Anal. Chem. 1990, 62, 27-31
27
Copper Determination in Urine by Flow Injection Analysis with Electrochemical Detection at Platinum Disk Microelectrodes of Various Radii D a r r y l L. Luscombe a n d Alan M. Bond* Department of Chemical and Analytical Sciences, Deakin University, Geelong, Victoria 321 7, Australia David E. Davey The South Australian Institute of Technology, The Levels, P.O. Box 1, Ingle Farm, South Australia 5098, Australia John
W.Bixler
Chemistry Department, State University College a t Brockport, Brockport, New York 14420
The lncorporatlon of platlnum dlsk microelectrodes of various radli (2.5-50 pm) in a wall-jet flow cell offers reduced limns of detection for the determlnatlon of copper In urine by flow Injection analysis compared wHh standard methods based on a conventional sized glassy carbon dlsk macroelectrode (radius 1.5 mm), in a thin-layer cell. The radius of the platlnum disk microelectrodewas found to be crltlcai with respect to both the limit of detection and flow rate dependence. An optimal radius value of 28 pm was found wHh detection ilmHs Increasing with both larger and smaller electrode radii. I n contrast, as theoretically expected, a diminlshed flow rate dependence was observed the smaller the radii of the piatinum dlsk microelectrodes. Sample cleanup and preparation is convenlently achleved by the use of Sep-Pak cartridges and formation of a copper dithiocarbamate complex. The metal complex Is easily oxidized at platinum disk microelectrodes In acetonltriie, whkh was the solvent used in the flow Injection method of analysis.
INTRODUCTION The properties that distinguish microelectrodes from conventional sized electrodes in stationary solution measurements include a predominance of radial diffusion to the electrode, yielding a true steady-state response in experimentally accessible times; a reduced requirement for deliberately added supporting electrolyte; a reduced ohmic iR drop in solution; a fast cell response time; and the ability to operate by using a simple two-electrode system rather than the more complex potentiostated system due to the low iR drop involved (1-3). The use of microelectrodes as analytical detectors in flowing rather than stationary solution systems, to which the above benefits apply, has already been demonstrated (4-9). The retention of many of the advantages deduced from measurements on stationary solutions in conventional cells, together with a reduced dependency of peak current on flow rate in flowing solutions, makes the use of microelectrodes as electrochemical detectors for chromatography and flow injection analysis (FIA) particularly appealing (4-9). In this paper we will discuss a flow injection analysis procedure for the determination of copper in urine, using electrochemical detection in a wall-jet cell with platinum disk microelectrodes of variable radii over the range of 2.5-50 pm. A 'comparison with detection a t a conventional sized glassy carbon disk electrode (radius 1.5 mm) in a thin-layer cell is presented to demonstrate the advantages of the microelectrode method over a standard method using commercially available instrumentation. Importantly, in this work it is shown that
while improved limits of detection are obtained via the use of microdisk electrodes, there is likely to be an optimal radius and that the limit of detection does not become more favorable as the electrode radius is decreased in size as may be deduced from solely considering the theoretically expected faradaicto-charging current ratio. This feature of microelectrode detectors having an optimal size is an important aspect of their use that has yet to be widely recognized. EXPERIMENTAL SECTION Chemicals. Chemicals were of analytical grade purity and used without further purification, unless otherwise stated. Solvents (Mallinckrodt) were of liquid chromatography grade. The water used in this study was obtained from a Millipore water purification system. Acetate buffer was prepared according to Vogel (10). Tetraethylammonium perchlorate (Et4NC104) (Southwestern Analytical Chemicals, Inc.) was recrystallized from methanol and then dried and stored in a vacuum desiccator prior to use. CIS Sep-Pak cartridges were obtained from Waters/Millipore Corp. and used as described in Figure 1. Flow Injection System. The microelectrode-based FIA system consisted of an IC1 LC1500 pump (IC1 Instruments, Melbourne, Australia), a Rheodyne 7125 sample injector (20- pL loop), and a Metrohm 656 wall-jet flow cell. The microelectrode cell was operated in the two-electrode mode, with a