170
Anal. Chem. 1985, 5 7 , 170-174
The financial support of the Natural Sciences and Engineering Research council, canada (Strategic E~~~~~programme), is gratefully acknowledged. C.M.G. also acknowledges the receipt of graduate scholarships from NSERC and FCAC (Quebec).
(15) Merritt, M. V.; Sawyer, D. T. Inorg. Chem. 1970, 9 , 211. (16) Adams, R. N. "Electrochemlstry at Solld Electrodes"; Marcel Dekker: New York, 1969; p 118. (17) Kim, J.; Faulkner, L. R. Anal. Chem. 1984, 56, 874.
RECEIVED for review May 7,1984. Accepted August 27,1984.
Determination of Standard Potentials and Electron-Transfer Rates for Halobiphenyls from Electrocatalytic Data Thomas F. Connors, James. F. Rusling,* and Azita Owlia
Department of Chemistry (U-60), University of Connecticut, Storrs, Connecticut 06268
Electron-transfer properties of halogenated biphenyls (PCB's and PBB's) are Important In thelr synthetic, environmental, and biological transformallons, many of whlch Involve radlcal pathways. The electrocatalytic method was used to obtaln standard formal potentlals (EO') and homogeneous electronfor five halobiphenyls of repretransfer rate constants (&,) sentative structures. A nonlinear regresslon method was developed to obtaln the best fit of theory to the log k , vs. Eo, data and yleided optimized values of parameters. Preclslons of 2-7 mV In Eo' and 0.1-0.4 in log k , were Improved over those obtalned graphically. A correlatlon was observed between Eo' of the halobiphenyls and blologlcal actlvlty.
Environmental contamination with chloro- ( 1 , 2 )and bromobiphenyls (PCB's and PBB's) (3) has created a need for knowledge of their electron-transfer properties, so that their behavior in complex natural systems can be more readily interpreted. Details of formation and reactivity of radicals formed by electron-transfer reactions of halogenated biphenyls are of interest because photodegradation (1, 4), chemical degradation with sodium naphthalide (5), electrochemical dehalogenation (6-8), and some biological transformations ( I ) occur by radical pathways. However, fundamental parameters of PCB's and PBB's, such as standard potentials and electron-transfer rates, remain largely unknown because their determination involves specific problems. Direct measurement of the standard formal potential ( E O ' ) requires attainment of equilibrium between oxidized and reduced forms of the redox couple (9). Reliable methods for obtaining the standard heterogeneous (electrochemical) rate constant for electron transfer (kosh) rely on forcing the reaction rate to be partially controlled by the kinetics of heterogeneous electron transfer. Neither type of experiment is feasible for the halogenated biphenyls, since, similar to other aryl halides (10, I I ) , they are reduced in totally irreversible electrochemical reactions (6-8). Their reduced forms are unstable anion radicals, and no equilibrium with the parent compound is possible under the time restrictions of most electrochemical experiments. Recently, as part of an elegant extension of the theory of homogeneous electrocatalysis, Saveant et al. (10-13) demonstrated the feasibility of determining standard potentials and rates for totally irreversible electron-transfer reactions, provided the electroactive species could be reduced by homogeneous redox electrocatalysis. The method employs a soluble, reversible redox couple (PI&)which shuttles electrons from the electrode to the substrate in a thin layer of solution close to the electrode. For aryl halide substrates (ArX), experimental conditions can be adjusted such that the reaction of 0003-2700/85/0357-0170$01.50/0
Scheme I P t e-Q I
(2) ArX2-S ArX'
ArX'
t Q
t XP
ArX-t
(41
t ArX'
ArX2
t RH
ArHX
t R-
(6)
A r X ' t SH
ArHX
t
S'
(7)
ArX2-s ArX-
t
ArX-
(5)
