Analyst, February 1995, Vol. 120

511

Square-wave Voltam metric Determination of Copper(ii) With a Nafion-Dimethylglyoxime Mercury-film Electrode

Jyh-Myng Zen, Nai-Yuen Chi, Fu-Shien Hsu and Mu-Jye Chung Department of Chemistry, National Chung-Hsing University, Taichung, 402, Taiwan The use of a dimethylglyoxime-modifiedNafion-coated mercury-film electrode in square-wave anodic stripping voltammetry for the determination of copper(I1) is described. Compared with a Nafion-coated mercury-film electrode, this modifiedelectrode showed improved resistanceto interferences from surface-active compoundsand much better selectivity. In addition, detection can be achieved without deoxygenation and the electrode can be easily renewed. Keywords: Square- wave anodic stripping voltammetry; mercury-film electrode; Nafion; copper determination; dimethylgly oxime

important, the relative advantages of using square-wave anodic stripping voltammetry (SWASV) in combination with an NC(DMG)MFE as the working electrode for the determination of nickel(i1) and copper(i1) were assessed. As opposed to differential-pulse stripping voltammetry (DPSV) and linear-sweep stripping voltammetry (LSSV), the advantages include higher sensitivity (especially for reversible reactions),20321 effective discrimination against the capacitive current ,2* speed of analysis22 and insensitivity to dissolved oxygen in the sample.23.24 This work was intended to extend the applications of SWASV on the NC(DMG)MFEs in order to investigate and establish the advantages of the technique in the trace determination of various species in real sample applications.

Introduction

Mercury-film electrodes (MFEs) have found wide application Experimental in anodic stripping voltammetry (ASV). The electrodes are prepared by electroplating a thin mercury film onto a suitable Apparatus substrate, but generally glassy carbon is claimed to be the most All voltammetric experiments were performed with a BASfavourable.2-8 The advantages of this type of electrode include lOOB electrochemical analyser. The three-electrode system simple preparation, high sensitivity and excellent resolution of consists of an NC(DMG)MFE working electrode, a Agneighbouring signals. However, one of the most common AgCl, NaCl (3 mol 1-1) reference electrode and a platinum problems when applying MFEs in ASV is the interference effects caused by organic constituents of the sample matrix.”*() wire counter electrode. Spectrophotometric measurements were carried out using a Hitachi U-3000 spectrophotometer. This problem can be largely solved by coating the working electrode with a permselective membrane, such as Nafion ionomer, cellulose acetate films and cellulose dialysis Reagents and Chemicals membranes.10-17 The function of the membrane is to prevent Doubly distilled, de-ionized water was used to prepare all the organic interferent from reaching the interface at which solutions. Nafion perfluorinated ion-exchange powder, 5% the deposition/stripping process takes place. A Nafion-coated m/v solution in a mixture of lower aliphatic alcohols and 10% mercury-film electrode (NCMFE) has been extensively water, was obtained from Aldrich. Butane-2,3-dione dioxime studied in this respect. Unfortunately, considering the (dimethylglyoxime), mercury(i1) nitrate, ammonium chloride interference effects of common ions, the NCMFE generally and sodium acetate were of analytical-reagent grade. All offers no more advantage than the MFE with regard to the buffers and supporting electrolyte solutions were prepared selectivity aspect. However, by incorporating a suitable from Merck Suprapur reagents. Copper(i1) standard solution modifying group capable of binding only a few metals, a (1000 mg 1-1) and all other standard metal solutions used in number of highly selective determinations should be possible. interference studies were also obtained from Merck. Model Recently, our group has been developing a new kind of organic compounds were used as received. Solutions of Triton NCMFE by incorporating suitable chelating agents into the X-100 (BDH), sodium dodecyl sulfate (SDS) (Wako), humic electrode. For example, an NCMFE containing appropriate acid (Fluka), albumin (Sigma) and gelatin (Sigma) were amounts of dimethylglyoxime (DMG), designated prepared by dissolution in water. D-Camphor (Eastman) was NC(DMG)MFE, was used for the determination of nickel(i1) dissolved in 1 ml of 95% v/v ethanol before dilution with by an adsorptive stripping process. 18 Further, NCMFE water. Starch (Fisher) solution was prepared (without presercontaining appropriate amounts of 2,2’-bipyridyl (Bpy), vative) using a standard procedure. Fulvic acid was extracted designated NC(Bpy)MFE, was used in the determination of from Taichung landfill leachate using XAD-8 (Sigma), lead(i1) by an anodic stripping process. 19 In this approach, one generally following the procedures reported earlier.25 can take advantage of three effects: the preconcentration effect of the ion-exchange polymer (Nafion), the selective chelating effect of the chelating agent (DMG) and the Procedure accumulation effect of the electrode (Hg). The glassy carbon disc electrode ( A = 0.0685 cmz, BAS) was In this paper, the analytical applications of the polished with the BAS polishing kit and rinsed with de-ionized NC(DMG)MFE are further demonstrated for the determinawater, then further cleaned ultrasonically in 1 + 1 nitric acid tion of copper(i1) using an anodic stripping process. Most

