48 PRISCOTT, HAND AND YOUNG: ANALYSIS OF [Analyst, VOl. 91 The Analysis of Electrolytic Capacitor Electrolyte The Determination of Chloride and Sulphate in the p.p.m. Range BY B. H. PRISCOTT, T. G. HAND AND E. J. YOUNG Fordrough Lane, Birmingham, 9) (Birmingham Materials Section, Test and Inspection Branch, Post O$ce Engineering Department, A method is described for the determination of chloride and sulphate ions in ethylene glycol - ammonium borate based electrolytes from electrolytic capacitors. The chloride concentration is determined by titration with electrogenerated silver ions by differential electrolytic potentiometry, and the sulphate concentration by titration with barium in the presence of Thorin with a photoelectric titrator. Concentrations of 0.3 p.p.m. of chloride and 0.5 p.p.m.of sulphate in the final solution have been satisfactorily determined. Both procedures have applications in a wide range of analyses. The electrolyte is extracted with 80 per cent. ethanol. MODERN telecommunications practice makes high demands on the complex equipment used, and high-reliability, long-life components are therefore imperative for inaccessible equipment such as submerged repeaters in transoceanic cables. The electrolytic capacitors used in such systems are of the conventional type in which aluminium-foil electrodes are separated by paper impregnated with a solution of ammonium borate in ethylene glycol.lY2 It is known that sulphate and chloride ions seriously reduce the reliability of such capacitors when present in concentrations exceeding 10 to 20 p.p.m.As the amount of electrolyte in one capacitor may be quite small (about 1 g) a procedure capable of determining 3 pg or less of each ion is therefore required. The use of barium and mercurous chloranilates3s4 was found to give high blanks and erratic results at these low levels, although they were found satisfactory for other applications in which higher concentrations were encountered. A procedure based on the chloride - mercury - diphenylcarbazone system has been proposed by KemulaJ5 but was not studied in detail. Gravimetric and nephelometric procedures are not applicable as the solubilities of the usual precipitates of silver chloride (1 pg ml-l) and barium sulphate (2 pg ml-1) are too high, and the procedures presented here rely on ion association in solution rather than in a precipitate.EXPERIMENTAL DETERMINATION OF SULPHATE- For this determination the procedure due to Fritz and Yamamura6 was studied, in which the sulphate ions are titrated with barium ions in 80 per cent. ethanolie solution with Thorin as a visual indicator. An E.E.L. photoelectric titrator is used to detect the end-point. Initial trials with 0.01 N barium chloride solution in an Agla micrometer syringe showed that the maximum response was obtained with an Ilford 604 green filter (with peak transmission at 520 mp). During the titration the added barium ions first associated with the sulphate ions and when all of the sulphate had reacted, the barium ions formed a coloured complex with the indicator, producing a change in instrument reading, the end-point being the point of inflection of the curve of instrument reading plotted against volume added shown in Fig.1 (a) and ( b ) . When aliquots of standard sodium sulphate solution were added to 80 per cent. ethanol it was possible to make reproducible titrations down to concentrations of 0.5 p.p.m. in each of the two vessels supplied with the instrument, representing total sulphate contents of 1.5 and 10 pg in the small and large vessels, respectively. At the lower concentration levels there was considerable delay before equilibrium was reached and in order to prevent the titration from becoming too lengthy, readings were taken at 60-second intervals followed immediately by an addition of a further aliquot of titrant to the solution.With similar titrations made in methanol solutions no end-point was detected. Titrations were next made with sulphate added to a synthetic capacitor electrolyte consisting of 20 g of ammonium borate, 45 ml of ethylene glycol and 15 ml of distilled water.January, 19661 ELECTROLYTIC CAPACITOR ELECTROLYTE 49 The ammonium borate was prepared by distilling analytical-reagent grade ammonium hydroxide into a solution of analytical-reagent grade boric acid until strongly alkaline, and then evaporating and crystallising it. Aliquots of sulphate-containing electrolyte were made 80 per cent. with respect to ethanol, and cations were removed by passage through a small column of Amberlite IR 120 resin in the hydrogen form when it was found possible to make titrations down to 0-5 p.p.m.of sulphate. When, however, attempts were made to cover smaller total contents by concentrating the solution by evaporation, poor results were obtained, as shown in Fig. 1 (c) because of the high glycol content that was obtained. This was overcome by evaporating the solution nearly to dryness (after the addition of a few milligrams of potassium carbonate) to remove the bulk of the glycol and taking up the residue in 80 per cent. ethanol. It was then found possible to titrate down to a total content of 1 to 2 pg as shown in Fig. 1 (b) and Table I. t I 1 1 1 1 1 1 ' 1 " 0. 