electrochemically generated Q with ArX to yield P and radical anion ArX-. is the rate-determining step, and its forward rate constant (kJ can be measured voltammetrically. The standard potential of the ArX/ArX-- couple and the standard rate constant (k,) for homogeneous electron self-exchange can then be obtained from a plot of log k, vs. EopQfor a series of catalysts for a single substrate (10,13). Combination of these quantities with data from the direct voltammetric reduction of ArX provides ko,h. This electrocatalytic method has been successful in determining such fundamental quantities for reductions of halobenzenes and halopyridines (10-13), aromatic epoxides (14), and bianthrones (15). Halobiphenyls are reduced in stepwise dehalogenation reactions at mercury electrodes by an irreversible ECE (electronation-chemical step-electronation) mechanism (8). The first electronation step in the direct reductions is the kinetically controlling step, which is a requirement for use of the electrocatalytic method. Thus, halobiphenyls seemed suitable candidates for measurement of Eo' and standard rates by the electrocatalytic method. We recently reported electrocatalytic dechlorination of 4-chlorobiphenyl (4-CB), 4,4'-dichlorobiphenyl (4,4'-DCB), and commercial PCB mixtures (16,17) and developed methods for computing rate constants from linear-sweep voltammograms (16,18). We have now measured kl for a series of electrocatalysts for five halobiphenyls of representative structures, and in this paper, we report the use of these data for determining standard potentials and rates. Previous applications of the electrocatalytic method used graphical analysis of the data. Considering the inherent variance in log k , (11,18, 19),the theoretical complexity of the plots, and the limited number of catalysts available, a graphical approach gives less than optimum precision and accuracy. Therefore, we also describe a general regression method to obtain the statistical best fit of the intersecting lines plots to the experimental data, yielding in the log k, vs. EopQ optimum parameter values. 0 1984 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 57, NO. 1, JANUARY 1985
171
of the log kl vs. Eom plot is found by equating the right-hand sides of eq 9 and 10 E1 = (9.68 - log k,, 8.62E0’)/8.62 (11)
+
Equations 9 and 10 can be fit simultaneously to experimental data by using nonlinear regression analysis, employing eq 11 to find the best value of EI within each iteration. This procedure yields optimized values for Eo’ and k,,, which can be combined with LSV data on the direct electrochemical reduction to yield the heterogeneous electron-transfer rate constant.
EXPERIMENTAL SECTION
- Eopp,
Y
VI.
SCf
Flgure 1. Plot of log k , vs. E o w for electrocatalytic reductions of
4-chlorobiphenyl with a series of catalysts. Solld lines are computed from regression analysis: diff. zone, eq 10; act. zone, eq 9.
THEORY As an example of a redox electrocatalytic reaction, Scheme I gives the elementary steps in the two-electron electrocatalytic reduction of 4,4’-dichlorobiphenyl (16, 18), ArX2, to 4chlorobiphenyl, ArHX. In Scheme I, Q produced at the electrode transfers electrons in homogeneous steps to the dichlorobiphenyl (eq 2) and to the radical (eq 4) formed from decomposition of ArX2-., thereby regenerating P. Proton donor R H may be an electrolyte or residual water (10, 11), and proton abstraction from the solvent (SH) may also be involved. With a large excess of dichlorobiphenyl over catalyst and relatively low scan rates ( u ) in linear-sweep voltammetry (LSV), the reaction is under pure kinetic control of eq 2, and the LSV curve has a characteristic sigmoid shape. The product of Scheme I is 4-CB, which can react further with Q but at a significantly smaller rate. For example, in N,N-dimethylformamide, kl for the reaction of 4-CB with Q was a t least 10-fold smaller than for 4,4’-DCB, with anthracene and 9,lO-diphenylanthracene (9,lO-DPA) as catalysts (16). Electrocatalytic reduction of 4,4’-DCB produces 4-CB at the electrode but a t a concentration in the reaction layer much smaller than that of 4,4’-DCB, which is usually at a 10-200-fold excess in our experiments. Thus, the actual rate of loss of the second chloride will be relatively small during the time of an LSV scan, and kl in eq 2 can be obtained with some confidence (16, 18).
