Anal. Chem. 1981, 53, 2275-2280
2275
Titration of Soil-Derived Fulvic Acid by Copper(I1) and Measurement of Free Cspper(I1) by Anodic Stripping Voltamrnetry and Copper(I1) Selective Electrode Gajanan A. Bhat, Robert A. Saar, Ronald B. Smart, and James H. Weber* Department of Chemistry, Parsons Hall, University of New Hampshire, Durham, New Hampshire 03824
This paper emphaslzes measurement of [Cu2+] during the Cu2+ tltretlon of soil-derlved fulvlc acid (SFA). We detected Cu2+ by anodic strlpplng voltammetry (ASV) and ion selective electrode (ISE) technlques under the same experimental Conditions. Correctlon of observed ASV stripping current ( I , ) after Wing the data to a theoretlcal eqiiatlon allowed calculatlon of SFA complexing capacities for 10, 20, and 40 mg/L SFA. The use of In situ calibration curves in the ASV oxperlments allowed us to convert klnetlc current-corrected I , values into [Cu"], and to compare [Cuzc]measured by ASV and ISE experlments. 'Three major conclusions arlse from thls work. Researchers can use ASV to measure [Cu2+] In the presence of fulvic acid wlth appropriate correctlons. It would be advantageous to compare the results of at least two technlques when measuring free metal lion concentrations in fulvlc acid solutions. For estlmatlon of free metal ion by ISE in fulvic acid sailutlons, researchers should exercise cautlon while calibratlng the ISE.
Recent attempts to understand the extent and nature of metal ion binding to isolated and in situ humic materials utilize dialysis (I), fluorescence spectrometry (2), ion selective electrode (ISE) potentiometry (2-9), and voltammetry (10-13). Although the most popular techniques are ISE and voltammetry, they are controversial. Sekerka and Lechner (14) noted that the Cu2+ISE responds to humic materials in the absence of Cu2+,but Buffle e t al. (15) using microprobe analysis observed no carbon adsorption on Cu2', PbZ+,or Cd2+ ISE. During voltamnietric experiments humic materials adsorb on mercury electrodes (II), and observed plating (reduction) currents should be corrected for kinetic current due to metal complex dissociation during deposition at the electrode (16-18). Both of these problems with voltammetry cause complications and necessitate assumptions in calculating free metal ion concentrations from observed currents. The metal-complexing ability of an aqueous solution of fulvic acid (FA) or of humic matter in a natural water sample is measuired by adding excess metal ion to the sample and distinguishing free metal ion and complexed metal ion concentrations. The major requirement of differentiating between free and total metal ion can be fulfilled by techniques such as ISE patentioinetry or anodic stripping voltammetry (ASV). The organic matter complexing capacit!? or total ligand concentration (C,) is typically obtained from the upper and lower portion of the titration curve of a free metal ion vs. total metal ion graph. CL values obtained by different experimental techniques do not always agree. For example, CL values €or different F A samples me lower for ASV (16) than ISE (15) results. The disparity might be caused by differences in propertieu of humic materials obtained from different sources, or in experimental conditions. For instance, ASV experiments usually use maximum metal ion concentrations of ca. 10" M compared t o CEI. M for ISE experiments. It is more
appropriate to adopt more than one technique under similar experimental conditions for comparison purposes. Such a comparison could also evaluate the validity of the assumptions made with different techniques. In this paper we discuss titrations of a fulvic acid sample by Cu2+in 0.1 M KN03and measurement of free Cu2+([Cu2+]) by ASV and ISE on a comparative basis. This comparison required manipulation of ASV parameters, particularly the drop size of mercury and time of plating, so that the identical Cu2+ concentration range could be employed for both techniques. To our knowledge no one has compared free metal ion concentration measured by ASV and ISE under the same experimental conditions. Such experiments enabled us to evaluate and