Effect of Scan Rate on Third Derivative Curves. No extensive study of the effect of scan rate was made for the third derivative case. However, for the reduction of 2.01 X 10-3A11cadmium in 0.231 KCI a t a %drop electrode, measurement< of the heights of the second peak, a t --0.65 volt os. SCE, (second peak, Figure 11)) were made a t two widely different bcan rates, 21.8 mv. per second and 129 mv. per second. At 21.8 mv. per second, the peak height was -32.9 x amps per second3; a t 129 mv. per second, it was -21.9 X 10-3 amps per second3. The ratio describing the increase in the peak value is 665. The theoretical ratio predicted from the v7/*-dependence is 504. When the spherical correction term is taken into account, the increase ahould be 497 with an uncertainty of about 3%. The reason for the disci-epancy between the experimental and calculated ratios of the derivative peak.> is uncertain and is being investigated imrrently. CONCLUSIONS
The voltammetric first derivative technique has sevrral advantages over the conventional voltammetric method. The preci4on obtained a t the 0.4-11-12 level with cadmium indicates a t least an order of magnitude increase in sensitivity ; closely spiced reduction waves can be resolved readily; and the masking effects due to the presence of an excess of more easily reducible sub-
stances are minimized. Furthermore, the basic instrumentation involved is not any more complicated than much of that already used for conventional voltammetric work. Several further studies need to be carried out, however, to develop the derivat,ive method along promising lines. For example, the use of fast scans or second and third derivative techniques have not been investigated extensively here. This was primarily because further refinements in the instrumentation are needed before these techniques can be utilized readily, since background noise is a serious interference in these cases. Kit,h improved instrumentation further enhancement in the sensitivity of the derivath-e allproach should be attemllted by utilizing faster scans and:'or second and t'hird derivative techniques. Ailso,it might prove advantageous to apply derivative techniques to theoretical voltammetric studies where fast scans are utilized, since the charging current interferences would be minimized. LITERATURE CITED
(1) Barker, G. C., Anal. Chim. Acta 18, 118 (1958'1. ( 2 ) Barker, 'G. C., Jenkins, I. L., Analyst 77, 685 (19.52). ( 3 ) Bauer, H. H., J . ElectroanaL C'hem. 1, 256 (1960). (4) Booman, G. L., AXAL.CHEM.29, 213 (1957).
(5) Davis, H. AI., Seaborn, J. E., Electronic Engineering 2 5 , 314 (1953).
( 6 ) IleFord, I). D., 133rd lleeting, ACS, Sati Francisco, Calif., April 1958.
( 7 ) llelahay, P a d , "Sew Iirstrrimental Alethods i n Electrochemistry," 2nd ed., p. 119, Interscience, New York, 1954. ( 8 ) De Xlars, 11. U., Shain, Irving, ANAL. CHEM.29, 1825 (1957). (9) Frankenthal, 11. P., Shairi, Irving, J . -4m.Chem. SOC.78, 2069 (1Oj6). (10) Kelley, A I . T., Joties, H. C., Fisher, I). J . , ASAL. CHEM.31, 1475 (1959). (11) Alartin, K . J., Shain, Irving, l b i d . , 30, 1808 (19.58). (12) Sicholsori, R . S., Shain, Irving, I b i d . , 36, 706 (1964). (13) Perorre, d. P., Oyster, T. J., Ibid., 36, 235 (1064). (14) Pozdeev, N . AI., "1)ifTerential AIethod of Oscillographic Polarography," (Akad. Xaitk. 1 o.59.
R&, J. \V., De AIars, li. I)., Shain, Irving, SSAL. C f i E h i . 28, 1768 (1%56). (16) Rrrlfs, C. I,., J . .am. f'hem. Soc. 76,
(1;)
2071 11954). (17) Shairi, Irving, Xlartiii, IC. J., J . Phys. ('hem. 65, 254 i1961). (18) Fhekirii, L. Ya,, Ya.,R i r s s . J . Phys. Chem. 36. 1962). 36, 239 ((1962). (19) Smith, 1). E., I1einrnr1th, \V. H.. ASAL. CHEM.33, 482 (1!961). 120) Cnderkofler, FV. I.., Shain, Irving, Ibid.,33, 1966 11961). (21) T 7 0 n Starkelberg, AI., Priritan, AI., Toome. Toome, I-.,%. Elektrochem. 57, 342 (1953).