The standard potential for a given substrate can be obtained from a plot of log kl vs. EOpQ (Figure l),which can be described by equations for two intersecting straight lines. Equation 8 represents the data (10, 13) at 22 “C when k1 is governed by the activation energy of eq 2 (act. zone) log kl
log It,, - ~ & ? 3 ’ p ~- E0’)/0.058
(8)
where a, is the homogeneous transfer coefficient and Eo‘ is the standard formal potential for the couple ArX2/ArXF, for the example in Scheme I. Taking as = 0.5 (10) yields log k1 = log K,, - 8.62(E0p, - EO’) (9) When eq 2 is controlled by diffusion (diff. zone) log k1 = log kd - 17.24(E0pQ- EO’)
(10) where k d is the diffusion limit (20) of the homogeneous rate constant. Eo’ is located (Figure 1) by the intersection with log k d of the line through the diffusion-zone data; log k,, is the intersection of the activation-zone line with Eo,, - Eo’ = 0. The intersection (E,) of the two straight-line portions
Chemicals and Solutions. Anthracene, 4-chlorobiphenyl, 4-bromobiphenyl (4-BB), 4,4’-dichlorobiphenyl, 9-phenylanthracene, 9,10-diphenylanthracene,4-cyanopyridine, benzophenone, benzo[flquinoline, phenanthridine, phthalonitrile, terephthalonitrile, tetracene, perylene, and 1,2-benzanthracene were obtained from Aldrich Chemical Co. 4-Methoxybenzophenone was synthesized by a standard procedure (21). a-Naphthonitrile was a gift from E. Cioffi, 2,2‘,5,5’-tetrachlorobiphenyl (2,2’,5,5’-PCB) was a gift from J. Stuart, and 3,3’,4,4’-tetrachlorobiphenyl (3,3’,4,4’-PCB) was a gift from U.S. EPA and also obtained from Ultra Scientific Co. 4-Methoxybenzophenone, 4-CB, benzolflquinoline, and phenanthridine were purified before use by crystallization from methanol; the other compounds were used as received. Acetophenone was Fisher Certified reagent grade and distilled under reduced pressure before use. Fisher tetrabutylammonium iodide (TBAI) was used as received. N,N-Dimethylformamide (DMF), distilled-in-glassgrade from Burdick and Jackson Labs, was stored over activated alumina under nitrogen at 5 OC, as described previously (18). Apparatus and Procedures. A Princeton Applied Research Corp. (PARC) Model 174A polarographic analyzer and a Bioanalytical Systems BAS-100 electrochemistry system with threeelectrode cells and procedures described previously (16,18) were used for cyclic voltammetry. Both a PARC Model 9323 hanging-drop-mercuryelectrode (HDME, A = 0.018 cm2)and a glassy carbon electrode (GCE, A = 0.07 cm2),constructed and polished as previously described (22),were used as working electrodes. A fresh drop was used for each scan on the HDME; the surface of the GCE was freshly polished before each scan. A Ag/AgI wire served as the reference electrode;a platinum wire was the counter. All potentials were corrected by reference to the standard potential of anthracene and are reported vs. the saturated calomel electrode (SCE). The supporting electrolyte for all experiments was 0.1 M TBAI in DMF which had been passed through a column of activated alumina immediately before the experiment. Activated alumina was also added directly to the cell to scavenge remaining impurities and water. DMF prepared in this way contained 2.81 A 0.06 X M water, as measured by gas chromatography. Rate constants were computed by previously described methods (1 6, 18) from voltammograms obtained under conditions pseudofirst-order in catalyst. Conditions were a 10-200-fold excess of halobiphenyl, scan rates of 0.01-0.50 V s-l, and