critically examine the relative merits of measuring free metal ion concentration by ASV and ISE. EXPERIMENTAL S E C T I O N Apparatus. For the ASV experiments we used a Princeton Applied Research Corp. (PARC) 174A polarographic analyzer, PARC Model 315 automated electroanalysis controller, PARC Model 303 static mercury drop electrode (SMDE), Houston Instrument Omnigraphic 2000 recorder, and Orion Research Model 701 pH meter. A homemade water jacketed cell of 20 mL capacity was used for the ASV titrations. The electrode support block of the SMDE was slightly modified to accommodate the pH electrode and Ag/AgCl reference electrode. A platinum coil was used as a counterelectrode. Tank nitrogen, passed through a VClzsolution, was used to purge oxygen from the solution, to stir the solution during plating, and to blanket the solution during stripping. Triply distilled mercury was used for the SMDE. The temperature of the solution was kept at 25 f 0.1 "C. ISE experimenb were described earlier (7). The pH was controlled to 0.01 units during all titrations. All solutions were prepared in C02-freewater. Cu2+hydrolysis at the pH values employed is insignificantly small and does not cause any serious error in the measured free [Cu2+]. Reagents. Stock C U ( N O ~solution )~ was Fischer certified SO-(2-194 1000 mg/L atomic absorption standard. Adjustment for pH was made by dilute (0.1 M or less) reagent grade "OB or KOH solution. Appropriate amounts of SFA isolated and characterized in this laboratory (19) were dissolved in deionized and distilled water just before the experiments. About 1.0 M KN03 (A.R. grade) solution was electrolyzed over a mercury pool until no traces of metal ions were found. The [KNOB]was determined by evaporation and diluted to 0.1 M before the experiments. Cu2+was added by means of a G h o n t Ultraprecision microburet (2.5 mL capacity). ASV Measurements. Differential pulse ASV was used throughout. SFA in 0.1 M KN03 was deaerated for about 15 min and background current was recorded. The solutions were titrated with Cu2+ ion at constant pH in duplicate experiments and stripping currents were measured in triplicate. The Cu2+-lqA cm2/s (16). The pH 5 settings diffusion coefficient (0)is are as follows: time of plating, 90 s; scan rate, 5 mV/s; modulation amplitude, 10 mV; plating potential, -0.400 V vs. Ag/AgC1; surface area of the drop, 2.50 X lo-' cm2. The pH 6 settings are as follows: time of plating, 25 s; scan rate, 5 mV/s; modulation amplitude, 10 mV; plating potential, -0.400 V vs. Ag/AgCl; surface area of cm2. the drop, 9.93 X The accumulation coefficient (y) was measured for the above experimental conditions by measuring the plating current by dc
0003-2700/81/0353-2275$01.25/00 1981 American Chemical Society
2276
ANALYTICAL CHEMISTRY, VOL. 53, NO. 14, DECEMBER 1981
Table I. Parameters Obtained by Using is and Total [Cuz+]Data from ASV Experiments PH 5
unfitted NONREG
fitted
10 20 40 10 20 40
2.8 4.8 6.7 5.2 6.0 7.9
0.90 0.80 0.67 0.96 0.82 0.67
17.4 24.6 37.4 18.2 26.7 40.5
0.18 0.16 0.13 0.18 0.16 0.13
10
7.4 8.7 13.9
0.85 0.75 0.62
19.4a 29.9‘“ 41.2‘“
0.14 0.12 0.09
av
kinetic current corrected
20 40
av a
PH 6
Calculated without computer program because of nonconvergence.
30 A
I
m
0
32 TOTAL
x
#i
I
1
~
0
I
64
[ C U I , pM
Figure 1. Stripping current vs. total [Cu2+] added in 0.1 M KNOBat pH 5: 10 mg/L SFA (x), 20 mg/L SFA (A), 40 mg/L SFA (m).
voltammetry, measuring the ASV stripping current, correcting both for background current, and calculating the stripping current/plating current ratio (7). The value of y is 2.9 at pH 6 and 1.7 at pH 5 (16). Nonlinear Regression Analysis. The program NONREG is a University of New Hampshire version of a University of North Carolina (Chapel Hill) statistical program. Our application included use of the theoretical relationship (16)between stripping current (i,) and total metal ion added (CM).The program fits the data and refiies the estimated parameters CL,S, (upper slope: slope of i, vs. CM plot at high CM),and K (conditional stability constant) shown in eq 1. All calculations were performed by a DEC-10 computer.