0.0 I 0.02 0.01 N barium chloride solution, ml Fig. 1. Titration of 0.01 ml of 0.01 N sulphate ions with barium chloride: curve A, in 80 per cent. ethanol; curve R, after the removal of glycol; curve C, in the presence of glycol THE DETERMINATION OF CHLORIDE- For the determination of chloride, the application of the technique of differential elec- trolytic potentiometry (D.E.P.) due to B i ~ h o p ~ , * ~ ~ ~ ~ O was studied, in which he has shown that titrations are feasible several orders below the levels concerned here.The theory and practice of this technique have been adequately described7,* and only the basis of the procedure will be given here. If in the titration of chloride ion in a suitable solvent by silver ions, two silver electrodes are immersed in the solution and a small current is passed between them from a constant- current high-impedance source, only a small potential difference exists between them, before and after, the end-point.A t the end-point, however, potentials of several hundred millivolts can be developed and the curve obtained, by plotting titrant added against potential difference, resembles the first derivative of a normal titration curve. This technique has been satis- factorily used for the determination of chloride ions of much lower concentrations than those at which a normal potentiometric titration would give an erratic response, and Bishop has shown that the electrode equilibrium is attained more rapidly when the titrant is electro- generated than when discrete additions are made from a microburette. The electrodes were prepared from 16-s.w.~. silver wires cast in pairs in epoxy resin, cut off square, ground and polished flat so that only the cross-sectional area of the wire was exposed.Silver and silver chloride electrodes were used in the initial trials and both were found to give satisfactory titration curves, the silver chloride electrode giving the higher response. Under the conditions required in the presence of ammonium salts the chloride coating spalled off fairly rapidly and plain silver electrodes were adopted as the standard. These electrodes require activating before use and cleaning with fine emery, only immersing then in 50 per cent. nitric acid for 30 seconds, washing and storing them in distilled water until required for use was found satisfactory.50 PRISCOTT, HAND AND YOUKG: ANALYSIS OF [Analyst, Vol. 91 The silver ions were electrogenerated from a ring anode of 16-s.w.~. silver wire placed symmetrically around the indicating electrodes.The cathode was a copper wire - copper nitrate half-cell separated from the analytical solution by an agar bridge. A current of 45.3 pA, equivalent to 1.0 pg minute-l, was used. As there is no common point between the generation and measuring circuits, high insulation of the generator battery and control equipment from the measuring circuits is imperative, and the battery was placed on rubber bungs to meet this requirement. The electrical circuits are shown in Fig. 2. A = Copper nitrate half-cell B = Indicator electrodes C = Silver anode M = Meter, 0 to 50 pA R, = 5000-megohm resistor R, = I-megohm resistor VR = I-megohm variable resistor Fig. 2. Differential electrolytic potentiometry circuits Trials showed that the optimum differentiating conditions were obtained by using a 112.5 volt supply with a ballast resistor of 5 x lo9 ohms giving a current density of 1.08 x amp cm-2.The electrode potentials were measured with a Pye Dynacap pH meter and recorded on an Evershed and Vignoles recording milliammeter with a 34-inch wide scale and a chart speed of 0.1 inch per minute. With this procedure it was found that good results were obtained down to at least 0.3 p.p.m. in 80 per cent. ethanol made 0.001 N in nitric acid, and that a greater electrode response was given in methanol, as reported by Bishop. In order to obtain a procedure compatible with that for the sulphate determination, the use of ethanol was adopted for the work reported here. It is possible to determine smaller concentrations of chloride by varying the differentiating current, but the level reached was adequate for the present purpose.Additions of boric acid and ethylene glycol in the amounts expected from capacitors were found to have no measurable effect on the titration. With ammonium borate, however, the nitric acid concentration had to be increased to 