catalyst concentrations of 0.1 mM (HDME) or 0.5 mM (GCE). Whenever possible, rate constants were estimated from purely kinetically controlled, sigmoid-shaped cathodic voltammograms. Computations. For obtaining kl from overlapped voltammetric data, a FORTRAN (WATFIV) version of a general nonlinear regression program (23) in double precision, modified as described previously (181,was used on an IBM 3081D computer. Background correctionswere handled in two ways: by subtraction of the linearly extrapolated base line (18) or by using a linear background term in the regression model. The two methods gave identical results. A BASIC version of the nonlinear regression program run on a Radio Shack TRS-80 Model I microcomputer (48K) was used for simultaneous fitting of eq 9 and 10 to experimental log kl vs. E 0 p q data. Usually, the parameters were Eo‘ and log k,, but for some fits we used a, as a third parameter. The subroutine for computing values of log k , required an initial guess for El. If E0pq IEI,the log k1 values are computed from eq 9; if Eopq> EI,log kl values are computed from eq 10. After a log kl is obtained for each E’PQ in a given iterative cycle, a new El is calculated from eq 11. E1 is confined t o values such that
ANALYTICAL CHEMISTRY, VOL. 57, NO. 1, JANUARY 1985
172
Table I. Estimation of Electron-Transfer Parameters from Electrocatalytic Rates" for Aryl Halides by Nonlinear Regression compound bromobenzene
Nb
P
9
2
chlorobenzene
8
3 2 3
3-bromopyridine 2-bromopyridine 2-chloropyridine 3-chloropyridine
6 8 5 6
2 2 2 2
log k,,, 5-l M-' found ref 10 5.75 f 0.10 5.70 6.00 f 0.14 6.42 7.01 f 0.37 7.11 7.49 6.67
-Eo'(VI VS. SCE ref 10
found
5.9
2.418 f 0.002 2.421 2.757 f 0.003 2.759 2.270 f 0.006
6.2 7.3 7.5 7.0 7.1
as
RSD,d %
2.44
8 31 9 13
0.43 2.78 0.55
2.29 2.30 2.40 2.39
2.272
2.387 2.313
11 7
3 27
" Rate data from ref 10; uncertainties obtained from pointwise variance analysis. Number of data. Number of parameters in regression analysis. Relative standard deviation of regression. Table 11. Standard Formal Reduction Potentials of Catalysts in DMF/O.l M TBAI
Table 111. Rate Constants for Electrocatalytic Reduction of 4,4'-Dichlorobiphenyln
no.
compound
-Eop~(V) VS. SCE
catalyst
Tb
1 2
benzo[flquinoline phenanthridine acetophenone 1,2-benzanthracene anthracene 9-phenylanthracene 9,lO-diphenylanthracene a-naphthonitrile 4-methoxybenzophenone 4-cyanopyridine benzophenone perylene phthalonitrile tetracene terephthalonitrile
2.105 2.005 1.960 1.955 1.905 1.855 1.845 1.835 1.825 1.725 1.705 1.623 1.570 1.540 1.518
3
10
3 4 5 6 7 8 9 10 11 12
13 14 15
Y,
V s-l
log kl,L mol-' s-l
av log k,
3.61 3.58 3.37 3.19 3.08 3.11 2.69 2.65 2.41 2.33 2.41 2.09 1.85 1.57 2.15 2.01 1.86 2.06 1.92 1.70 0.29 0.38 0.68 0.15 0.10 0.50
3.52 f 0.11
0.20 0.10
0.05 5
40
6
50
0.20 0.10
0.05 0.50 0.20 0.10
0.05 0.02
7
0.20
100
0.10 8
0.05 0.20
100
0.10
0.05
each straight-line portion of the log k, vs. EopQplot retains at least two data points. The process described above is repeated by systematically varying parameter values until the minimum in the sum of squares of deviations in log k , is reached, yielding the best values of the parameters. Standard errors were estimated by pointwise variance analysis (24),using a program provided in the nonlinear regression package (23). This latter program was modified to employ the above BASIC subroutine (available by writing to J.F.R.) for log kl vs. E O ~ Qregressions and run on the microcomputer.
