ANALYST, JANUARY 1993, VOL. 118 47 Simultaneous Determination of Nickel(ii) and Cobalt( 11) by Square-wave Adsorptive Stripping Voltammetry on a Rotating Disc Mercury Film Electrode Anastasios Economou and Peter R. Fielden" Department of Instrumentation and Analytical Science, UMIST, P.O. Box 88, Manchester, UK M60 IQD Adsorptive stripping voltammetry was used for the simultaneous determination of Nil' and Colt; the metal ions were complexed with dimethylglyoxime and the complexes were adsorbed on the surface of a glassy carbon rotating disc electrode, which was pre-plated with mercury. The stripping step was carried out by using a square-wave potential-time excitation signal. An electrochemical cleaning of the mercury film was employed, enabling the same mercury film to be used for a series of measurements.The limits of detection were 12 ng 1-1 for Co" and 14 ng 1-1 for Nil1 (for 120 s preconcentration), the relative standard deviation was typically about 2-3% (at the 1.2 pg I-' level) and linearity held for a concentration range of at least two orders of magnitude. Keywords: Nickel; cobalt; adsorptive stripping electrode voltammetry; square-wave voltammetry; mercury film Adsorptive stripping voltammetry (AdSV) has been estab- lished as a reliable trace analysis technique, especially over the last decade. 1.2 Traditionally the hanging mercury drop electrode (HMDE) has been used as the working electrode together with chemically modified electrodes or solid electrodes in special cases. However, mercury film electrodes (MFEs), which are prepared by electroplating a thin film of mercury on a suitable substrate,3 have a number of advantages compared with the HMDE.(a) More precise and controllable mass transfer to the electrode surface can be achieved by the use of an MFE (by rotating it or by incorporating it in a flow system) than with the stirring used in combination with the HMDE.4 ( b ) Mercury film electrodes are simpler to construct, to maintain and to control than the HMDE.4 (c) Mercury film electrodes are more stable in flow systems than the HMDE (especially at high flow rates)3 and are less seriously affected by vibrations (as with ship onboard operation). (d) Both the HMDE and the MFE have renewable surfaces. While the surface of the HMDE is rapidly and reproducibly renewed by using a new mercury drop for each measurement, the surface of an MFE can be regenerated either electrochemically or by plating a new mercury film.Although renewing the surface of an MFE is more time consuming, precision is high, especially if the operation is automated.5 Linear sweep (d.c.) voltammetry and differential-pulse (DP) voltammetry have so far dominated the stripping process. Square-wave (SW) voltammetry offers a number of advantages over the d.c. and DP excitation signals.6 (a) The background current is well suppressed compared with d.c. voltammetry, resulting in a flat baseline. (b) The sensitivity is high, expecially for reversible reactions. (c) The analysis time is shorter than when DP is used. (d) Oxygen interference is minimized. ( e ) Commercial availability for instruments cap- able of SW voltammetry is now wide, as most modern potentiostats offer an SW choice in their repertoire.In two early applications Eskilsson et aL.5 and Brett et aL.7 made use of flow systems to determine Ni" and CO" using chronopo ten tiome try and differen ti al-pul se vol tamme try, respectively. In this work the suitability of the rotating disc MFE for square-wave adsorptive stripping voltammetry (SWAdSV) is for the first time assessed and finally demon- strated for the simultaneous determination of Nil' and Co". * To whom correspondence should be addressed. Experimental Instrumentation An E G & G Princeton Applied Research (PAR) Model 273 potentiostat/galvanostat controlled by an ARC Proturbo 286 PC through the Model 270 electrochemical software was used for