0-1 N before satisfactory curves were obtained. Under these conditions spalling of the surface of one electrode was found t o occur, but this was minimised by reversing the polarity of the electrodes after each deter- mination. EXTRACTION OF SAMPLE FROM CAPACITOR- As the concentration of sulphate and chloride ions in the electrolyte is required, the complete extraction of the liquid material from the capacitor is not essential as long as no selective extraction occurs.It was found that 80 per cent. ethanol was a satisfactory solvent and that three extractions removed substantially all the titratable ions. Sulphate determinations on each of three extracts were found to give concentrations of 2.80, 3-00 and 3.03 p.p.m., the electrolyte extracts being 4.0, 3.0 and 1.95 g, respectively.January, 19661 ELECTROLYTIC CAPACITOR ELECTROLYTE 51 C o N s E c r T I J- E D E T E K M I K AT I o N o F c H L o R I D E AN D s u L P HATE- V'hile large capacitors yield sufficient material for the two determinations to be carried out in duplicate on separate aliquots of the extract, small capacitors require that both the determinations be made on a single aliquot, the silver that was added during the chloride titration having been removed with other cations by the ion-exchange step prior to the sulphate determination.A solution containing 2.0 pmoles of sulphate gave a titre of 0.0195 ml of 0.01 x barium chloride and a similar solution after a chloride titration corresponding to 50 pg of chloride gave a titre of 0.0197 ml, showing the consecutive determination of the two 10115 to be satisfactory. REAGENTS- 1IJZlHOD Etliaiiol, 80 per ceitt. Bariziin chloride, 0-01 s-Dissolve 1.2215 g of analytical-reagent grade barium chloride in distilled water and make up the volume to 1000 ml. Thoriiz idicntor--A 0.05 per cent. w/v solution of Thorin [2(Hydrox?.-3,6-disulpho- 1-naphthylazo) - phenylarsonic acid sodium salt: in distilled water. This solution is unstable and .hould be prepared daily.Sitvic acid-B.D.H. transktor grade or equivalent. A irzberlite 111120 resi7i-Aiial3.tical-rea~ent grade. APP.IRATVS- Photoelectric titrntor-That supplied by Evans Electroselenium Ltd. was found to be satisfactory. The glass vessels supplied have working volumes of 3 and 20 ml. pH wzetev-A Pye Dynacap meter was found satisfactory but, any pH meter or valve voltmeter with an input impedance of more than Chart recorder-An Evershed and l'ignoles hfurday recorder with a range of 1 mA full- scale deflection, 3000-ohms resistance and a chart speed of 0-1 inch minute-l was satisfactor!-. Agla micrometev s y i q e or similar equipment. Ion-exclzaiige coZzinz~z-A glass tube approximately 5 inches long by Q inch diameter with a tap at the lower end is suitable for this column.A small piece of quartz wool placed above the tap retains the resin within the column, and a suspension of Amberlite IR120H analytical-grade resin in water is poured in to a depth of about 8 inch. Wash the resin twice with water, twice with 80 per cent. ethanol and allow to drain. With care in construction the bed volume and dead space can be kept to a minimum so that washing volumes can be small. The resin is re-activated by stirring it with N hydrochloric acid for 5 minutes, then filtering and washing it until it is free from chloride. Store it under water until required. DiferePztiating circz~it-~4 battery of approximately 100 volts and stable resistance of 5 x lo9 ohms are used. Megistors obtained from the Morgan Crucible Co.are suitable. The circuit is built into a metal screening container and all external connections are made with co-axial cable, the outer conductor being used as a screen. Insulation of the ballast resistor from the core is critical as many materials regarded as insulators are of the same order of conductivity as the ballast resistors. Similarly, the insulation of the change-over switch must be good. Electrugeszeratiw circuit-Any batter>. - resistance combination giving a constant electro- generative current may be used. A high potential and a high resistance provide niore stable conditions. The effective insulation of all parts of this circuit (except the electrodes in the solution with the indicator electrodes) from the differential circuit is essential to prevent spurious potentials from being indicated on the pH meter.The circuit is shown in Fig. 2. DiffeerentiaL electrodes-Two 1-inch lengths of 16-s.w.g. silver wire are soldered on to lengths of tinned-copper wire, and cemented by means of Araldite AVlOO resin into two shallow grooves cut longitudinally in a 3 x 3 x +-inch piece of Perspex such that the free end of the silver wire slightly projects past the edge of the strip. When the resin is cured, the wires are covered with a layer of resin 9-inch deep, free from air