RECEIVED for review AIay 6, 1964. Arcepted Sovember 4, 1964. 1)ivision of Analytical Chemistry, 148th AIeeting, ACS, Chicago, Ill., September 1964. \Vork siipported i n part h>- L-. S. Atomic Energy Comniissioii utider Contract with the Union Carbide Corporation, arid ill part by the .4merieari Caiicer Society.
A pplication of Derivative Tech niq ues to Anod~icStripping Voltammetry S. P. PERONE and J. R. BlRK Department o f Chemistry, Purdue University, lafayette, Ind. Derivative voltammetry has been used in conjunction with anodic stripping analysis a t the hanging mercury drop elecirode. The derivative technique was compared in sensitivity, accuracy, and reproducibility to the conventional stripping technique, which involves direct measurement of the anodic voltammetric curve. A significant enhancement in sensitivity was attainable with the derivative technique. For solutions of cadmium(l1) as dilute as 8 X 10-W, only a 2-minute pre-electrolysis was required for a determination b y the derivative stripping technique, whereas a 10-minute preelectrolysis was necessary to obtain comparable sensitivity using direct voltammetric measurement. Cadmium (11) could b e determined in solutions as dilute as 6 X 10-"M b y the derivative stripping technique.
0
of the most sensitive techniques for trace analysis is anodic stripping voltammetry ( I , 3, 4, ?). Using this technique, heavy metal ions as dilute as 10-951 have been determined with electrolysis times of the order of 60 minutes. Accuracy of the order of +57G and precision of the order of i1 to 4% are obtainable, even though concentrations may vary over 4 or 5 orders of magnitude. The Iirimary disadvantage of the stripping analysis method is that a aizable time delay (electrol required before analyttral data may be obtained. This sacrifice is macle willingly in most rases in order to obtain the necessary semitivity. However, where short analysis times are critically import.ant, stripping analysis may be worthless at the lo-* to .If level. SE
The objectives of this study, therefore, were to develop a more sensitive stripping technique and to the trace level. 'The technique elected for use in the stripping step was deriyative voltammetry, n-hirh recent work ( 5 ) ha> hhon.n to be at least an order of magnitude 1 1 i 0 1 ~ ~>ensitive than conventional \-oltammetry. In addition, it i+ 1c.s sensitive to csharging current intc.rfPrrnce. to the erratic effects of + i d a c e film>. and to other interference> +iir.h a> 1)rrrding rharge transfers. The erlierimental Iirorcrlure for deriI.atiye btri1)1)iiig \-oltaninwtry i,s identical to that u h ~ dill coii\.entional strii>i)ingvoltanmirti~y,t3sccl)t that the drrivativr of the 5tril)l)iIig currentvoltagc r t i r v ~is r c ~ o r r l c d . Quantitative VOL. 37, NO. 1 , JANUARY 1965
9
correlation is made between the derivative Ijeak height and the product of bulk concentration and electrolysis time. In this study the concentration de1)endence was investigated for cadniiuin ion in KC1 solution. Results were obtained with bot’h the derivative and conventional voltammetric techniques for direct comparison. A linear relationship was found over the range of 2 X 10-5 to 4 X 10-loM for both techniques. In addition, the possible use of extremely short elect,rolysis times for quantitative determinations was studied using an 8 X 10-loLlf cadmium ion solution. 130th the derivative and conventional \-oltanimetric techniques were used. Electrolysis times as short’ as 2 minutes gave a measurable reproducible derivative stripping signal, whereas electrolysis times of the order of 10 minutes or more were necessary for the same sensitivity and reproducibility with conventional stripping voltammetry. The advantage of derivative stripping voltammetry in analysis of mixtures was also investigated. Comparison of derivative and conventional stripping curves for mixtures of Zn(II), Cd(II), Tl(I), and Ph(I1) ions indicates much iinliroved resolution with the derivative technique. EXPERIMENTAL
Instrumentation. The instrumentation for obtaining derivatives of linear potential sweep current-voltage curves has been described (5, 6). Some modifications in the electronic characteristics have been made for this study. .I IO-mf. capacitor (Stablex-D, Industrial Condenser Corp.. Chicago, Ill.) was substituted for the 1.0-nif. Alylar capacitor in the differentiator. Filtering was improved by adding more stage; with finer frequency selection. Care mas taken not to exceed input source impedance requirements whenever filtering preceded the input to any of the amplifier units or the recording device. Also. filtering was adjusted so that a test signal of 1.5 to 2 times the frequency of the experimental derivative signal aould get through unimpeded. These precautionary measures were periodically rechecked to assure valid measurements throughout the study. Cells and Electrodes. The cell and electrode assemblies have been described ( 5 ) . The cell and the Teflon lid supporting the electrodes, drop transfer scoop, and nitrogen disperser were mounted separately, so that it was not necessary to handle the various pieces of equipment, except for the outside of the cell, making it possible to minimize contamination and to keep a constant cell geometry. Reproducible magnetic stirring was obtained with a synchronous rotator (E. H. Sargent and Co., Chicago, Ill.) provided with a magnet attachment. The temperature was not controlled during these studies. The working electrode used for all 16