RESULTS ASV D a t a T r e a t m e n t P r i o r to Kinetic C u r r e n t Correction. Figure 1 (pH 5 ) and Figure 2 (pH 6) show i, vs. CM data at 10,20, and 40 mg/L SFA. In all cases the curves are linear well before and past the end point, and curved near the end point. The x axis intercepts and S, value obtained by linear regression at high CM values yielded “unfitted” CL estimates (Table I). As expected, CL values increase as [SFA] increases. For example, CL values at pH 5 vary from 2.8 pM for 10 mg/L SFA to 6.7 pM for 40 mg/L SFA, while S, values decrease with increasing [SFA]. Similar results occur at pH 6. As a second step in the data treatment we fitted the is and CM data (Figures 1 and 2) to eq 1 by NONREG. Equation 1,
I
0
0
120 TOTAL [CUI,
24 0 pM
Figure 2. Stripping current vs. total [Cu2+] added in 0.1 M KNO, at 40 mg/L SFA (m). pH 6: 10 mg/L SFA (x),20 mg/L SFA (A),
based on 1:l metal ion/humic matter stoichiometry, shows that i, is expressed in terms of CM,S,, CL,and K. For a given system CMand is are variables, and S,, CL,and K are fitted parameters based on initial estimates. The use of NONREG required initial estimates of CL and S, values as mentioned above and K estimates of 5 X lo6 a t pH 5 and 1.5 X lo6 a t pH 6 (16). Fitted CL values are larger than unfitted ones (Table I). For example, a t pH 5 and for 20 mg/L SFA, the unfitted CL value of 4.8 pM increases 25% to 6.0 pM in the fitted data. This increase reflects the use of CM data omitted in the linear regression analysis. In contrast, fitted S, values differ very little from unfitted ones. Fitted CLvalues obtained by using N O W G are approximate values, because the calculation excludes the effect of dissociation of the complex at the electrode interface. The dissociation necessitates conversion of is values into [Cu2+]via a kinetic current correction. Only then can we compare free [Cu2+]vs. total [Cu2+]curves from ASV and ISE data. Correction of Stripping C u r r e n t for Kinetic C u r r e n t Contribution. Stripping currents (i,) measured during the oxidation of Cuo from the mercury electrode cannot be used to calculate free metal ion concentrations before correcting for kinetic current contributions (16,20-23). The reason is that ASV measures free Cu2+ and labile Cu2+, Cu2+ that dissociates during the reduction step. The first step in explaining the calculation is to show how the plating convective-diffusion current (i,) is related to is (eq 2). In eq 2, y (2) i, = y(i, + ik) is the proportionality constant between analogous plating (reducing) and stripping (oxidizing) currents, and ik is the
ANALYTICAL CHEMISTRY, VOL. 53, NO. 14, ECEMBER 1981
plating kinetic current. Calculation oE ik requires the use of eq 3, which is strictly valid when the ligand is in excess. ik
= r~FAD1/2[ML]K1L/2kf1/2
75
(3)
Equation 3 relates, i k to kf, the first-order rate constant for metal complex dissociation. The constants n, F, and A have the usual defimitions; D is the diffusion coefficient of the compllex (cm2/s),[ML] is the concentration of the complex (mol/mL), and K1is [M]/[ML]. The correction for kinetic current contribution requires the subtraction of yik from is. The following paragraph describes the steps of the kinetic current correction. From the nonlinear regression calculation described above, we obtained fitted i, values. The kinetic current corrections are done by a computer program in eight steps. (1) CLvalues were estimated from the intersections of the upper and lower curves, and calculations of i, were done at several CW/(CLC,) ratios between 0.1 and 0.7. (2) Values of i, were converted into plating anialogs by dividing by y. l(3)Plots of is/ y vs. CL were done for all three [SFA] to obtain i,/y intercepts when CL = 0 These intercepts represent convective-diffusion currentb in the absence of SFA. (4) Plots of i,/y vs. CM/(CL - C,) for three [SFAI and the y intercepts from step 3 were done. Subtraction of the convective-diffusion currents (step 3) from total plating currents for three [SFA] yields ik values. ( 5 ) The rate constant Kf was obtained from the slope of i, vs. [ML]KILl2(eq 9). (6) Values of i k for the whole titration curve were obtained via kf and eq 3. The kfvalues of 49 s-l a t pH 5 and 40 s-l at pH 6 are in reasonable agreement. (7) The plating i k was converted into its stripping analog by dividing by y, and T/ik was subtracted from original stripping currents. (8) Iterative cailculations of steps 1 to 7 were done to obtain kinetic current-corrected stripping currents. Shuman and Cromer (16) did their calculations in a different fashion. They carried out steps 1 to 7 above to calculate yik. Then they added this term (in the form of eq 3) to the right-hand side of eq 1 and reran their nonlinear regression program. Effect of Kinetic Current Correction on ASV Titration Parameters. For every set of experimental conditions C L values increase from untreated data to fitted data to kinetic current corrected data (Table I). We already noted the effect on C L and S , of fitting the i, and C M data. The pH 5 data in the presence of 20 mg/L SFA exemplify the trend. The untreated data CL value of 4.8 pM yields the 15% higher 5.5 pM value after the data are fitted, and the kinetic current correction increases the value another 58% to 8.7 pM (Table I). It is not surprising that kinetic current corrections result in higher CL values. Subtraction of the kinetic current contribution from each i, value lowers the current values a t constant CM (Figures 1 and 2), causing a larger CL value (intersection of upper and lower curves). The average conditional stability constant ( K ) also increases when the fitted data are corrected for kinetic contributions. The enhanced K value, exemplified by a 60% increase a t pH 5 , is expected because uncorrected i, values are due to free and labile [Cu2+] and consequently underestimate the cornplexed [Cu2+], The S , values decrease during kinetic current corrections for the same reason. The pH 6 data show some of the trends of the pH 5 data, but there are some differences (Table I). CLvalues of pH 6 are much higher than the pH 5 values for all three SFA concentrations. NONREG treatment of the unfitted data increases CL values by about 5 4 % and kinetic correction increases further by about 2-10%. s, values are usually unaffected by the NONREG treatment; however, with kinetic correction S , values decrease as in the pH 5 experiments. The K and CL values of this work compare favorably to reported values (16) obtained for a fulvic acid sample at pH
2277
5 -
I
A
50
I
Y
I
TOT AL
[CUI. pM Flgure 3. Free [Cu2+]vs. total [Cu"] in 0.1 M KNOS with 20 mg/L SFA at pH 5: untreated data (*), kinetic current corrected ASV data (X), ISE data (A). The untreated data include free [Cu2+]and labile [CU2+].