512

Analyst, February 1995, Vol. 120

and de-ionized water successively. Nafion coating solutions (the total volume of the coating solution was always 4 p.1, except that there were different ratios between Nafion and ethanol) were spin-coated onto the electrode surface at 2800 rev min-l. A uniform thin film was formed by evaporating the solvent after about 5 min of spinning. Mercury was deposited on the glassy carbon-(Nafion, DMG) substrate by adding 5 x 10-4 mol I - * mercury(r1) nitrate solution to the supporting electrolyte medium at -800 mV versus Ag-AgCI for 6 min as described previously. 10 The supporting electrolyte medium contained 0.07 moll- * acetate buffer (pH 4.0), prepared from sodium acetate and nitric acid. Unless stated otherwise, a medium containing ammoniaammonium chloride buffer (pH 9.0) was used in the electrochemical experiments. The performance of the electrode was then optimized by improving the coating solutions in three stages as described previously.18 Electrodes prepared with the optimum coating solution of 1.7 ml of Nafion + 1 ml of the DMG (2% m/v in ethanol) were used throughout. Underground water was collected from the campus of Chung-Hsing University, located about 5 km from several electroplating facilities. In order to fit into the linear detection range of SWASV, the sample was diluted as described previously. 18 The underground water used for copper( 11) measurement was diluted by a factor of ten. Electroplating waste solution was first acidified to pH < 2.0 with an appropriate amount of H2S04and then filtered with a 0.45 pm sterilized membrane. Following the same dilution procedure as for the underground water, the dilution factor of electroplating waste solution used for copper(i1) measurement was ten.

be considered, the electrode and the detection. Concerning the electrode, the principal factors governing the performance of the Nafion-DMG thin mercury-film CME are the concentration of DMG in the film, the thickness of the Nafion-DMG film and the deposition of mercury. Regarding the detection aspect, the factors involved are the pH of the solution, the preconcentration time, the preconcentration potential and the SW parameters. The optimum conditions with respect to the electrode aspect for copper(i1) determination were found to be similar to those in a previous study on the use of this CME in the determination of nickel(ii).l8 CMEs prepared from coating solutions that contain 2 ml of Nafion + 1 ml of DMG (0.01, 0.1,0.5, 1, 1.5 and 2% m/v in ethanol) were examined under identical conditions. In all instances copper responses could be obtained and the peak current increased as the content of DMG in the coating solution increased, as shown in Fig. 2.

.5

t

-0.boo

I

I

-1.000

-1.100

-0.500

-0.600

Results and Discussion Electrochemical Be ha viour

The cyclic voltammograms clearly demonstrate that the Cu"-DMG reaction is reversible and the Ni"-DMG reaction is irreversible. After ten successive cycles, the reversible CullDMG peaks remain almost at the same height, whereas the irreversible NP-DMG peaks almost disappear. As NilLhas a much better chelating ability than Cu" with DMG, in order to observe both peaks at the same scale of sensitivity, the Cult concentration is 500 times higher than that of Nil1. It is well known that the sensitivity of SW stripping of adsorbed species is proportional to the degree of reversibilty of the electrochemical reaction.20.21 Therefore, the species that undergo reversible reactions benefit more by the application of SWs than those that undergo quasi-reversible or totally irreversible reactions. By carrying out the experiments with SWASV, DPSV and LSSV at the same effective scan rate, the above prediction agrees well with the observed behaviour, as shown in Fig. 1. Comparing the sensitivity between SWASV and LSSV, the reversible CuII-DMG reduction shows an obvious improvement, being at least 15 times more sensitive [Fig. l(b)J , whereas for the irreversible Nit*-DMG reduction , SWASV is only about three times more sensitive than LSSV [Fig. l(a)]. In terms of background rejection, LSSV gives the highest sloping baselines, which causes a difficulty in the measurement of low concentrations. SWASV competes with DPSV for background rejection ability, with DPSV having a small advantage in that respect. Overall, even though SWASV has small disadvantage in terms of background discrimination compared with DPSV, it makes up for that with its superior sensitivity in practical applications. Optimum Conditions for Electrochemical Determinations