RESULTS AND DISCUSSION Fitting Procedure. Electrocatalytic data for aryl halides reported by Saveant et al. (IO) were used to test the nonlinear regression procedure for fitting log kl vs. Eo,, data and to investigate its precision. Both two and three parameter fits were tried for several data sets. However, the use of a,as the third parameter increased the relative standard deviation of the regression (Table I), so we decided to rely only on the two parameter fits. Large variations of initial estimates of the parameters had no influence on their final values. In general, there was acceptable agreement between the graphical results (IO) and those from the regression analyses, but differences of up to 0.5 in log k,, and 77 mV in Eo' (Table I) are large enough to recommend use of the regression method to optimize values of the parameters. As the number of data available increased, the standard error in each parameter decreased. The uncertainty in E"' by the electrocatalytic method, recently estimated (11) at f 6 mV (determined graphically), can be improved to f 2 mV by using the regression method with nine data points. Uncertainties in log k,, are on the order of 0.1-0.4 (Table I). Electrocatalytic Reductions of Halobiphenyls. The catalysts used (Table 11) showed reversible electrochemical behavior by conventional cyclic voltammetric criteria. An illustration of the precision obtained in rate constants is given for reactions of 4,4'-DCB with various catalysts (Table 111).
9
100
10
200
0.10
200
0.05 0.02 0.20
0.20 0.10
0.05
11
0.10
0.05
3.14 f 0.11 2.50 f 0.14
1.84 f 0.21 2.00 f 0.12 1.89 f 0.15 0.45 f 0.17 0.25 f 0.18
"LSV data at HDME. bRatio of molar concentration of sub-
strate to catalvst. Table IV. Rate Constants for Electrocatalytic Reduction of 4-ChlorobiphenyP catalyst
yo
1 2
10
3 4 5 6 7
8
10
20 20 100
200 200 200
Y,
v
5-1
0.05-0.20 0.05-0.20 0.05-0.50 0.02-0.20 0.05-0.20 0.05-0.20 0.05-0.20 0.02-0.10
(N)
(3) (3) (3) (3) (3) (3) (3)
(3)
av log k , 4.45 f 0.14 3.20 f 0.14 2.68 f 0.12 2.57 f 0.18 1.85 f 0.20 1.24 f 0.11 0.60 f 0.08 0.42 f 0.22
" See Table 111. ~~
~
For some of the reactions there was a slight dependence of the measured log kl on the scan rate. A discussion of possible reasons for this small dependence is beyond this paper's scope, but comparison of a recently estimated ( 1 I) precision of *0.6 in log k , with our reproducibilities of f0.2or better suggests that the observed variations are insignificant. Values of log kl for electrocatalytic reduction of other halogenated biphenyls (Tables IV-VII) had similar reproducibilities. It was also of interest to see if values of kl could be successfully obtained on solid electrodes, such as glassy carbon,
ANALYTICAL CHEMISTRY, VOL. 57, NO. I , JANUARY 1985
Table V. Rate Constants for Electrocatalytic Reduction of 4-Bromobiphenyl* catalyst
yo
2’
80
3 5 6
40 40 40 40 40 40
7
9 10
v,
v s-l (N)
0.05-0.10 (3) 0.05-0.20 (3) 0.05-0.20 (3) 0.05-0.20 (3) 0.05-0.20 (3) 0.01-0.05 (3) 0.01-0.05 (3)
compound
3.52 f 0.40 3.58 f 0.11 2.94 f 0.11 2.49 f 0.13 1.99 f 0.20 1.13 f 0.40 0.27 f 0.16
4-CB 4-BB 4,4’-DCB 2,2’,5,5’-PCB 3,3’,4,4’-PCB
Table VI. Rate Constants for Electrocatalytic Reduction of 2,2’,5,5’-Tetrachlorobiphenyl“ catalyst 9 10
11 12
13 14
15 (I
yo
10 20 20 100 40 150 200
v,
v s-1 (N)
0.05-0.20 (3) 0.05-0.50 (4) 0.05-0.50 (4) 0.02-0.50 (5) 0.02-0.20 (4) 0.02-0.05 (2) 0.02 (1)
av log kl 3.42 f 0.32 3.07 f 0.27 3.05 f 0.37 2.09 0.29 2.06 f: 0.29 0.95 f 0.15 0.75
*
See Table 111.