7. At pH 7, the reported K value of 4.7 X lo5 is very close to our average K values (Table I) of 6 X lo5 (pH 5 ) and 3 X lo5 (pH 6). Our K values and the reported ones are surprisingly similar, although for at least two reasons similar values are unexpected. First K values generally increase with pH increases (1,3,8),and second, there is no reason to expect fulvic acid samples from different sources to have the same Cu2+ binding ability. Calculation of [Cu2+]and Comparison of ASV and ISE Experiments. T o compare ASV and ISE experiments we converted ASV currents into [Cu2+](free Cu2+concentration) by eq 4, and plotted data as [Cu2+]vs. CM. Figure 3 for the
i s - yik = su[cu2+]
(4)
Cu2+titration of 20 mg/L SFA at pH 5 exemplifies our results. The kinetic current corrected ASV [Cu2+]values are lower than corresponding uncorrected [Cu2+]values because Cu2+ that dissociates from the Cu2+fulvic acid complex during the plating step is included in uncorrected values. The upper slope of both corrected and uncorrected ASV curves is unity as expected for Cu2+added after the CL end point. The kinetic current corrections for other [SFA] at pH 5 and all [SFA] a t pH 6 show similar trends. The corrected stripping currents are lower than the uncorrected values and the upper slopes of free [Cu2+]vs. total [Cu2+]plots are all unity. Figure 3 also contains ISE results done under the same conditions as the ASV titrations. Calculations show that the ratio of [Cu2+]measured by ISE and ASV (ISE/ASV) is about 0.83 for 10 mg/L SFA. The ISE/ASV ratio decreases to 0.72 for 20 mg/L SFA and to 0.41 for 40 mg/L SFA. In addition the upper slopes for Cu'+ ISE data decrease from near unity as [SFA] increases (Table 11). The upper slope values of 0.97 (10 mg/L SFA), 0.93 (20 mg/L SFA), 0.84 (40 mg/L SFA), and 0.65 (80 mg/L SFA) vary inversely with [SFA]. S, values from i, vs. C M curves are also inversely proportional to [SFA] (Table I).
DISCUSSION Determination of CLValues by ASV. The CL values of humic matter determined by ASV depend on the experimental conditions especially on the range of metal ion concentration employed. The conditions of the pH 5 measurements correspond to those of ref 16 in which the maximum total Cu2+ is about 1-2 mg/L. Further addition of Cu2+ even in the absence of FA resulted in a distinctly nonlinear relationship (instrumental limitation) between in and CM, and these nd-
2278
ANALYTICAL CHEMISTRY, VOL. 53, NO. 14, DECEMBER 1981
-
Table 11. Parameters Obtained by Using Free [Cu"] and Total [Cu"] Data from ASV and ISE Experiments pH 5
status of data unfitteda (ASV)
CL,fiM
S,, PA/MM
CL, CLM
S,, fiA/fiM
10
3.2 4.6 6.8 3.4 5.3 6.2 5.7 9.8 11.5 17.0 11.0 36.5
1.00 1.00 0.99 0.98 0.99 0.98 1.00 1.00 1.00 0.97 0.93 0.84
15.8 25.7 38.8 17.8 25.7 37.6 19.4 29.9 41.2b
0.99 1.01 0.99 0.99 1.00 0.99 1.00 0.99 1.00
20 40 N O N R E G ~fitted (ASV)
10 20 40
kinetic current corrected (ASV)
10 20 40
ISE data
pH 6
SFA, mg/L
10 20 40
a Data treatment includes free [Cuz+]and labile [Cu*+]. vergence.