In order to arrive at the optimum conditions for the determination of copper(iI), there are two aspects that should

-0.200

EN versus Ag-AgCI

Fig. 1 ( a ) Comparison between A , LSSV; B, SWASV; and C. DPSV for 0.1 mg I-' Nil1 obtained with the NC(DMG)MFE at the same effective scan rate of 200 mV s-1. Conditions: B, modulation amplitude, 25 mV; modulation frequency, 100 Hz; C , pulse height, 50 mV, pulse width, 20 ms. (b) Same comparison as in ( a ) for 1 mg 1-1 Cu" except that the SW modulation amplitude was 50 mV.

34

0

0.5

1

1.5

2

DMG (m/m %)

Fig. 2 Effect of the amount of D M G in coating solutions on the peak current of Cu". CMEs prepared from coating solutions that contain 2 ml of Nafion + 1 mi of DMG (0.01, 0.1, 0.5, 1, 1.5. and 2% m/v in ethanol) were examined under identical conditions: [Cu2+] = 40 pg I-'. preconcentration time ( t p ) = 3 min and preconcentration potential (pp) = -0.8 V versus Ag-AgC1.

Analyst, February 1995, Vol. 120

Apparently, the chelating ability of DMG with copper(i1) functions properly in the CME. In summary, electrodes prepared with the optimum coating solution of 1.7 ml of Nafion + 1 ml of DMG (2% m/v in ethanol) at a spin-coating rate of 2800 rev min- and with a mercury deposition time of 6 min were used in all subsequent work. For the CME, the optimum conditions for copper(i1) accumulation were observed when the sample was slightly basic (pH 8-10), as shown in Fig. 3. The absorbance of the CuA2 complex in different pH solutions is also shown in Fig. 3 for comparison. As can be seen, the pH region was slightly different from the measured optimum absorption region of pH 6-8 for the CuA2 complex. This change is due to the fact that the SO3- sites of Nafion can attract protons and hold them inside the polymer matrix, an effect which is apparently equivalent to lowering the pH of the solution. The effect of preconcentration potential on the SW response for copper(r1) is shown in Fig. 4. As can be seen, between -0.6 and -0.9 V versus Ag-AgCI, the peak current increases as the potential of the electrode becomes more negative. This behaviour is explained by the fact that copper(i1) bears a positive charge, and as a result the accumulation of copper(1i) is favoured at more negative potentials. However, the peak current drops rapidly as the potential becomes more negative than -0.9 V. A preconcentration potential of -0.8 V was therefore chosen in all subsequent work. As preconcentration of copper(I1) takes place at -0.8 V, which is far on the negative side of the copper reduction peak, the preconcentration mechanism of copper is mainly based on electrolysis. Nevertheless, the contribution from the chelating process between DMG and copper(i1) cannot be ignored, as indicated by substantial evidence in this

1

0.8

# 0.6

z

study. In summary, the signal was attributed to the following reactions: Cu" + Hg -+ Cu(Hg) ~ 2H+ CU" + 2HA + C U A + CuA2 + 2H+ + 2e- + Hg -+ Cu(Hg)

+ 2HA

where HA = DMG. The dependence of the peak current on the preconcentration time for different copper(i1) concentrations is shown in Fig. 5. The rate of copper(i1) uptake is dependent on concentration. For higher concentrations of copper(n), (40 pg I - I ) , the peak current increases as the preconcentration time increases and starts to level off around 6 min. For a lower concentation of copper(i1) (1 pg I - I ) , it takes about 15 min for the peak current to level off. This phenomenon is as expected and further confirms the chelation process between DMG and copper(I1) in the CME. Apparently, as the electrodes were prepared at a spin-coating rate of 2800 rev min-1 with 4 p1 of the optimum coating solution, which is composed of 1.7 ml of Nafion + 1 ml of DMG (2% m/v in ethanol), the amount of DMG on the CME was actually fixed. Therefore, in order to chelate with all the DMG sites available, a longer time is needed for a lower concentration of copper(n). The SW parameters that were investigated were the frequency, the pulse height and the pulse increment. These parameters are interrelated and have a combined effect on the response. The response for copper(i1) increases with increasing SW frequency, but at frequencies higher than 160 Hz the sloping background current renders the measurement difficult. An increase in the pulse height causes an increase in the copper(r1) peak of up to 50 mV and the peak potential shifts in the positive direction with increasing frequency. The scan increment together with the frequency defines the effective scan rate; hence, an increase in either the frequency or the scan increment results in an increase in the effective scan rate. Overall, the best signal-to-background current characteristic can be obtained with the following instrumental settings: modulation amplitude, 50 mV; modulation frequency, 160 Hz; and effective scan rate 640 mV s-1. 60

4 0.4

50

6 .

