Analyst, January 1996, Vol. 121 (75-78) 75 Determination of Low Concentrations of Nickel and Aluminium in Membrane Electrolyser Liquors Michael Cullen and Susan Lancashire ICI Chemical and Polymers Ltd, R & T Department, The Heath, Runcorn, Cheshire, UK WA7 4QD Differential-pulse adsorptive stripping voltammetry (DPASV) has been used to determine nickel and aluminium concentrations in electrolyser cell liquors; these are composed of both NaCl and KCI brines of various concentrations up to saturated (30% m/v) and NaOH and KOH liquors up to 32% m/m. In the analysis procedure, nickel is complexed with dimethylglyoxime at pH 8.8 k 0.1; the complex is then adsorbed on a hanging mercury drop electrode (HMDE) at -0.8 V. In a separate procedure, aluminium is complexed with 1,2-dihydroxyanthraquinone-3-sulfonic acid at pH 7.5 k 0.1 and adsorbed on the HMDE at -0.85 V.The detection limits are 0.1 pg 1-1 nickel (60 s adsorption) and 0.2 pg 1-1 aluminium (30 s adsorption) in saturated brine. The linear working range is up to 20 pg 1-1 for nickel and up to 100 pg 1-1 for aluminium. Concentrations greater than this range have been determined by taking a smaller sample volume. In both procedures, hydrazinium sulfate is used to remove interference from free chlorine prior to analysis. These methods are in routine use for the quality control of ICI's FM21 membrane electrolyser brines with nickel and aluminium concentrations typically below 5 pg 1-1 and 30 pg 1-1, respectively. Keywords: Nickel; aluminium; membrane electrolyser liquors; differential-pulse adsorptive stripping voltammetry Introduction This paper describes the methods used to detect low concen- trations of nickel and aluminium in electrolyser liquors of the FM21 membrane electrolysis cells used for the manufacture of chlorine, and sodium hydroxide or potassium hydroxide.During the normal operation of the FM21 electrolysis cells, it was found that low concentrations of trace metals, such as nickel and aluminium, in the feed brine had an adverse effect on cell performance. In order to monitor these effects, sensitive procedures for the determination of nickel and aluminium in saturated brine were required. Our initial investigations using inductively coupled plasma optical emission spectroscopy (ICP-OES) and ETAAS could not achieve the required detection limits in the presence of very high levels of dissolved solids.ETAAS could only achieve a minimum quantification limit of 50 yg 1-1, with a detection limit of 10 yg 1-1 for nickel. Aluminium in a variety of matrices has been routinely measured in our laboratory by a colorimetric method using Solochrome Cyanine R, but the procedure was not sensitive enough for this application. Work has previously been reported using adsorptive stripping voltammetry to detect nickel using dimethylglyoxime (DMG)'.* and aluminium with 1,2-dihy- droxyanthraquinone-3-sulfonic acid (DASA)1,3.4 in sea-water and freshwater. The detection limits reported for sea-water are 0.1 nmol l-1 of nickel (6 ng 1-1) and 1 nmol l-1 of aluminium (27 ng 1-1) for a 60 s deposition time.However, the procedures described could not be directly applied to membrane electro- lyser liquors as these samples often contain free chlorine, have wide variations in pH, brine strengths up to saturated NaCl (30% m/v) and caustic strengths up to 32% m/m NaOH. The aim of this work was to produce a standard method for the determination of nickel and aluminium concentration in electrolyser liquors, which would be both sensitive and precise enough to correlate with variations in electrolysis cell perform- ance. A further requirement was that a short over-all analysis time was achieved in order to allow routine application of the method to large numbers of samples. Consequently, it was important to establish the optimum conditions for the analysis of nickel and aluminium in electrolyser solutions.This paper describes adsorption stripping voltammetry procedures which can achieve detection limits of 0.1 f 0.025 pg 1- * for nickel and 0.2 f 0.05 pg 1-1 for aluminium in electrolyser feed and exit brines. External standard calibrations are performed in blank brine solutions and the interference caused by free chlorine in the electrolysis cell exit brine is removed by adding hydrazin- ium