ditions were not included for the calculation of titration parameters. Due to dissociation of the complex, the i, vs. C M curve is smooth and nonlinear considerably after the end point. Therefore, the upper slopes vary continuously depending upon the range of highest total [Cu2+]chosen. This problem introduces some uncertainty in the estimated CL values at pH 5 , but neither the K nor k f values were much affected (Table
1)* The above difficulties were eliminated in the pH 6 experiments by shortening the time of plating and reducing the mercury drop size, which resulted in much smaller current for the same [Cu2+].The highest total [Cu2+]was about 10-12 mg/L. The upper slopes of is vs. CMcurves were constant for a considerable range of Cu2+additions and the resulting CL values were much higher than the corresponding pH 5 values. The pH 6 values are comparable to those obtained by other methods ( I , 2) including ISE. In addition the k f values at pH 5 and pH 6 are in agreement. C L values from ASV at pH 5 (Table I) are much lower than analogous ISE values (Table 11) and dialysis values (I) but comparable to Shuman's values (16). The anomalous pH 5 ASV values call for a close scrutiny of factors affecting them. The pH 5 conditions, which are similar to those of ref 16, have larger drop size and longer time of plating which restricts the titratable concentration range of metal ions to 1-2 mg/L due to instrumental limitations. Furthermore, upper portions of is vs. CMcurves for 1to 2 mg/L total [Cu2+]are curved. This introduces some uncertainties in the determination of upper slspes. Besides, the extension of curved part of i, vs. CMcurves to evaluate CL might cause a large error in the estimated CL values at pH 5. The pH 6 ASV data further clarify this point. Manipulation of experimental conditions at pH 6 enabled us to extend the total [Cu2+]to the higher values typically used for ISE and dialysis (1)experiments. As a result, i, vs. CM curves are extended to a higher range of total [Cu2+]and upper portions are linear unlike the pH 5 conditions. CL values obtained at pH 6 are reliable and are higher than pH 5 values and comparable to those of ISE or dialysis (1). If we restrict the analysis of i, vs. CMpH 6 data to a maximum of 1 to 2 mg/L of total [Cu2+],the calculated CLvalues are close to the p H 5 values. The obvious conclusions from the above discussion are (1)one should choose the experimental ASV parameters in such a way that i, vs. CM curves should attain constant slopes over a t least some concentration range of added Cu2+. (2) The metal ion concentration range employed for different techniques should be similar to make a meaningful comparison of parameters obtained by them. Use of ASV in Determining Free Metal Ion Concentrations in the Presence of Humic Matter. Although there is an increasing use of voltammetric techniques (particularly
Calculated without computer program because of noncon-
ASV) in environmental studies, doubts often have been expressed about their utility as a precise analytical tool because of adsorption of humic materials on mercury electrodes. However, no attempts have been made to clarify the exact influence of adsorption on estimated free metal ion concentrations using current-concentration relationships. No doubt the adsorption of complexing agent or organic matter on the electrode affects the measured current. In some extreme cases adsorption can cause several undesired effects like erratic drop behavior, inhibition of the electrode process, catalytic current, shift of peak potentials, split or drawn-out waves, etc. (22). The impact of such phenomena is to cause nonlinearity in the current-concentration relationship of electrochemical measurements. Under such circumstances attempts to estimate free metal ion concentrations from the measured currents are futile. However, in certain cases researchers can minimize or eliminate such adverse effects by careful modifications of experimental parameters and measure free metal ion concentrations (22). The effect of adsorption of FA is manifested by shifts of peak potentials in some cases (24). The determination of stability constants on the basis of peak potential shifts must take adsorption into consideration. This requires a priori knowledge of the type of interaction due to adsorption at the electrode interface. Thus adsorption limits the direct application of the peak shift relationships of De Ford and Hume (25). Weak or moderately weak adsorption quite often does not cause abnormalities in polarographic measurements. The detailed study of Buffle and co-workers (11)shows that the adsorption of FA on mercury electrodes does not result in irregularities of the measured current. The absence of such anomalies suggests that the linear relationship between the free metal ion concentration and the voltammetric current is not affected. Indeed, several studies (16) including this work confirmed this. Therefore, estimation of free metal ion concentration from the measured current is perfectly justifiable when a calibration curve is done under the same experimental conditions. An additional point is important in ASV experiments. Quite often longer plating times cause a nonlinear relationship between the plating time and stripping current due to adsorption (26). The exact length of time that causes deviation from linearity depends on the nature of adsorption and the species involved, Therefore, one should minimize the plating time when adsorption occurs or is suspected. The upper slopes (Table I), though constant for a fixed FA concentration, decrease with increases in FA concentration at all three stages of the data treatment indicating that for the same free metal ion concentration the stripping current
ANALYTICAL CHEMISTRY, VOL. 53, NO. 14, DECEMBER 1981