0.2

0

513

1

40 30 2

3

4

5

6

7

8

9

101112

PH

20

Fig. 3 Effect of pH on A, the peak current of Cull and B, the absorbance of the CuA2 complex. Conditions for the SW measurement: [Cu2+] = 1 mg ] - I , tp = 3 min, and pp = -0.8 V versus Ag-AgC1. Conditions for the spectrophotometric measurement: [Cu2+] = 10 mg 1-1, [DMG] = 8.5 x 10-3 mol 1-1.

10

g o

2

6

4

10

8

12

14

3

60 I

20 -1.2

I

I

-1.1

-1

-0.9

-0.8

-0.7

-0.6

-0.5

Preconcentration potentialN versus Ag-AgCI

Fig. 4 Effect of preconcentration potential on the peak current of Cu" determination obtained at the NC(DMG)MFE. [Cu2+] = 50 pg I-'. t,, = 5 min.

0

5

I 10

I 15

I

20

I 25

30

Preconcentration time/rnin

Fig. 5 Effect of preconcentration time on the peak current of Cull determination obtained at the NC(DMG)MFE. [Cu2+]:( a ) 40 pg I - * ; and (b) 1 pg I-'. pp = -0.8 V versus Ag-AgC1.

514

Analyst, February 1995, Vol. 120

Effect of Oxygen

As the reduction of oxygen is a totally irreversible reaction, the SW waveform is expected not to respond to it,20,21 thus providing a means of carrying out the analysis without deoxygenating the sample. Fig. 6 shows the results for copper(i1) measured both before and after deoxygenation of the sample, with identical sensitivity. As can be seen, the peak currents are almost identical. A similar comparison was also made for nickel(u), but oxygen interfered more with the measurement, probably owing to the Nil'-DMG reduction that occurs near the oxygen reduction wave. The higher background current for the voltammogram obtained without deoxygenation is good support for the above explanation. Therefore, deoxygenation is required for the precise determination of nickel(ii), as indicated in our previous study.18 Electrode Renewal

The optimum conditions for copper(i1) accumulation were observed when the sample was slightly basic (pH 9-11). This behaviour proved in fact to be useful, as it provided a chemical method for removal of the deposit. A 5 s immersion in 1 moll-' nitric acid was generally sufficient to remove all visual and electrochemical evidence of the preceding copper(i1) deposition. However, the first three cycles were generally found to be less efficient for preconcentration than virgin CME surfaces, but after a few cycles the deposition efficiency generally reached a steady value. Nevertheless, this phenomenon was not observed when 1 ml 1-1 hydrochloric acid was used. For ten successive preconcentration-determinationrenewal cycles, the SW voltammetric response could be reproduced with a 4% relative standard deviation for the detection of 40 pg 1-1 of copper(i1). Analytical Characteristics of CME

Our previous study showed that nickel(i1) can be easily determined at the pg 1 - 1 level by SWASV on the NC(DMG)MFE.Ig After a 5 min preconcentration period, a blank of 0.1 pg 1-1 nickel could be detected and a linear response was observed from 1 to 80 pg 1 - 1 . The limit of detection can be further decreased by using purer chemicals and a longer deposition time. The same NC(DMG)MFE can also be used in determining copper(1i) at the pg 1-1 level even in the presence of oxygen. Calibration data were obtained for ammonia buffer spiked with copper(i1) following a 5 min preconcentration. Fresh sample solutions were used for each

-0.200

-0.5 EN versus AglAgCI

4.800

Fig. 6 Square-wave stripping voltammograms for 40 pg I-* of Cu" obtained with the NC(DMG)MFE ( a ) after deoxygenation; and ( b ) before deoxygenation.