sulfate prior to addition of the other reagents. Experimental Equipment The following Metrohm (Herisau, Switzerland) systems were used: the VA646 processor, VA675 sample changer, multi- mode electrode (MME), Ag/AgCl reference (3 mol 1-1 KC1) and platinum auxiliary electrodes. The MME was set to a hanging mercury drop mode with a surface area of 0.55 mm2 and a -75 mV pulse amplitude was used unless otherwise stated.A fresh mercury drop was used for each scan. All adsorption steps were carried out with stirring using the medium stirring rate on the VA646 followed by an unstirred 10 s equilibration time. A scan rate of 10 mV s-1 was used with a 6 mV step and a pulse time of 0.6 s. All glassware was stored in approximately 10% v/v hydrochloric acid and rinsed well with de-ionized water prior to use. Reagents De-ionized water obtained via a Milli-Q water system was used in the preparation of all solutions. Working standard nickel and aluminium solutions were prepared daily by serial dilution of the nickel and aluminium 1000 mg 1-1 Spectrosol standard solutions (Merck, Poole, Dorset). The hydrochloric acid used was Aristar grade (d 1.18, Merck) and the ammonia was a PrimaR ammonia solution (d 0.88; Fisons, Loughborough, Leices ters hire).Blank saturated brine (30% m/v NaCl). This solution was prepared by dissolving 30 g of solid NaCl (Aristar grade) in 100 ml of de-ionized water.76 Analyst, January 1996, Vol. 121 Ammonium buffer (0.8 mol l-l)-DMG solution. This was prepared by dissolving 0.1 g of DMG (AnalaR Grade, Merck) in 10 ml of methanol then diluting to 500 ml with 0.8 mol 1-1 ammonia-ammonium chloride buffer (adjusted to pH 8.8). BES buffer (N,N'- bis-(2- hydroxyethy1)-2 -aminoethanesul- fonic acid), 1 mol 1-l. This buffer was prepared by dissolving 21.3 g of BES buffer solid (Aldrich, Gillingham, Dorset) in 100 ml of water, adjusting the pH to 7.5 ( f O .l ) using ammonia solution or hydrochloric acid as required and then adding 0.01 g of manganese(1v) oxide in order to remove trace metals.3 DASA solution, 1 X 10-3 moll-'. This reagent was prepared by dissolving 0.032 g of DASA (Aldrich) in 100 ml of de- ionized water. Hydrazinium sulfate (1% m/v). This was prepared by dissolving 1 g of the AnalaR grade solid in 100 ml of de-ionized water. This solid is extremely toxic and all the necessary safety measures were applied. Starch iodide paper (Merck). This was used for checking chlorine removal. Procedure for the Determination of Nickel in Electrolyser Brine Calibration Hydrazinium sulfate (0.1 ml of a 1% solution), 10.0 ml of de- ionized water and 10.0 ml of ammonia-DMG solution (pH 8.8) were added to a polarographic cell and mixed.The solution was de-oxygenated by a purge with pure nitrogen for 4 min prior to the application of -0.8 V for 60 s stirred adsorption time followed by 10 s equilibration time. A differential-pulse stripping scan was carried out to a final potential of -1.3 V. Standard additions of nickel to the cell produced a linear calibration up to 20 pg 1-1 of nickel. The nickel peak occurred at -0.99 V and a typical cell blank of 0.2 pg 1-I was found. Standard scans are illustrated in Fig. 1. Sample preparation Hydrazinium sulfate (0.1 ml of a 1% solution) and 10.0 ml of brine sample were added to a clean polarographic cell and Z a n -300 W > a 3 0 -250 >E -200 0 v) I I I -150 7 1 > -100 E 7 -50 t- W 2 0 v! = 00 Fig. 1 Typical nickel scans: 0, 1 and 10 pg 1-l spiked Ni cell concentrations.mixed; the pH was then adjusted to 1 or less by adding hydrochloric acid. Ammonia-DMG solution ( 10.0 ml) was then added and the solution pH adjusted to 8.8 f 0.1. For samples with a nickel concentration greater than the linear range, the sample was diluted with water and then 10 ml of the diluted sample were taken into the cell. The prepared sample solution was next scanned under the same instrumental conditions as used for the nickel calibration standards, allowing the nickel concentration to be determined from the calibration graph after correcting for any dilution used. Procedure for the Determination of Aluminium in Electrolyser Brine Calibration Hydrazinium sulfate (0.1 ml of a 1 % solution), 1 .O ml of blank 30% m/v NaCl solution, 18.6 ml of water and 0.2 ml of 1 moll-' BES buffer (pH 7.5) were added to a polarographic