microelectrode as the working electrode and either a Ag/AgC1(3 M KC1) reference or large-area gold disk (radius 2.5 mm) as the second electrode. The gold disk was found to be suitable for use as a pseudo reference electrode if the current was maintained below 1.0 nA to avoid polarization of the electrode. The current was measured with a Keithley 614 electrometer, and the cell voltage was applied by using a home-built precision voltage source, as described previously ( 4 , 5 ) . The macroelectrode FIA system consisted of a Waters 6000A pump coupled to a Rheodyne 7125 sample injector (20 or 50 p L loop) and then to a BAS LC4B thin-layer cell detector, using a glassy carbon working electrode (radius 1.5 mm), stainless steel auxiliary electrode, and Ag/AgCl (3 M KC1) reference electrode. Microelectrode Fabrication. All microdisk electrodes used in this work were fabricated from platinum wire (Goodfellows, Cambridge, England) of nominal diameters 2.5-50 pm. A length of soda glass tubing was collapsed in a flame, leaving a 1-2-mm hole in the end. A 10-15-mm length of the microwire was placed in the hole and then sealed into the glass. Electrical contact was made by using Wood's Metal (mp 72 "C) to solder the microwire to a length of copper wire, and the body of the electrode was filled with polystyrene to immobilize the connection. The end of the electrode was ground on emery paper and then successively finer grades of alumina slurry (5-0.3 pm) until a mirror smooth finish was obtained (11). Microdisk Electrode Calibration. The radii of all microdisk electrodes fabricated as described above were calibrated prior to use. Slow scan rate voltammetry in a conventional cell of a solution of ferrocene (1.0 X M) in acetonitrile (0.10 M
0 1989 American Chemical Society 0003-2700/90/0362-0027$02.50/0
28
ANALYTICAL CHEMISTRY, VOL. 62,
NO. 1, JANUARY
1, 1990
Sample Dilution
Sep-pak Cartridge Pretreatment
Sample aliquot of 3 ml
Sep-pak soaked in lo-'
NaDDTC/
diluted with H20 to 5.0 ml.
acetonitrile overnight.
Cartridge
rinsed with H20 (10 ml).
I Sample Preparation To sample solution add 5 ml 3 x
DDTC in acetonitrile
Reaction Time
Slowly load total solution '
Urine sample
>
15 mins
volume onto Sep-pak over 3 minutes.
Sample eluted into a 5 ml standard flask with 4 ml 80/20a acetonitrile/H20, pH 6 over a 2 min
-
period.
Standard flask made up to 5 ml with 80/20a acetonitrile/H20, pH 6.
Volume suggested f o r analysis 20
-
100 p l .
a
When microelectrodes were used the sample was eluted with 100% acetonitrile. Figure 1. Sample preparation and cleanup procedure used for copper in urine by flow injection analysis with electrochemical detection.
Et,NC104) was employed under conditions where a steady-state response is obtained to calibrate the microelectrodes by using the relationship (12) id = 4nFDCr (1) where id is the diffusion-limited current, n is the number of electrons transferred, F is Faraday's constant, C is the bulk cmz/s ( 4 ) ) , concentration, D is the diffusion coefficient (2.3 X and r is radius of the microelectrode. Procedures. Urine Collection. All urine samples were collected as random specimens by the clean-catch/midstream technique described by Schumann (13)and acidified and stored as described in ref 14. Glassware. All glassware used in these determinations was precleaned in an acid bath (2 M HN03) for a minimum of 7 days and then throughly rinsed with distilled/deionized water prior to use.
RESULTS AND DISCUSSION General Methodology. Determination of Copper as the
Diethyldithiocarbamate Complex. Sodium diethyldithio-
carbamate (NaDDTC) and other dithiocarbamate ligands form stable complexes with a wide variety of transition metals (15). The nonspecific nature of NaDDTC has meant that careful selection of experimental conditions is necessary to increase the selectivity of the reagent. In this study, selectivity and sample cleanup are obtained from a combination of the sample preparation procedure and differences in oxidation potentials of the metal-DDTC complexes. For example, from data contained in ref 16 it can be noted that an applied potential of 0.60 V vs Ag/AgCl will selectively oxidize the CU(DDTC)~ complex, according to a reversible one-electron oxidation reaction CU(DDTC)~+ [Cu(DDTC),]+
+ e-
(1)
in the presence of nickel, cobalt, or chromium, which are oxidized at more positive potentials. In the present work, the separation previously performed by the chromatographic column (14) is achieved by judicious choice of applied potential.
ANALYTICAL CHEMISTRY, VOL. 62, NO. 1, JANUARY 1, 1990
28
240
d
-2
n
C
160
E .-a!