all measurements.The data were collected on the hard disc of the PC and plotted on a Roland plotter connected to the computer. A 10-point moving average filter was used to smooth the data. pH measurements were made with a Kent EIL 7045/46 pH meter calibrated with standard buffer solutions (at pH 7 and 9). The rotating disc electrode (RDE) assembly was a PAR Model 616 RDE with a glassy carbon (4 mm in diameter) working electrode. The analytical cell was a 50 ml vessel, the reference electrode was an Ag-AgC1 (sat. KCl) electrode and the auxiliary electrode was a Pt wire. A Model 303A HMDE (PAR) was also used for comparison purposes. Reagents and Glassware All reagents were of AnalaR grade. The water was doubly distilled. Standard Co" solutions were prepared daily from a 1000 ppm BDH atomic absorption standard solution.Stan- dard Ni" solutions were prepared from a 600 ppm solution [prepared by dissolving the appropriate amount of Ni(N03)2 in water]. The supporting electrolyte was an NH3-NH4CI buffer, pH 9, with 0.1 moll-' total NH3. AO.l moll-1 dimethylglyoxime (DMG) solution was prepared by dissolving the appropriate amount of DMG in 95% ethanol. The mercury plating solution was 1 mmol 1-1 Hg" in 0.1 mol 1-1 KN03-0.01 mol 1 - 1 HN03. The flasks used for the standards were soaked in 2 moll-' HCl for 1 week, thoroughly rinsed with water, filled with the appropriate standard solution, left to equilibrate with it for 1 week and were then ready for use; in this way adsorption on the walls of the flasks was avoided.The cell was kept filled with 6 mol 1-1 HN03 between successive analyses for at least 30 min. For the calculation of the limit of detection, for the determination of Ni and Co in iron and for calculation of accuracy, the water was passed through an Elgastat UHQ water purification system, Aristar grade chemicals were used for the supporting electrolytes and the supporting electrolytes were further purified by equilibration with 1 x 10-4 mol I-' Mn02 solution overnight followed by filtration.848 ANALYST, JANUARY 1993, VOL. 118 Procedure Preparation of the electrode The electrode was polished successively with 600 and 800 grade metaliographic grinding paper, diamond paste and 0.3, 0.075 and 0.015 pm aluminium oxide until a scratch-free and mirror-like surface was achieved.The electrode was rinsed with acetone and water and was then ready for use. When not in use the electrode was kept in the supporting electrolyte. Repolishing of the electrode was carried out when its performance deteriorated or when its surface was deactivated by accidental overpolarization. Results and Discussion Mercury film plating The mercury film was plated from the 1 mmol l-1 solution for 2 min on the glassy carbon working electrode at - 1 .O V (versus Ag-AgCI) and at a rotation speed of 10 Hz. Preparation of the iron sample A 0.250 g amount of high-purity iron was dissolved in 10 ml of 4 mol 1-1 HN03, 0.5 g of potassium tartrate were added (in order to complex Fell') and thc solution was diluted to 1 1. For the voltammetric analysis 5.0 ml of this solution and 45.0 ml of supporting electrolyte were placed in the cell, the pH was adjusted to 9 with concentrated ammonia solution and the sample was ready for the analysis. A blank solution not containing the sample (iron) was prepared in exactly the same way.Determination of Ni" and Co" The blank solution (50 ml) was de-aerated for 10 min, 500 pl of the 0.1 mol 1-1 DMG solution were added (to give a final DMG concentration of 1 X 10-3 mol 1-I), the preconcentra- tion was carried out at -0.7 V (versus Ag-AgC1) at a rotation speed of 10 Hz, the solution was left to equilibrate for 15 s and the analytical current-potential (i-E) response was recorded. The solution was blanketed with Ar during the preconcentra- tion, equilibration and measurement steps.Cleaning of the mercury film The mercury film was cleaned of the remaining adsorbed complexes by keeping the potential of the electrode at -1.2 V (twsus Ag-AgCl) for 60 s; at this potential the complexes of Ni" and Co" with DMG are exhaustively reduced. After a series of measurements the mercury film was removed by wiping the electrode with a wet tissue. 