cell.DASA (0.1 ml of a 1 X 10-3 moll-' solution) was added and the cell solution de-oxygenated for 4 min using a pure nitrogen purge. An initial potential of -0.85 V with a stirred adsorption time of 30 s was then applied, followed by a 10 s equilibration time before starting the scan. A differential-pulse stripping scan was then carried out to a final potential of -1.25 V. The aluminium peak occurred at - 1.08 V with a typical cell blank of less than 1 pg 1-1 of aluminium. The calibration was linear up to 100 pg 1-1 of aluminium (for 2 ml of a 30% m/v NaCl sample) using the above conditions. Typical standard scans are illustrated in Fig. 2. Sample preparation Two millilitres of sample plus 0.1 ml of hydrazinium sulfate were added to a clean polarographic cell and then mixed to destroy any chlorine; starch iodide paper was used to check for completion of the reaction.Water (17.6 ml) and 0.2 ml of BES buffer were then added and the pH adjusted to pH 7.5 f 0.1 using ammonia solution or hydrochloric acid as required. The prepared solution was examined under the same instrumental conditions as for the calibration standards and the aluminium concentration determined using the calibration graph. For samples with concentrations greater than the linear range (100 pg 1-1) the sample was diluted with water and then 2 ml of the diluted sample were taken and analysed. The sample concen- tration was then corrected for dilution as appropriate.-70 7 Fig. 2 concentrations. Typical aluminium scans: 0, 5 and 50 pg 1-1 spiked A1 cellAnalyst, January 1996, Vol. 121 77 Procedure for the Determination of Nickel and Aluminium in Caustic Liquors Calibration The calibration procedures for nickel and aluminium in caustic liquors (KOH, NaOH) are the same as those detailed for electrolyser brine solutions. Sample preparation For 32% m/m caustic liquors, 25.0 ml of sample were pipetted into a clean 150 ml beaker and then water added to give a total volume of approximately 50 ml. Next, the pH of the solution was adjusted to pH 1 by using hydrochloric acid and the solution transferred into a 100 ml calibrated flask and diluted to the mark with water. The prepared solution was then analysed following the procedure described for electrolyser brine samples.A correction for the dilution used (X4) was applied to obtain the concentration of nickel or aluminium in the sample before acid treatment and dilution. Results Optimized Conditions The optimum conditions for the analysis of electrolyser liquors described in the procedures were chosen after investigation of the effects of pH, adsorption potential, drop size and pulse amplitude on both the Ni and A1 responses. Effect of Sodium Chloride Concentration on the Nickel Response The different types of electrolyser brine contain different concentrations of NaCl ( e . g . , cell feed brine is 30% m/v NaC1; cell exit brine is about 20% m/v NaCl). The effect of changes in sodium chloride concentration on the nickel response was investigated by carrying out nickel calibrations in firstly 10 ml ammonia-DMG solution plus 10 ml of water, and secondly 10 ml of ammonia-DMG solution plus 10 ml 30% m/v NaCl solution.In both instances the experimental conditions were -0.8 V stirred adsorption potential for 30 s, -75 mV pulse amplitude and a final pH of 8.8. Known additions of 2 mg 1-l nickel standard were made and the nickel response recorded after each addition. Little variation in the slope of the nickel calibration was found with changes in NaCl concentration. Hence, for the nickel method, an external calibration could be carried out in a solution of 10.0 ml water and 10.0 ml of ammonia-DMG solution (pH 8.8) and for the sample analysis up to 10.0 ml of 30% m/v NaCl could be used in the cell. Effect of Sodium Chloride Concentration on the Aluminium Response Solutions of sodium chloride with concentrations from 0 to 15% m/v in the cell, 0.1 ml of DASA, 0.2 ml of BES, at a final pH of pH 7.5, were studied using a 30 s adsorption time at -0.85 V.In each experiment a calibration graph was established and the gradients compared. Unlike the small effect that changes in brine strength had on the nickel response, a reduction of approximately 25% was found in the aluminium response when 1% m/v NaCl was present in the cell as compared with the response with no added NaC1. As is shown in Fig. 3, calibration curves obtained in the range 1.54% m/v of NaCl in the cell gave similar slopes and provided sufficient sensitivity for the analysis of electrolyser liquors.Above 6% m/v of NaCl in the cell, it was found that the aluminium response decreased significantly. Hence, based on these results, a brine strength range of 1 . 