury(1) chloride electrode with films of small volume (G400 p1) is markedly less than when immersed in much larger volumes of solution. This is so even for dilute aqueous chloride solutions i e . neither the sulphite reaction nor the detergent is the main factor in this loss of sensitivity. Marshall and Midgley4ps found that in unstirred bulk solutions mercury(1) chloride electrodes responded slowly but the equilibrium e.m.f .s differed little from those in stirred solution. In this work the response time was longer with small volumes (<400 p1) of chloride film solution than in stirred bulk solution (5-10 min compared with about 2 min) but not as long as in unstirred bulk solution (20-30 min). Although the sensing electrodes gave reproducible e.m.f.s (&-2 mV) when directly immersed in bulk sulphite solutions and when in small volumes of chloride solution e.m.f.s in the air-ga Jzcne 1983 OF SULPHITE BY FLOW INJECTION ANALYSIS 71 1 mode were very variable (&8 mV) with a standard deviation equivalent to about 50% in concentration terms.With such poor reproducibility air-gap electrodes cannot be considered a practical means of determining sulphite. Conclusion Flow injection analysis using a mercury(1) chloride membrane electrode as detector is a use-ful alternative to the established methods of determining sulphite at concentrations down to 0.1 mg 1-1 of sulphite and is more sensitive than previously reported flow injection methods for sulphite. The air-gap electrode is not a suitable means of utilising the sensitivity of the mercury(1) chloride membrane electrode to sulphite.The authors acknowledge the help of Mr. C. Gatford with the experimental work. This work was carried out at the Central Electricity Research Laboratories and is published by permis-sion of the Central Electricity Generating Board. 1. 2. 3. 4. 6. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. References Tseng P. K. C. and Gutknecht W. F. Anal. Chem. 1976 48 1996. RbfiEka J. and Hansen E. H. Anal. Chim. Acta 1974 69 129. RbfiCka J. and Hansen E. H. “Flow Injection Analysis,” John Wiley Chichester 1981. Marshall G. B. and Midgley D. Analyst 1978 103 438. Marshall G. B. and Midgley D. Analyst 1979 104 55. Frant M. S. Ross J . W. and Riseman J . H. Anal. Chem. 1972 44 2227. Sekerka I. and Lechner J. F. Water Res. 1976 10 479. Midgley D. unpublished work. Midgley D. and Torrance K. “Potentiometric Water Analysis,” John Wiley Chichester 1978, Dimmock N. A. and Goodfellow G. I. unpublished work. Scaringelli F. P. Saltzman B. E. and Frey S. A. Anal. Chem. 1967 39 1709. Forrest J. and Newman L. J. Air Pollut. Control Assoc. 1973 23 761. Marshall G. B. and Midgley D. Anal. Chem. 1981 53 1760. Marshall G. B. and Midgley D. Anal. Chem. 1982 54 1490. Steiber R. and Statnick R. M. in Sawicki E. Mulik J . D. and Wittgenstein E. Editors “Ion Chromatographic Analysis of Environmental Pollutants,” Ann Arbor Science Ann Arbor 1978, p. 141. Frazier C. D. in Mulik J. D. and Sawicki E. Editors “Ion Chromatographic Analysis of Environ-mental Pollutants,” Volume 2 Ann Arbor Science Ann Arbor 1979 p. 211. Received November 29th 1982 Accepted January 14th 1983 p. 271