1.400 1.200 1 .ooo Q 0.800 0.400 0.200 Eo'600E -0.200 0.000 -0.600 -0.700 -0.800 -0.900 -1.000 --1.100 -1.200 -1.300 EN versus Ag-AgCI Fig. 1 Two successive cyclic voltammograms for a solution contain- ing 6 pg 1-' of Col' and 10 pg I-' of Nil1 after adsorptivc accumulation of their complexcs with DMG on an MFE. Conditions: preconcentra- tion timc, 60 s; prcconcentration potential, -0.7 V (versus Ag-AgC1); electrode rotation speed.10 Hz; scan rate, 100 mV s-l; supporting clcctrolytc. NI-14Cl-NH3 (pH 9); and DMG concentration, 1 mmoll-1 Cyclic Voltammetry Two successive cyclic voltammograms of 10 pg I-' of NilL and 6 pg 1-1 of Co" in the presence of 1 x 10-3 mol 1-1 DMG are shown in Fig. 1 after preconcentration for 60 s. The peaks at -1.03 and -1.15 V arise from the reduction of Nil1 and Co", respectively, in their complexes with DMG which are adsorbed on the MFE. The absence of a peak in the anodic branch of the cyclic voltammograms indicates that the reduction of the complexes is an irreversible process. Irrever- sibility is also implied by the shift of peak potentials to more negative values on increase of the scan rate (a negative shift of 60 mV was observed for both peaks when the scan rate was increased from 20 to 500 mV s- 1).On the other hand the small peaks for the second scan indicate that the reduction of the metal complexes is exhaustive during the first scan. Comparison Between the MFE and the HMDE Two typical square-wave stripping voltammograms for the same sample containing Nil' and Co" adsorbed on an MFE and on an HMDE are shown in Fig. 2. Despite the fact that d.c. stripping gives better results with the HMDE than with the MFE, for SW stripping both voltammograms exhibit good resolution, excellent background rejection, comparable sensi- tivity (after normalization of the electrode area) and a similar linear range. Linearity and Limit of Detection The calibration graph for the simultaneous determination of Nil1 and Co" at the same concentration is linear with the linear range and sensitivity depending on the preconcentration time.For 60 s preconcentration the calibration graph is linear up to 3.500 i I 3.000 2.500 < 2.000 5 1.500 1.000 a 0.500 - B -0 500 I I I I I I I I I -0.800 -0.900 -1.000 -1.100 -1.200 -0.850 -0.950 -1.050 -1.150 -1.250 EN versus Ag-AgCI Fig. 2 Square-wave stripping voltammograms for a solution contain- ing 1.2 pg 1-1 of both Nil1 and Co" after adsorption of their complexes with DMG on: A, an MFE; and (B). an HMDb. Conditions: preconcentration potential, -0.7 V (versus Ag-AgC1); electrodc rotation speed. 10 Hz; preconcentration timc, 60 s; frequcncy, 40 Hz; scan increment, 2 mV; pulse height, 10 mV; supporting electrolyte.NH4Cl-NH3 (pH 9); DMG concentration, 1 mmol I - ' and HMDE arca, 2.5 mm2 Table 1 Calibration parameters for Ni" and Co" determined by SWAdSV (referring to the linear range of the calibration graphs: sce text). Conditions as in Fig. 3 Preconcentration time ~~~~ ~ 60 S 120 s Parameter Ni Co Ni Co Slope/pA 1 pg- 0.20 0.85 0.46 1.23 Intercept/pA 0.11 0.09 0.006 0.07 Correlation coefficient 0.998 0.997 0.999 0.999ANALYST, JANUARY 1993, VOL. 118 49 12 pg 1-1, while for 120 s preconcentration linearity holds only up to 6 pg 1-1; on the other hand for longer preconcentration times the sensitivity is increased, as shown in Table 1. For concentrations higher than 12 p,g 1-1 (for 60 s preconcentra- tion) or higher than 6 pg 1-1 (for 120 s preconcentration) the graphs are no longer linear, as saturation of the electrode occurs.This behaviour, ie., an initial linear increase of adsorbed analyte with