/+’I
/
0)
I Y
m
p“
80
/+
b
+’+/
+
/;+b
+/
’ t
i
I
100
/
11
Figure 2. Flow injection response of Cu(DDTC), standards in 80/20 acetonitrile/acetate buffer containing 0.2 M NaN03. Conditions: flow rate, 0.7 mL min-’; sample injection, 50 ML;applied potential, +0.60 V vs Ag/AgCI. The working electrode is a conventional glassy carbon electrode in a thin-layer cell. Concentrations (as Cu) are (a) 10,(b) 20, (c) 40, and (d) 60 ppb.
Sample Determination. In a FIA method where no chromatographic column is employed, no separation of interfering components present in urine is possible during the copper determination. That is, unless a sample cleanup procedure is used, gross interference occurs. A simple and efficient means of sample preparation has been developed using the small disposable solid-phase extraction cartridges (Sep-Pak (Waters)), which are available in a wide range of solid phases. A reversed-phase (C18)Sep-Pak was found to be a suitable choice for the sample preparation and cleanup, as detailed in Figure 1. This method, which forms the copper dithiocarbamate complex in an in-situ mode, achieves the equivalent of the chromatography in ref 14 with respect to both cleanup and removal of potentially interfering species. However, the procedure is faster and does not involve dilution. Macroelectrode Studies. ( a ) Mobile Solvent Composition. The optimum mobile solvent composition for FIA experiments when macroelectrode detection with a glassy carbon electrode was employed in a thin-layer cell was found to be 80% acetonitrile/20% acetate buffer (pH = 6.0), with 0.2 M NaN03 added to increase the conductivity. A decrease in the aqueous component of the mobile solvent leads to a decrease in the limit of detection for copper. The limit of detection (signal-to-noiseratio of 2:l) with 50% acetonitrile was 20 ppb Cu and with 80% acetonitrile was 6 ppb Cu. Concentrations of acetonitrile higher than 80% cause problems with ohmic i R drop and cannot be used with a macroelectrode in a thin-layer cell. An example of the FLA response to a standard series of copper solutions is shown in Figure 2. The calibration curves for both the 50150 and 80120 (acetonitrile/ buffer) mobile solvents are illustrated in Figure 3. Although both mobile solvents yield an excellent linear response, the response per unit concentration of the 80120 mixture is considerably greater than that of the 50/50 mixture, well above the increase expected due to the flow rate difference present in the two sets of data; hence the lowered limit of detection found with the former. The addition of 5 ppb copper as the Cu(DDTC)*complex to the mobile solvent (80/20 mixture) was found to improve long-term base-line stability and lowers
200
Concentration (ppb) Figure 3. Concentration dependence of FIA peak height for copper with the use of conventional glassy carbon electrode in a thin-layer cell. Conditions: applied potential, +0.60 V vs Ag1AgCI; (a) mobile sotvent, 80120 acetonitriWacetate buffer containing 0.2 M NaN03, flow rate, 0.4 mL min-’; (b) mobile solvent, 50150 acetonitrile/acetate buffer, flow rate, 0.6 mL min-’.
the detection limit to 3 ppb Cu. This deliberate addition of copper meant that slight variations of background current due to variable copper impurity levels was effectively eliminated. Copper levels in urine are considerably higher than this level, so detection of copper in urine is possible by the procedure of FIA (80/20 solvent mixture) with amperometric detection at a glassy carbon electrode in a thin-layer cell. ( b ) Urine Samples. Recoveries from deliberately spiked urine samples were determined over the concentration range of 20-100 ppb. For all spiked samples excellent copper recoveries of 100 f 2% were obtained. Results obtained on three samples obtained from healthy adults (38.8,40.2, and 37.1 ppb) were also in excellent agreement with data obtained by the previously described chromatographic method (14) at the f5% level, thus validating the procedure developed in this work. The possibility of contamination from the Sep-Pak cartridges was examined and found to be negligible by comparing results from a duplicate series of standards. The first was prepared without the use of Sep-Pak cartridges, and the second according to Figure 1. Within the limits of experimental error no difference was noted, indicating minimal contamination. Microelectrode Method. ( a ) Instrumental Considerations. Most of the measurements with microelectrodes were performed by using a very low impedance 2.5-mm-radius gold disk as reference electrode. If the current exceeded 1.5 nA, the base-line current was found to oscillate before settling out to an equilibrium value. This is attributed to electrode polarization problems. Therefore, as a general rule, the peak current was maintained under 1.0 nA for all microelectrode experiments reported in this paper. ( b ) Mobile Solvent Composition. The effect of mobile solvent composition on detection limits noted for macroelectrode detection was also observed for microelectrodes. However a further decrease in the aqueous component of the mobile solvent and further increase in sensitivity were possible without introducing problems from ohmic iR drop. The use of a microelectrode detection system enabled a 100% acetonitrile mobile solvent to be used, with the addition of only 0.001 M Et,NC10, as the supporting electrolyte. (c) Analytical Response. In the analytical problem being considered, FIA provides a simple and efficient means of transferring the samples and standards to the electrochemical cell. The copper concentration may be derived either from
30
ANALYTICAL CHEMISTRY, VOL. 62, NO. 1, JANUARY 1, 1990 d
Table I. Detection Limits" for the Determination of Copper Using the Microelectrode and Macroelectrode Methods Coupled with Flow Injection Analysis
electrode radius 65.1 pmb 28.1 pmb 13.2 pmb 4.1 pmb 2.5 pmb 1.5 mmc
Figure 4. Flow injection response of Cu(DDTC), standards in pure acetonitrile containing 0.001 M Et,NCI04, for a 28-pm-radius R disk electrode in a Walt-jet cell: flow rate, 1.O mL min-'; a p p i i i potential, +0.70 V vs Au pseudo reference electrode. Concentrations of standards (calculated as Cu) are (a)6.4, (b) 12.8, (c) 25.6, and (d) 64.0 PPb.