bubbles, and the resin is allowed to cure fully. The end of the assembly is then ground-off square and polished wet, up to 500-mesh carborundum paper. The copper wires at the other end of the strip are used to make connection to the differentiating circuit through coaxial cable. The electrodes are activated by immersing them in 50 per cent.nitric acid for 30 seconds, then washing and storing them in distilled water until required. When the electrodes become inactive in use they are polished on fine abrasive paper and re-activated as described. ohms may be used. The circuit is shown in Fig. 2.52 PRISCOTT, HAND AND YOUNG: ANALYSIS OF [Analyst, VOl. 91 Electrogenerative electrodes-A glass tube 3 x & inches is half filled with a hot 10 per cent. solution of agar containing 5 per cent. of analytical-reagent grade potassium nitrate. When the agar is cold and set, a 20 per cent. solution of analytical-reagent grade copper nitrate is placed in the empty end until this is two-thirds full.A copper wire is immersed in the copper nitrate solution, and the whole mounted vertically in a suitable holder. This forms the cathode half-cell with its connecting agar - salt bridge. The anode is made by bending a length of 16-s.w.~. silver wire into the form of a horizontal loop with a vertical connecting length. Electrode stand-The electrodes are mounted on a suitable stand constructed from Perspex sheet. A symmetrical arrangement is essential to minimise any potential induced in the measuring circuit by the generating current. PROCEDURE- Extraction of eZectroZyte-Cut, or grind away the metal case of the capacitor, carefully extract the inner coil and weigh it on a watch glass. The coil is then loosened to allow ready ingress of solvent, and the capacitor coil is extracted with the minimum amount of 80 per cent.ethanol in a suitable vessel. Three extractions are generally sufficient, after which the coil is dried on the watch-glass in an air oven at a temperature of 105" C, cooled and re-weighed. The loss in weight is taken as the weight of the electrolyte. Determination of chloride-A suitable portion of the capacitor electrolyte in 80 per cent. ethanol is placed in a small squat beaker together with a stirrer magnet. Sufficient N nitric acid is added to the solution to make it 0.1 N with respect to nitric acid. The electrode assembly is introduced, the stirrer started and the differential circuit switched on. After about one minute, during which the potential settles down, the electrogenerative circuit is switched on at 45.3 WA and the chart recorder is started simultaneously. The recorder now plots the course of t'he titration and a characteristic peaked curve I 0 4 8 1 2 1 6 ; Time, minutes 0 I " ' 1 " 1 " is produced, as shown i& Fig.3. 1 Fig. 3. Differential clcctrolytic potentiometry titration curve (titra- tion of 9 pg chloride ions at 45.3 FA) The comparatively high nitric acid concentration used causes spalling-off of the surface film on one electrode after prolonged use, which gives rise to a high differential-electrolytic poten- tiometric base potential. This can be avoided by reversing the indicator electrode polarity between determinations. The time taken to reach the peak of the curve is measured. The peak is rather rounded and the best value is obtained by considering the mean position of the comparatively flat portion at the top.At 45.3 pA, each minute represents 1 pg of chloride ion. Determination of sul$hate-A suitable volume of the extracted electrolyte obtained is taken and a few milligrams of potassium carbonate are added. The solution is carefully evaporated nearly to dryness in a small squat beaker and then cooled. A small amount of water is added to the solid and boiled to dissolve the residue and to rinse the sides of theJanuary, 19661 ELECTROLYTIC CAPACITOR ELECTROLYTE 53 beaker. When the solid is dissolved, the solution is cooled and made 80 per cent. ethanolic. The solution is then passed through the cation-exchange column, the effluent being collected in the small titration vessel. The column is washed twice with 80 per cent.ethanol. Since the column is not working as a true chromatographic column it can be drained before the sample is introduced and after each washing. The solution is then placed in the titrator and an Ilford No. 604 filter is inserted in the instrument; the stirrer is started and Thorin indicator is added to the solution until the instrument, with the sensitivity turned to maximum and the zero control turned almost fully clockwise (i.e., the condition of maximum sensitivity of the instrument), reads 20 divisions on the upper red logarithmic scale. The microburette is filled with 0.01 N barium chloride solution and fitted so that the tip of the jet is just above the surface and can be just dipped