Table VII. Rate Constants for Electrocatalytic Reduction of 3,3’,4,4’-Tetrachlorobiphenyl”
(I
catalyst
yo
10 11 12 13 14
10
10 1.5 20
40
Y,
v
€9-1
(N)
0.03-0.13 (7) 0.05-0.13 (4) 0.05-0.20 (3) 0.02-0.10 (3) 0.02-0.05 (2)
Table VIII. Standard Potentials and Homogeneous Electron-Transfer Rates for ArX/ArX-* Couples
av log kl
‘See Table 111. bLSV at GCE, except for catalyst 2.
av log k,
3.48 f 0.23 4.10 f 0.45 2.87 f 0.14 2.37 f 0.18 1.79 t 0.10
See Table 111.
since mercury had been used almost exclusively up to this time. The mean log k1 for reaction of the 9,lO-DPA anion radical with 4,4’-DCB at a GCE was 2.19 f 0.25, which was in good agreement with the value obtained at an HDME (Table 111). The data for 4-bromobiphenyl (Table V) shows that comparable precision to that obtained at the HDME is possible with a GCE but higher catalyst concentrations were needed because of larger background currents at the GCE. Regression procedures developed to extract log kl from severely overlapped voltammograms (18)were successful with data obtained at glassy carbon. Determination of E”’ and k,, for Halobiphenyls. Nonlinear regression of eq S11 onto the data in Tables 111-VI1 demonstrated good agreement between experiment and theory (Figure 1). Standard errors in log k,, and E”’(Table VIII) were comparable to those found for aryl halides. Even with the limited number of catalysts available, precisions of 2-7 mV in Eo’ and 0.1-0.4 in log k , were obtained. However, the accuracy of these values is more difficult to assess, since at present no referee method applicable to these compounds exists. Nevertheless, the trends in E”’are as expected: The most negative value is found for 4-chlorobiphenyl; the value for the more activated 4-bromo derivative is 95 mV more positive. Generally, the more chlorines on the biphenyl system, the more positive the E”’. Furthermore, 3,3’,4,4’-PCB, with a planar structure (I),has an E”’value more positive than that of its 2,2’,5,5’-isomer, which is less planar and thus should have less resonance stabilized activation toward electron attack. Structural factors which affect E”‘ appear also to influence biological activity, and we observe a crude correlation between Eo’ of the halobiphenyls and induction of cytochrome
173
(I
EO’ (V) vs. SCE -2.367
f
0.004
-2.272 f 0.007
-2.253 -2.034 -1.991
f f f
0.002 0.005 0.006
log k,,, s-l M-’
*
6.34 0.37 5.95 f 0.16 5.70 f 0.12 5.71 f 0.21 6.11 f 0.15
RSD,” % 18 27 15
15 9
Relative standard deviation of the regression.
P448 enzymes and toxicity. The compound with the most positive E”’, 3,3’,4,4’-PCB, is the best inducer of cytochrome P448 hydroxylase enzyme (25)and the most toxic of the PCB congeners investigated. 2,2‘,5,5‘-PCB, which has a lower cytochrome P448 induction activity because of its nonplanarity (25),has an Eo’ 43 mV more negative than 3,3‘,4,4’-PCB. The monohalobiphenyls, with the most negative E”’ values, are the least toxic and nearly inactive as inducers. In conclusion, the precision of the electrocatalytic method is measurably improved by the simultaneous nonlinear regression method. The method yielded standard potentials and kinetic constants on fairly complex irreversibly reduced redox couples, as illustrated by the halobiphenyls studied. Structural factors which influence standard potentials of the halobiphenyls appear similar to those which govern toxicity. Additional thermodynamic and kinetic information on halobiphenyls will be presented in a separate paper on their electrochemistry.