individual concentration and preconcentration time and at least six determinations for each data point. For 5 rnin preconcentration experiments, the relative standard deviation was 1-5%. The calibration graphs thus obtained show a very linear behaviour with slope, intercept and correlation coefficients of 0.8112 pA pg-1 I, 0.0269 pA and 0.9997, respectively. The linear range for 5 min preconcentration was investigated by adding copper(ii) at concentrations ranging from 0.1 to 100 pg 1 - 1 . The copper(1i) concentrations calculated from the observed signals and the slope of the regression line were in good agreement with the experimental concentrations, thus indicating essentially constant sensitivity from 1 to 80 pg 1-1 copper(l1). The sensitivity started to decrease when the concentration of copper(i1) was higher than 80 pg 1 - l . The detection limit is defined as the concentration of the analyte resulting in a signal three times the standard deviation of the blank, as recommended by IUPAC. For preconcentration times of 5 and 10 min, the detection limits for copper(i1) were 1 and 0.1 pg 1-1, respectively. Of course, with shorter preconcentration times, higher concentrations were required for detection. However, an even lower detection limit could be achieved for copper( 11) provided that the preconcentration time is longer than 10 min. Interferences Any electroactive species that forms a precipitate with DMG might interfere in the determination of copper(i1). The metals commonly cited as forming stable DMG complexes are nickel(n), palladium(ii), cobalt(ii) and bismuth(Ii1). However, the chemical conditions in effect during the preconcentration step might easily be optimized to favour the formation of the copper(ii) complex over that of most of the other metals. Therefore, the number of species interacting in this manner is limited. For 0.01 mg 1-1 copper(ii), the results show that over 1000-fold excess concentrations of aluminium(iii), boron(ii1), cobalt(Ii), calcium(I1) and magnesium(i1) did not influence the copper response. Zinc(u), mercury(i1), antimony( iii) and bismuth(n1) were found to interfere at a 1000-fold excess, palladium(i1) at a 100-fold excess and cadmium(i1) and nickel(1i) at a 10-fold excess. The most serious interference was from lead(Ii), for which only a one-fold excess affected the copper signal. The interference effects caused by surface-active compounds in ASV when using MFEs are recognized. 1" Coating of the MFE with permselective membranes has been introduced as a means of circumventing the organic interferences in ASV.10-17 The function of the membrane is to prevent the organic interferents from reaching the interface at which the deposition/stripping takes place. For comparison, the effects of the surfactants on both the NCMFE and the conventional MFE were also recorded. The two surfactants chosen were non-ionic (Triton X-100) and anionic (fulvic acid) solutions. For the NCMFE, in comparison with the MFE, the tolerance was improved by a factor of two for both surfactants, as can be seen from Fig. 7. For the NC(DMG)MFE, the tolerance was improved by a factor of 20 up to about 10 mg 1 - 1 for both surfactants. The effects of model surfactants (SDS, Triton X-loo), proteins (albumin, gelatin) and carbohydrates (starch) and also D-camphor and humic acid on the NC(DMG)MFE in SWSV were further studied. Comparison of the 50 pg 1 - 1 Cull normalized peak current data for model organic compounds (15 mg 1 - 1 ) with the bare MFE, the NCMFE and the NC(DMG)MFE are summarized in Table 1. As can be seen, for each of the seven model organic compounds, the NC(DMG)MFE shows a clear improvement over the bare MFE. The NCMFE can prevent most of the interferences, with the exception of those of Triton X-100 and humic acid. The NC(DMG)MFE shows an even better

Analyst, February 1995, Vol. 120

improvement for all the interferents, especially Triton X-100 and humic acid. Hoyer et a1.10 pointed out that a membrane coating ultimately cannot eliminate all interferents in a sample from reaching the electrode surface. Even interfering components that are excluded from the electrode surface may still complicate the signal by their accumulation at the outer membrane surface. The role of DMG in the protection against surfactants apparently has something to do with the improvement of the two factors mentioned above. Practical Applications

The analytical utility of the NC(DMG)MFE was assessed by applying it to the determination of copper(r1) in underground water and electroplating waste solution. The results are summarized in Table 2 and indicate good accuracy for all the samples. Real values for the copper(n) content in underground water and electroplating waste solution were calcu-

I

5 15

lated from the dilution factors for each sample during preparation. Conclusions It has been demonstrated that SWASV in conjunction with the NC(DMG)MFEs is a potentially interesting development for the determination of copper(i1). The CME offers considerably higher resistance to organic interferences than does the NCMFE and the mechanical stability of the mercury film is also better than that of the MFE. The NC(DMG)MFE's characteristics of good sensitivity and selectivity and the capability to be easily regenerated by exposure to acid make the technique very attractive in the trace determination of various species in real sample applications. Future work in this direction is being carried out in order to assess the applicability of the technique in flowing streams.