normalized sensitivity, pA ppm-' m-2
analytical sensitivity, pA ppm-' 1732 768 143 58 20 6.0
X
1.27 X 3.10 X 2.61 X 8.22 X 1.06 X 8.45 X
lo5
10" 10" 10" 10" 10l2 1O'O
lim of detectn, ppb 0.83 0.70 1.08 1.40 3.80 3.0
"Detection limits ( S I N ratio of 2:l) and sensitivities calculated in units of concentration of copper added as CU(DDTC)~. Conditions: platinum microelectrode in a wall-jet cell; mobile solvent, acetonitrile (W3M Et4NC104);applied potential, 0.65 V vs Au disk reference electrode; flow rate, 1.0 mL mi&. Conditions: large-area inlaid glassy carbon disk in thin-layer cell; mobile solvent, 80% acetonitrile/20% acetate buffer (0.02 M; pH = 6.0) containing 0.2 M NaNO,; applied potential, 0.60 V vs Ag/ AgCl (3 M KCI); flow rate, 1.0 mL m i d .
Table 11. Magnitude of Noise and Signal-to-NoiseRation for Microelectrodes of Different Radii* under Conditions of Flow Injection Analysis with a Wall-Jet Cell
3 1
2-
electrode radius, pm
L
c
.-
El)
Q
65.7 28.1 13.2 4.7 2.5
I
El)
0 -I
1 0
1 2
1 4
1 6
1 8
2 0
Log (Concentration (ppb)) Figure 5. Concentration dependence of FIA peak height for copper observed with a range of Pt microdisk electrodes of different radii: (a) 65.7, (b) 28.1, (c) 13.2, (d) 4.7, (e) 2.5 pm. Concentrations are in parts per billion (calculated as Cu); other conditions are as in Figure 4.
the peak height (current) or peak area (charge), as both are a linear function of concentration. A typical FIA microelectrode response to a series of CU(DDTC)~ standards is shown in Figure 4. The dependence of injected copper concentration on peak height or peak area was determined to be linear for a range of microelectrode sizes, in the concentration range of interest; data for peak height are shown in Figure 5. It was anticipated that decreasing the size of the electrode would lead to greater analytical sensitivities (more favorable signal-to-noise (S/N) ratios) in flowing systems, as would be predicted for stationary solutions. However, experimental data obtained in this work indicate that decreasing the size of the electrodes below a radius of 28.1 pm leads to a less favorable limit of detection (poorer S / N ratio), as shown in Table I. The area normalized analytical sensitivity did indeed increase as the radius was decreased, but this did not result in a concomitant decrease in detection limits because of the dependence of noise on the electrode radius. The reason for this behavior is revealed by a detailed examination of the short-term variation in the background current (noise). The background noise was found to vary with electrode radius (Table 11),where a decrease in radius corresponds to a decrease in observed noise. However the noise level decreases at a slower rate as the radius is decreased than does the faradaic response, as evidenced by the S/N ratios calculated for the 2 X lo-' M CU(DDTC)~ standard. Similar results were obtained for all concentrations of CU(DDTC)~. An increase in S / N from varying the radius from 65.7 to 28.1
normalized noise level, pA noise, pA m-2 2.95 x 6.85 x 9.13 x 4.96 X 1.27 x
0.40 0.17 0.050 0.035 0.024
107 107 107
lo8 109
S / N ratio 65 70 45 31 13
OSignal-to noise ratio calculated for injections of 2 X M Cu(DDT& standard. Noise levels calculated as peak-to-peak values, using experimental conditions contained in Figure 4.