ACKNOWLEDGMENT We are grateful to Eugene Cioffi of Yale University for the a-naphthonitrile, to James Stuart of the University of Connecticut for 2,2’,5,5’-PCB, and the U S . EPA Pesticide and Industrial Chemical Repository for 3,3’,4,4’-PCB. Registry No. Benzo[flquinoline, 85-02-9; phenanthridine, 229-87-8; acetophenone, 98-86-2; 1,2-benzanthracene, 56-55-3; anthracene, 120-12-7; 9-phenylanthracene, 602-55-1; 9,lO-diphenylanthracene, 1499-10-1;a-naphthonitrile, 86-53-3; I-methoxybenzophenone, 611-94-9;4-cyanopyridine, 100-48-1;benzophenone, 119-61-9;perylene, 198-55-0;phthalonitrile, 91-15-6; tetracene, 92-24-0;terephthalonitrile, 623-26-7;bromobenzene, 108-86-1; chlorobenzene, 108-90-7; 3-bromopyridine, 626-55-1; 2-bromopyridine, 109-04-6; 2-chloropyridine,109-09-1; 3-chloropyridine, 626-60-8;4,4‘-dichlorobiphenyl, 2050-68-2; 4-chlorobiphenyl, 2051-62-9; 4-bromobiphenyl, 92-66-0; 2,2’,5,5’-tetrachlorobiphenyl, 35693-99-3; 3,3’,4,4’-tetrachlorobiphenyl,3259813-3.
LITERATURE CITED Hutringer, 0.; Safe, S.; Zitko, V. “The Chemistry of PCB’s”; CRC Press: Cleveland, OH, 1974. Cairns, T. C.; Siegmund, E. G. Anal. Chem. 1981, 53, 1183A-1193A. Wolff, M. S.;Anderson, H. A.; Selikoff, I. J. JAMA, J . Am. Med. Assoc. 1982, 247, 2112-2116. Bunce, N. J. Chemosphere 1982, 1 1 , 701-714. Oku, A.; Yasufuku, K.; Kataoka, H. Chem. Ind. (London) 1978, 21, 841-842. Farwell, S.0.; Beiand, F. A.; h e r , R. D. J . Electroanal. Chem. 1975, 6 1 , 315-324. Maruyama, M.; Murakami, K. Nippon Kagaku Kaishi 1976, 536-539. Rusiing, J. F.; Arena, J. V., unpublished work. Bard, A. J.; Faulkner, L. R. “Electrochemical Methods”; Wiiey: New York, 1980. Andrieux, C. P.; Biocman, C.; Dumas-Bouchlat, J. M.; Saveant, J. M. J . Am. Chem. SOC.1979, 101, 3431-3441. Andrieux, C. P.; Saveant, J. M.; Zann, D. N o w . J. Chim. 1984, 8 , 107-1 16. Andrieux, C. P.; Dumas-Bouchiat, J. M.; Saveant, J. M. J. Elechoanal. Chem. 1978, 8 7 , 39-53. Andrieux, C. P.; Dumas-Souchiat, J. M.; Saveant, J. M. J . Elechoanal. Chem. 1978, 8 7 , 54-65. Boujiei, K.; Martigny, P.; Simonet, J. J . Electroanal. Chem. 1983, 144, 437-442. Evans, D. H.; Xie, H. J . Am. Chem. Soc. 1983, 105, 315-320. Connors, T. F.; Rusling, J. F. J . Electrochem. Soc. 1983, 130, 1120-1 12 1. (17) Connors, T. F.; Rusling, J. F. Chemosphere 1984, 13, 415-420.
174
Anal. Chem. 1985, 57, 174-179
(18) Rusling, J. F.; Connors, T. F. Anal. Chem. 1983, 55, 776-781. (19) Andrieux, C. P.; Blocman, C.; Dumas-Bouchlet, J. M.; M'Halla, F.; Saveant, J. M. J. Electroanal. Chem. 1980, 113, 19-40. (20) Kojima, H.; Bard, A. J. J. Am. Chem. SOC. 1975, 9 7 , 6317-6324. (21) Miller, A. "ExDeriments I n Oraanlc Chemistrv": Buraess: Mlnneaoolls. - . MN, 1980. (22) Rusllng, J. F. Anal. Chem. 1983, 55, 1719-1723. (23) Meltes, L. "The General Non-Llnear Rearassion program C F T ~ A." .:Drivately published, 1983.
-
-
(24) Meltes, L. CRC Crlt. Rev. Anal. Chem. 1979, 8 , 1-53. (25) Goldstein, J. A. Ann. N. Y. Aced. Sci. 1979, 320, 164-177.
RECEIVED for review Julv 26, 1984. Accevted Sevtember 18. 1984. This work was supported by U.S. PHS Gr& ES03154 awarded by the National Institute of Environmental Health Sciences.