decreases with increasing amount of FA. Such a decrease in i, values reflected by decreased upper slopes is probably clue to an increasing inhibition of electron transfer rate due to increasing amounts of adsorption of FA. Alternatively, the possibility that the decrease in slopes of i, vs. CM curves for three [SFA] may also be due to increased complexation with increase in [SFA] is unlikely. Well past the end point, further increase in comlplexation is insignificantly small due to large K values, and its effect on i, is negligible. The variation of upper slopes indicates the need of a new calibration curve for each [FA] to eetimate free metal ion concentrations. Calculations based on the complexation of metal ions by FA require the use of upper slopes of titration curves as in situ calibration curves. Although i, values do not rise substantially in the beginning of the titration due to metal ion complexation with FA, they rapidly increase after the complexation is essentially completed (Figures 1 and 2). After FA'S complexin,gcapacity has been expended, i, values should increase linearby with added metal ion. This is found true in our studies as indicated by the constant S, values of the i, vs. CM curves. We used the S, values from the kinetic current-corrected data as in situ calibration curves to estimate the free metal ion concentrations from the i, values. A second difficulty in using ASV measurements to determine CL and b: values is the necessity of correcting i, for current due to dissociation of the metal complex a t the electrode interface during the plating step. The calculation (16) adapted in this study for kinetic current corrections is not wholly sati,sfactory. First, the similarity of average K values in Table I at pH 5 and 6 is contrary to typical results (3, 6), because K usually increases as pH increases. The absence of typical trends is probably due to the fact that K values are obtained from titration data well before the end point, where the kinetic current is greater than the convective-diffusion current. Second, data in Table I show that CL values for NONREG fitted data are significantly smaller than those for kinetic. current corrected data. However, the latter calculation of CL values has the limitation of including data taken in the presence of excess Cu2+beyond the end point where eq 3 is not strictly valid. Despite the above difficulties, the omission of kinetic current contributions probably leads to erroneous conclusions. For example, Wilson et al. (10) contended that the initial increase in stripping current as SFA was added to Cu2+solution was due to adsorption and ignored the kinetic current contribution. Also Tuschall and Brezonik. (13) in ASV studies of filtered lake water determined CL values via eq 1 but ignored the kinetic current corrections. Srna et al. (27) algo neglected the kinetic current corrections in determining CL values of seawater. Our data in Table I demonstrate that without kinetic current corrections the various parameters determined could have large errors. In the above data treatment a single K value is used throughout the i, vs. curve. Such an assumption that the conditional stability constant is the same throughout complexation for a mixture of ligand like SFA is not perfectly valid. In the titration initially Cu2+ binds to stronger sites and subsequently to weaker sites. The conditional stability constants, consequently, are higher initially and decrease gradually with increase in Cu2+loading. This has been demonstrated by Gamble et al. in a recent paper ( 5 ) on Cu2+-SFA complexation. They find that the equilibrium constant (function) initially is 10-40 times higher than that of the average value. Can the basis of their theoretical and mathematical modeling, they showed that a single K value for a mixture of ligands is not a thermodynamic equilibrium constant and is not related tfi Gibb's free energy in a simple way. In brief, a single K value is a weighted average equilibrium
2279
function. More details pertaining to exact evaluation of thermodynamic equilibrium constants for a mixture of ligand can be found in ref 5. Determination of [Cu2+]by ISE. A typical plot of free [Cu2+]vs. total [Cu2+]obtained by ISE is given for 20 mg/L of SFA in Figure 3. The S, value of this plot is not unity and in general S, values decrease with increasing [SFA] (Table 11). A correction for [Cu2+]based on the expected unity slope is not satisfactory. For example, S, is 0.93 for 20 mg/L SFA. If we assumed the slope of Figure 3 should be corrected to unity as observed for ASV results (Table 11),the [Cu2+]would be 1.07 larger than the measured values. This small factor will not bring [Cu2+]to the kinetically corrected ASV value at the same CM. Instead a factor of 1.4 is required. This factor increases as [SFA] increases. For 20 mg/L SFA such a correction in terms of potential requires an increase of approximately 4 mV in the observed ISE potential when total [Cu2+] is 47 WM. There is no obvious explanation for the nonunity slopes of free [Cu2+]vs. total [Cu2+]plots of ISE data (Table 11). As explained earlier, S, values are calculated from data well past the end point. Therefore the negligible increase in complexation due to different [SFA] does not affect [Cu2+]or cause nonunity S, values. One reasonable explanation is that the evaluation of free [Cu2+]is based on calibration curves done in the absence of FA. Ligands are known to shift the equilibrium potentials measured by ISE to new values (14, 28,29), and our results confirm this behavior with SFA. The reason for such shifts is the interaction of analate ions along with the ligand in the double layer region (28). The result of such an interaction is to alter the intercept but not the slope of potential vs. log (concentration) curve. In essence the apparent standard potentials in the Nerstian relationship of ISE in the presence and absence of ligand may be different. Under such conditions the use of calibration curves obtained in the absence of ligand to calculate the free metal ion in the presence of ligand could cause a systematic error in the estimated values. The above explanations are tentative and much further study is required before we can draw definite conclusions. This work leads to three major conclusions. First, voltammetric techniques for measurement of free metal ion concentrations are feasible when two precautions are followed. Researchers should be aware of various effects due to adsorption, should adjust parameters to minimize or eliminate them, and should make kinetic current corrections before calculation of free metal ion concentrations. Second, since all metal ion speciation techniques in the presence of FA are based on assumptions, we suggest a comparison of two or more measurement techniques under the same experimental conditions. Third, we have reservations about calibrating the Cuz+ selective electrode in the absence of FA while studying the Cu2+binding with FA by ISE and are further studying this problem.