concentration followed by a gradual saturation of the electrode, is typical of Langmuir-type adsorption. At concentrations higher than 24 pg 1-1 double peaks start to appear for Co, indicating multilayer adsorption; this fact may also affect the Nil1-DMG adsorption because the Ni response becomes irreproducible. The limits of detection are 14 ng I- 1 for Ni" and 12 ng 1-1 for Co" at the 30 level (99% confidence level) for 120 s deposition. For Co" the limit of detection can be further decreased by longer deposition times but for NiI1 long deposition times (>I20 s) do not offer any improvement in the limit of detection, however, as even for the blank well defined Nil1 peaks are obtained.The limit of detection can be decreased by isothermal distillation of the ammonia and nitric acid used to prepare the supporting electrolyte and by recrystallization of DMG .9 Simultaneous presence and deter- mination of Nil1 and Co" at the same concentration levels narrows the linear range for both metals compared with the situation where only one metal ion is present, as expected. For instance, for 60 s deposition, the calibration graphs for solutions containing only one metal ion were linear up to the 48 pg 1 - 1 level. Effect of the Deposition Potential As shown in Fig. 3 the adsorption of the NilL-DMG complex is essentially the same from open circuit to -0.8 V. In contrast \ -0.9-0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 Deposition potentialN versus Ag-AgCI Fig.3 Effect of the preconcentration potential on the square-wave stripping response for: A, Nil1; and B, Co", adsorbed on an MFE as their DMG complexes. Ni", 0.6 pg 1-1 and Co", 1.2 pg I-'. Conditions: frequency, 60 Hz; and scan increment, 1 mV; other conditions as in Fig. 2 2.5 1 4 2.0 g . h v c. 1.5 3 m a 1.0 0.5 0 1 2 3 4 5 6 7 8 9 10 (Electrode rotation speed);/(Hz)i Fig. 4 Effect of the electrode rotation speed on the square-wave stripping response for: A, Ni"; and B, Co", adsorbed on an MFE as their DMG complexes. Nil' and Co", 1.2 pg 1-1 each. Other conditions as in Fig. 2 the adsorption of the Co"-DMG complex remains constant between 0 and -0.4 V and then starts to increase rapidly as the potential of the electrode becomes more negative.This behaviour is explained by the fact that the Ni"-DMG complex is adsorbed on the electrode as a neutral species10 while the Coil-DMG complex is adsorbed as a positively charged species;ll as a result the adsorption of Co"-DMG is favoured at potentials more negative than the potential of zero charge for a mercury electrode (about -0.5 V). The Ni"-DMG complex, being neutral, does not show a tendency to be adsorbed strongly at any extreme negative or positive poten- tial. Effect of the Electrode Rotation Speed The effect of the electrode rotation speed on the stripping response is shown in Fig. 4. For a mass-transport controlled adsorption step within the linear part of the adsorption isotherm, the response should increase linearly with the square root of the rotation speed until the equilibrium surface concentration (satisfying the Langmuir adsorption isotherm) is established on the electrode surface.12.13 For the Co"-DMG complex, which adsorbs strongly at -0.7 V, the equilibrium adsorption surface concentration is relatively high and adsorp- tion surface equilibrium is not achieved before a rotation speed as high as 36 Hz.For the Ni"-DMG complex, the adsorption of which is weaker at -0.7 V, the response increases linearly up to 4 Hz and then levels off as soon as the adsorption equilibrium surface concentration is reached. Effect of pH and of DMG Concentration The effect of pH on the stripping responses for Ni and Co has been reported previously14 and similar results were obtained in this work.Resolution between the Ni and Co peaks increases but the sensitivity for the Co peak decreases with increasing pH. For analytical purposes the optimum