pm is followed by a progressive decrease in S/N as the radius decreases. We can therefore find no analytical benefit in incorporating microelectrodes less than 28.1 hm for the determination of copper in wall-jet type detectors. The higher background found with small-radii electrodes is presumably a result of imperfect sealing and other nonidealities introduced during the fabrication and polishing of the smaller electrodes. The question of noise in flowing solutions is complex (17,18) and needs to be further addressed with respect to microelectrodes. ( d ) Flow Rate Effects. The variation of current with rate of mass transport, or flow rate, is expected to vary with the geometry of the flow cell and the nature of the electrodes employed. The dependence of peak height on flow rate has the form
i,
U*
(2) where i, is peak current, U is flow rate, and x is flow rate dependence. When a sample of electroactive species is injected into the flowing stream, a transient or peak shaped response is the result. When the electroactive species is present in the mobile solvent, such that a constant uniform concentration of the analyte is flowing into the detector cell, then a steady-state response is observed. The experimentally determined flow rate dependencies for these two kinds of experiments are shown in Table 111. The range of flow rate dependencies observed with the microelectrodes does not correspond to any particular flow model described in the literature (19-22). However, the smallest electrode (radius = 2.5 pm) examined is the least dependent on flow rate, as theoretically expected. Thus, while the greater analytical sensitivity is not achieved with the smaller electrodes, the theoretically decreased dependence on 0:
ANALYTICAL CHEMISTRY, VOL. 62, NO. 1, JANUARY 1, 1990
Table 111. Exponential Flow Rate Dependencies in a Wall-Jet Cell for a Range of Microelectrodes of Different Radii" flow rate dependence (x) electrode radius, pm
transient response
steady-state response
65.7
0.80
28.1 13.2 4.7 2.5
1.11
0.90 0.85
0.61
0.95 0.72 0.43
0.59 0.40
" F l o w r a t e dependence o f t h e f o r m i a Uxwhere U i s t h e flow rate a n d I is t h e f l o w r a t e dependence. T h e concentration o f copp e r was 1.0 X lo4 M C u ( D D T Q 2 in a c e t o n i t r i l e M Et4NC104). The f l o w r a t e was varied f r o m 0.4 t o 1.6 mL mi&, in 0.2 mL min-l increments.
d
31
conventional approach. Both methods are sufficiently sensitive for the determination of copper in urine a t naturally occurring levels. The absence of the column present in the previously described chromatographic method (14) leads to lower detection limits than previously reported. This is a considerable advantage for determining copper in urine of healthy adults. 3. There was found to be limited benefit in using extremely small microelectrodes. An increase in the limit of detection was observed with a decrease in electrode radius below 28.1 pm. However, the dependence of peak current on flow rate does decrease as the electrode radius decreases, and this could be an advantage in some circumstances. 4. The decrease in signal-to-noise ratio with decreasing microelectrode radius is interesting. Data obtained from chromatographic separation and electrochemical detection indicate that the same trend is found when high-performance liquid chromatography is associated with the analytical scheme. It therefore appears that this effect may be a general phenomenon associated with electrochemical detection at microelectrodes. Registry NO.NaDDTC, 148-18-5;CU,7440-50-8; Pt, 7440-06-4. LITERATURE CITED
Flgure 6. Flow injection response of copper in human urine. Concentrations of standards (calculated a$ Cu) are (a) 0.7 ppb, (b) 3.2 ppb, (c) 6.4 ppb, (d) 31.8 ppb, (e) diluted urine sample (seeFigure 1). Other conditions are as in Figure 4.