LITERATURE CITED Truitt, R. E.; Weber, J. H. Anal. Chem. 1981, 53,337-342. Saar, R. A.; Weber, J. H. Anal. Chem. 1980, 52,2095-2100. Bresnahan, W. T.; Grant, C. L.; Weber, J. H. Anal. Chem. 1978, 50, 1675-1679. Saar, R. A.; Weber, J. H. Can. J. Chem. 1978, 57, 1263-1268. Gamble. D.S.;Underdown, A. W.; Langford, C. H. Anal. Chem. 1980, 52, 1901-1908. Buffle, J.; Deladoey, P.; Greter, F. L.; Haerdi, W. Anal. Chlrn. Acta 1980, 776, 255-274. Saar, R. A.; Weber, J. H. Geochim. Cosmochim. Acta 1990, 4 4 , 1381-1384. Saar, R. A.; Weber, J. H. Envlron. Sci. Techno/. 1980, 74, 877-800. Sposito, G. S.;Holtzchlaw, K. M.: LeVesque-Madore, C. S. Soil Sci. SOC.Am. J . 1980, 43, 1148-1155. Wilson, S. A.; Huth, T. C.; Am&, R. E.; Skogerboe, R. K. Anal. Chem. 1980, 52,1515-1518. Cominoli, A.; Buffle, J.; Haerdi, W. J. Electroanal. Chem. 1980, 170, 259-275.
2280
Anal. Chem. 1981, 53,2280-2283
(12) Figura, P.; McDuffie, B. Anal. Cbem. 1980, 52, 1433-1439. (13) Tuschall, J. R., Jr.; Brezonik, P. L. Limnol. Oceanogr. 1980, 25, 495-504. (14) Sekerka, I.; Lechner, J. F. Anal. Lett., 1978, A l l , 415-427. (15) Buffle, J.; Greter, F.-L.; Haerdi, W. Anal. Cbem. 1977, 49, 216-222. (16) Shuman, M. S.; Crorner, J. L. Environ. Scl. Tecbnol. 1979, 13, 543-545. (17) Shuman, M. S.; Woodward, G. P., Jr. Envlron. Scl. Tecbnol. 1977, 7 1 , 809-813. (18) Shuman, M. S.; Woodward, G. P., Jr. Anal. Cbem. 1973, 45, 2032-2035. (19) Weber, J. H.; Wilson, S.A. Wafer Res. 1975, 9 , 1079-1084. (20) Delahay, P. "New Instrumental Methods in Electrochemistry"; Interscience: New York, 1954; Chapter 5. (21) Brainina, Kh. 2 . "Stripping Voltammetry in Chemical Analysls"; Wiiey: New York, 1974; Chapters 11, 111. (22) Bond, A. M. "Modern Polarographic Methods in Analytical Chemistry"; Marcel Dekker: New York, 1980; Chapters 2-6, 9.
(23) Davison, W. J. €lecfroanal. Cbem. 1978, 87, 395-404. (24) Greter, F. L.; Buffle, J.; Haerdi, H. J . Nectroanal. Cbem. 1979, 101, 2 11-229. (25) Crow, D. R. "Polarography of Metal Complexes"; Academic Press: New York, 1969; Chapter 4. (26) Lukaszewski, Z.; Pawiak, M. K. J. Electroanal. Cbem. 1979, 703, 225-232. (27) Srna, R . F.; Garrett, K. S . ; Miller, S. M.;Thurn, A. B. Envlron. Sc/. Tecbnol. 1980, 14, 1482-1486. (28) Hansen, E. H.; Lamm, C. G.; Ruzicka, J. Anal. Cbim. Acta 1972, 59, 403-426. (29) Hansen, E. H.; Ruzicka, J. Anal. Cbim. Acta 4974, 72, 365-373.