into the surface to introduce each increment of titrant.Titration is carried out by introducing equal increments, usually 0.001 ml, at fixed-time intervals, normally of one minute, and reading the galvanometer at the end of each minute until the reading has risen to a high value (about 50 to 80). A graph is then constructed of galvanometer reading against titration, and the end-point is determined from the point of inflection of the curve. Consecutive determination of chloride and sulphate-The chloride concentration is deter- mined by the method described, and after completion of the titration the electrodes and magnetic stirrer are removed, washed down into the vessel with 80 per cent. ethanol andthe solution is used for the determination of sulphate as described before. REPRODUCIBILITY A synthetic capacitor electrolyte was prepared containing 20 g of ammonium borate in 100 ml of ethylene glycol.Additions of chloride or sulphate were made to 10-ml aliquots of the electrolyte which were then diluted to 100 ml with 80 per cent. ethanol and the additions were determined by the procedures described above. The results obtained are shown in Table I. TABLE I DETERMINATION OF CHLORIDE AND SULPHATE IN A SYNTHETIC CAPACITOR ELECTROLYTE Concentration, p.p.m. Coefficient of r A \ variation, Addition Added Found mean per cent. Sulphate . . , . 2.88 1.68, 1.78, 1-94, 1.94 1-84 6-8 Sulphate . . . . 0.94 1.019, 0.966, 0.966, 0.815, 0.938 9.4 0.815, 0.966, 1.019 Chloride . . . . 0.70 0.98, 0-95, 0.97, 1.04, 0.99 3.2 Chloride . . . . 0.35 0.60, 0-68, 0.64, 0.62, 0.62 6.1 0.99, 1.01 0.60, 0.57 The results show that the reproducibility is adequate for the present application, the difference between the two chloride figures of 0.37 p.p.m.agreeing with the difference of 0.35 p.p.m. in the amounts added. CONCLVSIONS It has been shown that the electrolyte can be extracted from capacitors with 80 per cent. ethanol and good yields were obtained by three extractions, the contaminant-to-elec- trolyte ratio remaining constant. The sulphate can be satisfactorily titrated in 80 per cent. ethanol by barium chloride with Thorin indicator on a photoelectric titrator. In methanol solutions or in ethanol solutions containing more than 5 per cent. ethylene glycol, poor results are obtained. The interference from glycol can be removed by evaporating the solution nearly to dryness in the presence of a few milligrams of potassium carbonate and taking up the residue in 80 per cent.ethanol. The interference from the cations is removed by passage through Amberlite IR120 resin in the hydrogen form. The chloride can be determined by differential electrolytic potentiometry titration by electrogenerated silver, by using the electrical conditions described above. Ethylene glycol and boric acid have no effect on the titration but in the presence of ammonium salts the nitric acid concentration must be increased from 0.001 to 0.1 N. This higher acidity causes spalling of silver chloride electrodes and plain silver electrodes are to be preferred. The54 PRISCOTT, HAND AND YOUNG [Analyst, Vol. 91 slight surface spalling encountered was obviated by reversing the electrode polarity between each determination.The consecutive determination of chloride and sulphate has been achieved with good results and the reproducibility is found adequate for the present purpose. These procedures are of general application and have been used for other determinations of halides and sulphate at low concentration levels (the latter preferably in solutions half saturated with boric acid) such as sulphur in nickel alloys, residues on printed-circuit boards and water analysis. Acknowledgement is made to the Engineer-in-Chief of the General Post Office and to the Controller of Her Majesty’s Stationery Office for permission to publish this paper. REFERENCES 1. 2. 3. 4. 5. 6. 7 . 8. -,- , Ibid., 1962, 87, 845. 9. Dunimer, G. TV. A., “Fixed Capacitors,” Pitnian, 1956, p. 125, 3lcKnight Deely, P., “Electrolytic Capacitors,” Cornell - Dubilier Electric Corpn., New Jersey, Brrtaloccini, K. J., and Barney, J . E., .I>zalyf. Clzcnl., 1958, 30, 202. Klipp, R. \V., and Barney, J. E., Ibid., 1959, 31, 596. Kemula, TV., Hulanicki, A., and Janowski, A., Talanta, 1960, 7, 65. Fritz, J . S., and Yamamura, J. S., drzalyt. Climz., 1955, 27, 1461. Bishop, E., and Dhaneshwar, R. G., -4i?nl~~st, 1962, 87, 207. Bishop, E., Dhaneshwar, R. G., and Short, G. D., i n \Test, P. \V., lfacdonald, A\. nl. G., and \T7est, T. S., “Ahalytical Chemistry 1962 : The Proceedings of the lnternational Symposium, Birmingham, in Honour of Fritz Fcigl,” Elsevier Publishing Company, -41nsterdan1, London and Sew I’ork, 1963, p. 236, 1938, p. 68. 10. Bishop, E., and Ilhaneshwar, I<. G., Analyt. Chrtn., 1964, 36, 726. Received iwavch 21id, 1965