The authors gratefully acknowledge financial support from the National Science Council of the Republic of China under grant NSC 82-0208-M005-053.

1

40

References

4 2 30

a 20

Y

2

a 10 0 -2

-1

0

I

2

Log([surfactantyrng t')

Fig. 7 Effect of the surfactants Triton X-100 (dashed line) and fulvic acid (solid line) at different concentrations on the SW stripping response for 40 pg I-' of Cu" on the A, MFE; B, NCMFE; and C, NC(DMG)MFE; tp = 5 min, p p = -0.8 V versus Ag-AgCI.

Table 1 Comparison of normalized peak currents for model organic compounds (15 mg 1-l) at 50 pg I-' Cu"

Normalized peak current (100%) Compound Gelatin Albumin SDS Triton X-100 Humic acid Camphor Starch

MFE 40.6 45.1 82.3 28.0 49.3 85.8 88.7

NCMFE 97.8 91.1 97.1 81.8 88.8 102.6 104.6

NC(DMG)MFE 99.9 97.7 98.8 97.1 95.1 100.5 100.7

Table 2 Determination of copper(i1) in underground water and electroplating waste solution

Parameter Detected value,* original/yg I-' Spike/pg 1-1 Detected value* after spiking/ Pg 1-1 Recovery (%) Real value*/pgI-' *n=6.

Underground water 4.3 20.0

* 0.2

24.2 f0.5 99.5 43.1 k 2.0

Electroplating waste solution 31.1 k 2.8 20.0 50.2 k 4.1 95.5 311.2 f 28.2

1 Wang, J . , Stripping Analysis-Principles, Instruction and Applications, VCH, Deerfield Beach, FL. 1985. 2 van der Linden, W. E., and Dieker. J. W., Anal. Chim. Acra, 1980, 177, 1. 3 Florence, T. M., J. Electroanal. Chem.. 1970, 27,273. 4 Copeland, T. R., Christie, J. H., Osteryoung, R. A., and Skogerboe, R. K., Anal. Chem., 1973,45, 2171. 5 Lund, W., and Salberg, M., Anal. Chim. Acta, 1975. 76, 131. 6 Valenta, P., Mart, L., and Rutzel, H., J. Electroanal. Chem.. 1977. 82.327. 7 Economou, A., and Fielden, P. R., Anal. Chim. Acta, 1993, 273, 27. 8 Frenzel, W., Anal. Chim. Acta, 1993, 273, 123. 9 Brezonic, P. L., Brauner, P. A., and Stumm, W., Water Res.. 1976, 10.605. 10 Hoyer, B.. Florence, T. M., and Batley, G. E., Anal. Chem., 1987, 59, 1608. 11 Hoyer, B., and Florence, T. M., Anal. Chem., 1987, 59,2839. 12 Wang, J., Bonakdar, M.. and Pack, M., Anal. Chim. Acta, 1987, 132,215. 13 Morrison, G. M. P., and Florence, T. M.. Electroanalysis.1989, 1. 485. 14 Wang, J.. and Hutchins-Kumar. L. D., Anal. Chem., 1986,58, 402. 15 Stewart, E. E., and Smart, R. B., Anal. Chem., 1984.56,1131. 16 Smart, R. B . , and Stewart, E. E.. Environ. Sci. Technol.. 1985, 19, 137. 17 M. J. Kelly, M. J., and Heineman, W. R., J. Electroanal. Chem., 1987, 222, 243. 18 Zen, J.-M.. and Lee, M.-L., Anal. Chem.. 1993. 65, 3238. 19 Zen, J.-M., and Huang, S.-Y.,Anal. Chim. Acta, 1994,296,77. 20 Lovric. M . , and Branica, M., J. Electroanal. Chem., 1987,226, 239. 21 Lovric, M., and Komorsky-Lovric, S . , J. Electroanal. Chem., 1988, 248, 239. 22 Osteryoung, J. G.. and Osteryoung, R. A.. Anal. Chem., 1985. 57, 101A. 23 Woljciechowski, M., Go, W.. and Osteryoung, J. G.. Anal. Chem., 1985, 57. 155. 24 Ostapzuk. P., Velenta, P.. and Nurnberg. H. W., J. Elecrroanal. Chem., 1986, 214, 51. 25 Town, R . M., and Powell, H. K. J., Anal. Chim. Acta, 1993. 271, 195.

Paper 4102335H Received April 20, I994 Accepted August 4, 1994