5 4 % m/v in the cell was chosen for the optimized procedure. Precision and Limit of Detection for Nickel and Aluminium For nickel, following the optimized procedure, the limit of detection (LOD) for saturated brine (30% m/v of NaCl) at 3 times the standard deviation of the blank was 0.1 f 0.025 pg 1-l (for n = 5). Using a nickel cell concentration of 1 pg 1-l the relative standard deviation was 1% (n = 5). For aluminium, following the optimized procedure, the LOD for saturated brine (30% m/v NaC1) at 3 times the standard deviation of the blank (n = 5) was 0.2 A- 0.05 pg 1-1. For a 5 pg 1-1 aluminium cell concentration, the relative standard deviation was 0.6% (n = 5).In both the nickel and aluminium procedures, the LOD could have easily been improved by using an increased adsorption time, but at the cost of a reduction in the linear range. Thus, the choice of optimum conditions is the best compromise between sensitivity, linear range and speed of routine analysis for nickel and aluminium in electrolyser samples. Samples of brine were spiked with 5 pg 1-1 additions of both nickel and aluminium and analysed by the optimized proce- dures. From this, the over-all recovery of both methods was found to be > 95%. Typical Results A set of typical membrane electrolyser sample results for nickel and aluminium using the optimized procedures is given in Table 1.Interferences For electrolyser brine analysis, the most significant interference in both the nickel and aluminium methods is from dissolved chlorine. This was easily removed by the addition of hydrazin- ium sulfate which, at the levels described in the procedures, did not have any noticeable effect on the response of either method and could be omitted if the samples were known not to contain free chlorine, bromine or iodine. 100 80 a 5 60 c m al c Y a .- 40 20 0 a / 0 2 4 6 8 10 12 14 16 Al concentration/pg I-’ Fig. 3 NaC1. Calibration curves for A1 in (a) 0, (b) 1.5, (c) 3, (d) 6, (e) 15% m/v Table 1 Typical membrane electrolyser results Sample No. Sample matrix Al/pg 1-1 Ni/pg 1 - 1 1 30% m/v NaCl 60 12 2 18% m/v NaCl 30 3 3 30% m/v NaCl 10 3 4 32% m/m NaOH 50 100 5 32% m/m NaOH 80 1000 6 32% m/m NaOH 80 8078 Analyst, January 1996, Vol.121 A high level of Zn is a potential interferent in the aluminium method but it can be masked by the addition of EDTA.3 However, no interference was seen at the levels of Zn (< 0.05 mg 1-1) normally found in membrane cell electrolyser brine. Components commonly found in electrolyser brine and which do not interfere are Ca at 0.05 mg l-l, Mg at 0.05 mg l-l, 0.5 mg 1-1 of Sr, 1 mg 1-1 of Ba, 10 mg 1-1 of soluble silica, 10 mg 1-1 of total I, 10 mg 1-1 of sulfate, 0.5 mg 1-1 of Hg and 0.05 mg 1-1 of Mn, Sn, Ti or Pb. Electrolyser cell exit brine can contain up to 20 g 1-1 of sodium chlorate, 1 mg 1-1 of fluoride and 1 g 1-1 of active chlorine, but following the procedures described in this paper no interference was observed. Conclusions The methods described above have been found to be suitable for the determination of nickel and aluminium in membrane electrolyser liquors and are routinely applied in our laboratory for monitoring their effect on FM2 1 electrolyser performance. In addition to NaCl brines and NaOH liquors, these methods have also been successfully applied to various KCl brines up to 30% m/v of KC1, and KOH electrolyser process liquors. The authors thank the members of the ICI Electrochemical Technology Business in their support of this work. References 1 van den Berg, C. M. G., Chem. Oceanogr., 1988,9, 197. 2 Pihlar, B., Valenta, P., and Nurnberg, H., J . Electroanal. Chem., 1986,214, 157. 3 van den Berg, C. M. G., Anal. Chim. Actu, 1986, 188, 177. 4 Hemhndez-Brito, J. J., Gelado-Caballero, M. D., Ptrez-Pefia, J., and Herrera-Melirin, J. A., Analyst, 119, 1994, 1593. Paper 51033563 Received May 25, 1995 Accepted September 19, I995