Reliable Use of Calibration Curves in Voltammetric Analysis with a New Technique of Microcomputer-Based Data Eva1uation A. M. Bond* and I. D. Heritage Diuision of Chemical and Physical Sciences, Deakin Uniuersity, Victoria 321 7, Australia
Cybernetic analytlcal voltammetry offers a powerful approach to efflclent management of various methods of quantltatlon when uslng translent techniques such as dlfferentlal pulse polarography and strlpplng voltammetry. I n thls work a mlcrocomputer system has been developed with capabllltles of asslgnlng the rellablllty of analytlcal determlnatlons based upon direct callbratlon from multltlme domaln experlments. The system Incorporates peak height, peak wldth, and potentlal measurements as a functlon of tlme and concentration to construct a comprehenslve data base as a reference. Unknown solutions are compared agalnst thls extensive data base to detect If Interference Is present and ultlmately to declde whether the dlrect callbratlon method Is approprlate or the more tlme-consumlng method of standard addltlon must be employed. Examples conddered Include the determlnatlon of sterolds by dlfferentlal pulse polarography or cathodlc stripping voltammetry and cadmium or lead by slmllar technIque8.
The calibration curve method of quantitation to perform large numbers of analytical determinations is attractive due to the inherent time efficiency compared to the alternative time-consuming standard addition method. The important prerequisite for reliable use of the direct calibration method is that the standard solution accurately reflects the matrix of the sample. Invariably, however, the analyst is confronted by a sample with little information concerning the matrix composition, and as voltammetric techniques such as differential pulse polarography and stripping voltammetry are very sensitive to matrix effects, the method of standard addition is almost universally recommended. The peak heights of these transient voltammetric methods are often kinetically controlled by rates of electron transfer, chemical steps, adsorption, etc. unlike limiting currents in DC polarography which are often diffusion controlled. Thus, the advantage of far greater sensitivity with the transient methods relative to DC techniques has associated with it the necessity for more frequent use of standard additions methods. From the viewpoint of time saving it would be of significant benefit to the analyst if a computerized form of instrumentation could be developed which enabled calibration curves to be used with transient voltammetric methods as the first
analytical strategy. Interference effects detected via inbuilt validation procedures might be identified. Remedial steps, which may include the method of standard addition, could then be judiciously applied as a second strategy. This integrated approach would then be far more efficient than the ad hoc use of either calibration or standard addition methods. Commercially available instruments for voltammetric analysis, even though they may be computer based, are not programmed to comment upon the integrity of the response. This is fundamental to the cybernetic-type approach to electrochemistry described by Faulkner et al. (1). In some laboratories sophisticated methods based upon pattern recognition and related learning techniques or multifrequency fast Fourier transform can and have been proposed for interference detection (2-8). In this laboratory we are endeavoring to use simpler procedures readily implemented with microprocessor-based systems. The approach adopted within this study is a new method of analyzing voltammetric experiments by a multitime domain procedure to assess whether or not the direct calibration method is valid. Implicit in this multitime domain approach is the comment that interference effects, behaving phenomenologically as a change in the electrode kinetics, will be statistically easier to detect than with conventional single time domain experiments. Acquiring experimental data in various time domains is made possible by the microcomputer-based data acquisition system of the kind described elsewhere (9-1 1). To use this system in the cybernetic approach described by Faulkner et al. (1) requires the addition of appropriate software for data evaluation. Differential pulse and stripping voltammetry were selected as examples of analytical voltammetry in this study due to their widespread use. In both instances, a peak-shaped output is the result of each experimental procedure. In differential pulse polarography, where the theory has been well developed, any detectable interferent affecting the charge-transferprocess will result in the alteration of the response in a number of ways (12-14), It is reasonable to assume the same is true of peak-shaped stripping voltammetric curves. The decision-making process of a computer system must by definition be a comparative procedure whereby the computer software compares a sample response with a value predicted either from an appropriate theory or from a previously compiled data base. As an example of the former in
0003-2700/85/0357-0174$01.50/00 1984 American Chemical Society