pH value was about 9. The effect of the DMG concentration has also been studied for the HMDE'S and its effect is the same for the MFE. Comparison Between DP, SW and SC Stripping In Fig. 5 three stripping voltammograms of the same solution are shown, obtained by using typical values of staircase (SC), square-wave (SW) and differential-pulse (DP) stripping. It is clear that DP and SW stripping offer an excellent discrimina- tion against background current. On the other hand SW and SC stripping are typically 4-20 times faster than DP stripping [in this instance the scan rates are (in mV s-1): SC 100, DP 10 8.000 I 1 7.000 6.000 5.000 6 4.000 a 3.000 2.000 1.000 a 7 0.000 L 1 I I I I -0.800 -0.900 -1.000 -1.100 -1.200 -1.300 EN versus Ag-AgCI Fig.5 Comparison between DP. SC and SW modes of stripping for a solution containing 1.2 pg I-' of Nil1 and 0.6 pg 1-I of Co" adsorbed on an MFE as their DMG complexes. Conditions for DP: scan rate, 0.01 V s-1; scan increment 2 mV; drop time, 0.2 s; pulse height, 20 mV; pulse width 0.05 s; and 5-point moving average filter. For SW: frequency, 60 Hz; scan increment, 3 mV; and pulse height, 10 mV. For SC: scan rate, 0.1 V s-l; and scan increment, 2 mV. Other conditions as in Fig. 25 0 ANALYST, JANUARY 1993, VOL. 118 9.000 8.000 (4 - 7.000 6.000 5.000 4.000 3.000 2.000 1.000 0.000 1 I 1 I I I I I A a -0.850 -0.950 -1.050 -1.150 -1.250 m -0.900 - 1 .ooo -1.100 - 1.200 0 10.000 9.000 8.000 7.000 6.000 5.000 4.000 3.000 2.000 1.000 0.000 I I I I 1 -0,800 -0,900 -1.000 -1.100 -1.200 -1.300 EN versus Ag-AgCI Fig.6 ( u ) Effect of the SW frequency on the stripping response for a solution containing 0.6 vg 1 - 1 each of Ni" and Coif adsorbed on an MFE as their DMG complexes. A, 10; B, 20; C , 40; D. 80; and E, 100 Hz. Conditions: pulse height, 1.5 mV; and scan increment, 1 mV; other conditions as in Fig. 2. ( b ) Effect of the scan increment on the SW stripping response for Ni" and Co" adsorbed on an MFE as their DMG complexes. Nili and Co". 1.2 pg1-I each. A. 1; B, 2; C , 5 ; D, 10; and E, 20 mV. [In ( h ) no filtering was used for the voltammograms.] Other conditions as in Fig.2 and SW 801 and give much higher peaks. The SW waveform seems to combine the relative advantages of SC and DP waveforms. Square-wave Parameters The SW parameters that were investigated were the fre- quency, the pulse height and the pulse increment. These parameters are interrelated and have a combined effect on the response but here only the general trends will be examined. Frequency The response for both Ni and Co increases with SW frequency but at frequencies higher than 100 Hz sloping background current renders the measurement difficult, especially for the Ni peak [Fig. 6(a)]. This behaviour may be because, in order to achieve higher frequencies, the pulse width is shortened; as a result the measurement is taken at a time when the capacitive current is still significant and contributes to the measured response.16 Pulse height Increase in the pulse height causes an increase in the Ni peak up to 20 mV and in the Co peak up to 30 mV and a similar behaviour was observed in an earlier study with the HMDE.6 The peak potential shifts to the positive direction with increasing frequency. Scan increment The scan increment, S I , together with the frequency,f, define an effective scan rate, v, according to the relationship: v = S I x f (1) hence increase of the scan increment is expected to result in an increase of the response as the scan rate is increased. However, increase of the scan rate also results in 'aliasing' as fewer points are sampled during the experiment and conse- quently in a less accurate representation of the actual response.This effect is shown in Fig. 6(h) where no smoothing of the data was carried out: at scan increments greater than 10 mV too few points are sampled and the peaks are distorted, whereas at small scan