RECEIVED for review May 6, 1981. Accepted September 2, 1981. National Science Foundation Grant No. OCE 79-10571 provided partial funding for this research.
Epoxy-Bonded Graphite Microelectrodes for Voltammetric Measurements Joseph Wang Department of Chemistry, New Mexico State University, Las Cruces, New Mexico 88003
A mlcroelectrode composed of a commercial expoxy bonded graphite, which appears well suited for voltammetric measurements, is discussed. The electrode ls easy and inexpensive to fabricate and can be used for numerous voltammetric measurements. The technlques used In evaluating the electrode are linear scan voltammetry, dlfferentlal pulse voltammetry, pulsed-stlrringvoltammetry (at the submilllmolar concentratlon level) and dlfferentlal pulse anodlc strlpplng (for the nanomolar level). Well-defined current-potential curves are reported. Background currents are low and the usable potential range Is wide. Ascorbic add, dopamine, ferroeyanlde, lead, and cadmium were used as test systems.
The application of electroanalysis to the in vivo monitoring (1) or microliter volume batch analysis (2)requires miniature voltammetric electrodes due to the sample limitation. Microelectrodes based on various forms of carbon (mounted in glass or Teflon capillaries) have been used in connection with these applications. Most recent in vivo studies have been carried out with carbon paste microelectrodes (3). An epoxy can be added to this graphite-Nujol mixture, providing a better mechanical strength (4). Carbon fibers have been used to fabricate voltammetric microelectrodes with satisfactory response for the oxidation of catecholamines (5, 6). While significant progress has been made in this direction, the preparation of sensitive and reproducible miniature electrodes remains one of the major problems associated with in vivo electroanalysis and microelectroanalysis (3, 4). The present paper describes the relative merits of different analytical techniques using microelectrodes. Anodic stripping voltammetry (ASV) and hydrodynamic modulation voltammetry (HMV) are incorporated, for the first time, with the operation of microelectrodes for the measurement of very low concentrations of electroactive species. The WMV response is compared with that of differential pulse voltammetry (DPV) and linear scan voltammetry. The microelectrode is made of a commercial epoxy bonded graphite (Grade RX, Dylon, 0003-2700/81/0353-2280$0 1.25/0
Cleveland, OH). It is a two-component, graphite-filled, epoxy resin bonded adhesive filler. Specifications and characteristics are available along with several suggested nonelectrochemical applications (7). However, as was suggested recently by Jwtice (8) it appears to be well suited as an electrode material for various analytical applications. I t has high electrical conductivity, high mechanical strength, low residual currents, wide operating voltage range, and reproducible performance. I t is inexpensive and may be machined into various shapes. Various combinations of carbon and epoxy resin have been exploited for fabricating conventially sized voltammetric electrodes, mainly for ASV (9-12). Large and nomeproducible background currents limit their application to other forms of voltammetry. In addition, their preparations involve relatively cumbersome procedures (e.g.,grinding, centrifugation, mixing, overnight curing, etc). In contrast, microelectrodes made of commercial graphite epoxy are easily fabricated within less than 4 h. Their voltammetric characteristics and applications are described in the following work.
EXPERIMENTAL SECTION Electrode Fabrication. Two parts (by volume) of the epoxy bonded graphite resin were mixed thoroughly with one part of the correspondingaccelerator. Mixing continued for 5 min after all indications of the two individual components were gone. The end of a 3 cm length of Teflon tube (0.3 or 0.5 mm i.d., No. 24009 or 24005, respectively, Pierce Chemical Co., Rockford, IL) was then dipped in the epoxy. The epoxy filled the tip t o a height of about 3 mm. The electrode was then cured at 70 "C for 3 h. Electrical contact to the cured epoxy was made by filiing a portion of the Teflon tube with graphite powder and pushing a copper wire into it. The epoxy face was polished with a 0.1 ym alumina slurry and then washed with a stream of deionized water. Electrodes fabricated in this manner typically had resistances of approximately 80 Q. Since the currents measured in this work were in the nanoampere range, such a resistance had a negligible effect on the electrochemical response. A completed graphite f i e d epoxy microelectrode is shown in Figure 1. Apparatus. The electrochemical cell was a 22-mL (2.4 cm diameter, 5.0 cm high) glass container (Model VC-2, Bioandytical Systems, West Lafayette, IN). The cell was joined to the working electrode, reference electrode (Ag/AgCl, Model RE-1, Bioana0 1981 American Chemical Society