flow rate is observed and is a distinct advantage when flow rate variations are introduced into a flowing system by the Pump. ( e ) Urine Samples. The results for the determination of copper in a urine sample using FIA with amperometric detection a t a 28.1-pm microelectrode are shown in Figure 6. The concentrations determined on individually collected samples on consecutive days were 26.7 f 2 and 34.4 f 2 ppb Cu, respectively. These results fall within the limits expected for a normal healthy adult (23-25) and are in agreement with results obtained by the high-performance liquid chromatography method described previously (29.1 f 1.5 ppb) (14). In the previous study (14), extensive studies on the medical status, sample collection, storage, and sample treatment were undertaken. Similar considerations with respect to sample treatment and storage apply to the present method. Readers are referred to ref 14 for these aspects of urine analysis. CONCLUSIONS 1. The method for the determination of copper in urine using the sample preparation method described in Figure 1 with electrochemical detection a t both conventional glassy carbon electrode-thin-layer cell configurations and microelectrode-wall-jet cell configurations has been found to be simple to perform. 2. The microelectrode method exhibits a considerable improvement in the limit of detection and in some instances decreased dependence on flow rate over that found with the
Bond, A. M.; Oldham, K. B.; Zoski, C. G. Anal. Chim. Acta 1989,216, 177-230. Wightman, R. M.; Wipf, D. 0.I n Electroana~calChemistty; Bard, A. J., Ed.; Marcel Dekker: New York, 1986; Vol. 15, pp 267-353. Ultramlcroelectrodes; Fleischmann, M., Pons, S., Rollson, D. R., Schmidt, P. P., Eds.; Datatech Systems Publishers: Morganton, NC, 1987. Bixler, J. W.; Bond, A. M. Anal. Chem. lW8, 5 8 , 2859-2863. Bixier, J. W.; Bond, A. M.; Lay, P. A,; Thormann, W.; van den Bosch, P.; Fleischmann, M.; Pons, B. S . Anal. Chlm. Acta 1988, 187, 67-77. Caudill, W. L.; Howell, J. 0.; Wightman, R. M. Anal. Chem. 1982,5 4 , 2532-2535. Khoo, S.8.; Gunasingham, H.; Tay, B. T. J. Electroanal. Chem. InterfacialElectrochem. 1987,216, 115-126. St. Claire, R. L., 111; Jorgenson, J. W. J. Chromatogr. Sci. 1985, 2 3 , 186-19 1. Knecht, L. A.; Guthrie, E. J.; Jorgenson, J. W. Anal. Chem. 1984,5 6 , 479-462. Vogel, A. A Textbook of Quantitative Inorganic Analysis; Longmans: London, 1968; p 1161. Bond, A. M.; Luscombe, D. L.; Oldham, K. B.; Zoski, C. G. J. Electroanal. Chem. Interfacial Electrochem. 1988,249, 1-14. Oldham, K. 8. J. Electroanal. Chem. Interfacial Electrochem. 1987, 237, 303-307. Schumann, G. B. Urine Sediment Examination; Waverly Press: Baltimore, 1980; p 13. Bond, A. M.; Knight, R. W.; Reust, J. B.; Tucker, D. J.; Wallace, G. G. Anal. Chim. Acta 1988, 182, 47-59. Fries, J.; Getrost, H. Organic Reagents for Trace Analysis; E. Merck: Darmstadt, FRG, 1977; pp 125-152. Bond, A. M.; Wallace, G. G. Anal. Chem. 1982,5 4 , 1706-1712. Fox, K.; Armstrongdams, M.; Mlllar, J. J. Neurosci. Methods 1980, 3 , 37-48. Morgan, D. M.; Weber, S. G. Anal. Chem. 1984, 5 6 , 2560-2567. Yamada, J.; Matsuda, H. J. Electroanai. Chem. Interfacial Elechochem. 1973,44, 189-198. Chin, T. S.; Tsang, C. H. J. Electrochem. SOC. 1978, 125, 1461- 1470. Albety, W. J.; Bruckenstein, S. J. Electroanal. Chem. Interfacial Electrochem. 1983, 144, 105-112. Dalhuijsen, A. J.; van der Meer, Th. H.; Hoogendoorn, C. J.; Hoogvliet. J. C.; van Bennekom, W. P. J. Electroanal. Chem. InterfacialElectrochem. 1985, 182, 295-313. Sunderman, F. W., Jr. I n Chemical Diagnosis of Disease; Brown, S . S., et al., Eds.; Elsevier/North Holland Biomedical Press: Amsterdam, 1979; pp 1015-1021. Piscator, M. I n Handbook on the Toxicology of Metals; Frlberg, L., et al., Eds.; ElsevierlNorth Holland Biomedical Press: Amsterdam, 1979: pp 41 1-420. Sunderman, F. W., Jr.; Roszel, N. 0. Am. J. Clin. Pathol. 1967,4 8 , 286-300.
RECEIVED for review November 28,1988. Accepted September 27, 1989.