increments (1-2 mV) the response is more accurately recorded but higher frequency noise is also present . An increase of either the frequency or the scan increment results in an increase in the effective scan rate [eqn. (1)); because the reduction of Ni and Co complexes with DMG is an irreversible process, increase in the effective scan rate results in a shift of the peak potentials to the negative direction [Fig. 6(a) and (b)]. Determination of One Ion in Great Excess with Respect to the Other As the relative concentrations of Nil1 and Co" in real samples vary within a wide range it is of importance to be able to determine both metals simultaneously irrespective of the sample.It is possible to control the adsorption process in order to achieve preferential adsorption of a particular DMG complex and thus facilitate the determination of a specific ion; this can be mainly achieved in the following ways. ( a ) By controlling the preconcentration potential it is easy to adsorb the Ni"-DMG complex selectively at potentials more positive than -0.4 V while both the ColI-DMG and Ni"-DMG complexes are adsorbed at potentials more negative than -0.6 V (Fig. 3). ( h ) By varying the preconcentration time it is possible to control the adsorption of the two complexes as long preconcentration times (>lo0 s) result in an enhanced Co"-DMG adsorption.( c ) By increasing the rotation speed of the electrode to more than 4 Hz the adsorption of the Co-DMG complex is promoted, while that of the Ni-DMG complex is not affected (Fig. 4). (d) By using large pulse heights (>20 mV) an enhanced Co response can be achieved. (e) High DMG concentrations (> 1 x mol 1-*) result in an increased Co response. 15 By using the optimum conditions it is possible to determine one metal ion in the presence of more than a 100-fold excess of the other. Reproducibility and Stability of the Mercury Film The reproducibility of the measurements was assessed by carrying out eight successive analyses on the same mercury film for a solution containing 1.2 pg 1-1 each of Nil1 and Co"; the reproducibility in terms of the relative standard deviation was 2.5% for the determination of Ni and 2.1"/0 for the determination of Co.The stability of the mercury film is very important for the reliability of the results. It was found that plating for 2 min produced mercury films that would be stable for at least 2 h if care was taken to exclude oxygen. Moreover, solutions containing chloride ions could be safely analysed if the electrode was polarized only to potentials more negative than 0 V. No deterioration of the film was observed at very high rotation speeds, as a reproducible and well defined response was obtained even at 83 Hz. Both strongly alkaline and acidic solutions could be safely used without deterioration of the mercury film. However, as the plating step can be carried out while the sample is being de-oxygenated the approach taken in this work was to plate a new mercury film after a series of standard additions (4-5 scans).The reproducibility of an alternative method of measure- ment was also assessed: instead of cleaning the mercury film a new film was used for each measurement. This method wasANALYST. JANUARY 1993, VOL. 118 51 0.9 I c 2 0.5 2 0.4 TY2Ll 0.0 0.1 -5 -4 -3 -7 -6 ym 03c \ I 0.211q Log([Triton X-iOO]/g 1-1) Fig. 7 Effect of the surfactant Triton X-100 at different concentra- tions on the square-wave stripping response for: A, Ni"; and B. Co", adsorbed on an MFE as their DMG complexes. Other conditions as in Fig. 3 Table 2 Determination of Ni and Co in a high-purity iron sample and in a laboratory-prepared standard Ni (YO) Co (% ) Sample Certified Found" Certified Found" High Purity Iron 0.0036 0.0040 0.0073 0.0072 BCS-CRM 149/3 +0.0008 f O.o()03 * 0.0005 +0.0006 Ni/pg 1- 1 Co/pg 1 - 1 Added Found-/- Added Found? Laboratory standard 0.48 0.50 * 0.03 0.39 0.42 ? 0.03 * n = 8 .t n = 3 . clearly less satisfactory as far as reproducibility is concerned giving values for the relative standard deviation of 26.8% for Ni and 19.9% for Co. It is assumed that the decrease in reproducibility was mainly due to irreproducible generation of the mercury films rather than irreproducible conditions of adsorption. Interferences Two major sources of interference were investigated. (a) Surfactants present in most real samples are the more serious interference in AdSV. Triton X-100 was used to simulate the effcct of a typical non-ionic surfactant and Fig.7 shows how the stripping peak current is affected for different concentra- tions of Triton X-100. Similar experiments carried out by the authors with an HMDE have demonstrated that the MFE is as tolerant to the presence of Triton X-100 as the HMDE. ( b ) Metal ions can also interfere with the measurement, hence a number of common metal ions were examined: Pb", Hg", Cu", Fe", AIIII, Cd", Tiib, Ca" and Mnll added at a 1000-fold excess over Nil' and Co" did not interfere. Zinc(ii) was found to interfere severely at a 500-fold excess over Ni" and Co" but no interference was observed for excesses lower than 100-fold. For samples with high Zn" concentration the addition of nitrilotriacetic acid (NTA) is recommended.l 4 Accuracy and Applications The accuracy of the method was assessed by determining Ni" and Co" both in a laboratory-prepared standard and in the British Chemical Standard (BCS) Certified Reference n 1.200 I 1 1.000 0.800 5 0.600 0.400 0.200 0.000 z -0.4 0 0.4 Concentration/pg I- -0.200 I I I I I I -0.800 -0.900 -1.000 4 . 1 0 0 -1.200 -1.300 -4.400 EN versus Ag-AgCI Fig. 8 Standard additions (after background subtraction) for a sample containing 0.48 btg I-' of Ni" and 0.39 pg I-' of Co" in 0.5 mol I-' NaCl. Curves: A, sample; B. C and D, standard additions of 0.15, 0.30 and 0.45 pg 1 - 1 , respectively, of Ni" and Co". Other conditions as in Fig. 2 Material (CRM) 149/3 High Purity Iron by the standard additions method. The results are shown in Table 2, indicating good accuracy for both samples. A series of standard additions for a sample containing 0.48 pg 1-1 of Ni" and 0.39 pg I-' of Co" is shown in Fig.8; in this instance a background subtraction was carried out resulting in an extremely flat baseline. The authors express their gratitude to the Royal Society of Chemistry for financial support to A. E., who is in receipt of an SAC studentship. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 References Wang, J . , in Electroanalytical Chemistry, ed. Bard. A. J . , Marcel Dekker, New York, 1988, vol. 16, pp. 1-86. Kalvoda, R., and Kopanica, M., Pure Appl. Chem., 1989, 61, 97. Wang, J . , Stripping Analysis, VCH, Deerfield Beach, FL, 1985, pp. 66-75. Batley, G. E., and Florence. T. M., J. Electroanal. Chem., 1974,55,23. Eskilsson, H.. Haraldson, C., and Jagner, D., Anal. Chim. Acta, 1985, 175, 79. Ostapczuk. P., Valenta, P., and Nurnberg, H., J . Electroanal. Chem., 1986, 214, 51. Brett, C. M. A., Oliveira Brett. A. M. C. F.. and Pereira, J. L. C., Electroanalysis, 1991. 3, 683. Newton, M. P., and van den Berg, C. M. G., Anal. Chim. Actu, 1987, 199, 59. Pihlar, B., Valenta, P., and Nurnberg, H. W.. Fresenius' Z. Anal. Chem., 1981, 307, 337. Pihlar, B., Valenta, P., and Nurnberg, H. W., J . Eleclroanal. Chern., 1986, 214, 157. Jin, W., and Liu, K., J. Electroanal. Chem., 1987, 216, 181. Levich, V. G., Physicoc/zernicul Hydrodynamics, Prentice-Hall, Englewood Cliffs, NJ, 1962. Bard, A. J., and Faulkner, L. R., Electrochemical Methods, Wilcy, Ncw York, 1980, pp. 516-519. Gilbert, M. G. M . , Powell, H. K. J . , and Fardy, J . J . , Anal. Chim. Acta, 1988, 207, 103. Adeloju, S. B., Bond, A. M., and Briggs, M. H., Anal. Chim. Actu, 1984, 164, 181. van den Berg, C. M. G., Anal. Chim. Acta. 1991, 250, 265. Paper 2/02234F Received April 30, 1992 Accepted October 7, 1992