ANALYTICAL CHEMISTRY
experiments was a hanging mercury drop (HMDE). The preparation and use of the electrode assembly have been discussed ( 5 ) . Electrodes formed with 3 drops from the DME capillary were used in this mork, a fresh electrode being used for each run. The electrode area \\as 0.065 sq. em. The reference electrode was a large saturated calomel electrode, and the counter electrode was a X 3 inch graphite rod immersed in 0.1JI KC1 solution. All apparatus that came into contact with the sample solution were rinsed thoroughly with 1 to 1 HC1 solution, deionized water, 0.01M EDTA solution, and again with deionized water. The volumetric equipment and electrolysis cell were equilibrated with a given sample solution a t least three times for no less than 1 hour for concentrations of 2 X l O - * M and less. Reagents. All solutions mere prepared in water purified by distillation and passage over a mixed cationanion exchange resin bed. Solutions of less than 10-5M Cd+2 were prepared freshly as needed. All chemicals were reagent grade, used without further purification. The inert electrolyte was 0.0002L11 KC1 for all Cd+2 solutions except 2 X 10-5AM Cd+2, where 0,002M KC1 was used. These low electrolyte concentrations were necessary to niinimize the introduction of interfering heavy metal impurities. Because of the low electrolyte concentration, the cell resistance became as high as 24,000 ohms. No adverse effects were observed, however, since very small currents always were involved and a threeelectrode system was employed. High purity nitrogen was used to remove oxygen from the cell. I t was first
\
Figure 1 . Derivative and conventional stripping voltammetry
of Cd+2 Electrolysis time, 3 minutes Conventional, 2 X lO-’M Cd+* B. Derivative, 2.5 X lO-’M Cd+2 Curves ore photographic reproductions of original data
A.
passed through a gas-washing bottle containing zinc amalgam and chromous chloride solution to remove any oxygen present ( 3 ) and then through another gas-washing bottle containing the inert electrolyte solution. The minimum deaeration time used was 25 minutes. RESULTS AND DISCUSSION
Procedure. T h e electrolysis step was carried out a t -0.95 volt us. SCE for all Cd+2 determinations. When the mixture of Cd+2, Zn+2, Pb+2, and T1+ was studied, electrolysis was carried out at -1.25 volts us. SCE. At the end of each timed electrolysis, the electrolysis cell was switched out of the circuit and a dummy cell was switched in. Stirring was stopped and solutions were allowed to settle for 30 seconds. (The substitution of a dummy cell to end the electrolysis was necessary for maximum accuracy in measurements of short electrolysis times. Although spontaneous oxidation of the amalgam was not encountered in this work when the electrolysis cell was out of the circuit, it is a potential difficulty, and this procedure is not recommended ordinarily.) At the end of the settling time, the electrolysis cell was switched back in. An anodic potential sweep was initiated, and either the conventional or derivative voltammetric stripping curve was recorded. A sweep rate of 31.3 mv. per second was used throughout this study. For conventional voltammetric curves, peak measurements were made in the usual manner (1). With the derivative peaks, the height of the first peak was measured. The base line is simply the zero line when low sensitivity is used. For high sensitivity the irregular background signal results in an irregular derivative base line, and blanks must be obtained to get an accurate base line. In the case of very short electrolysis times (less than 10 minutes) and high sensitivity (concentration less than l O - 9 M ) , base lines could not be reproduced from one run to the next. Therefore, an assumed base line was sketched in for each of these runs. This type of base line was admittedly inaccurate, but was used only when sensitive measurements were desired with little need for maximum accurary. Concentration Dependence. The concentration dependence of t h e derivative stripping method was investigated for Cd+* in KC1 solution. The Cd+Z concentration was varied from 2.0 X to 6 X 10-I1M, and the resultant data are summarized in Table I. For comparison, the data obtained by conventional stripping voltammetry for the same solutions are given in Table 11. Typical derivative and conventional stripping curves are shown in Figures 1, 2, and 3. The precision of
I-
z w a a
3 0 V
B
0
3 5
VOLTS vs. S.C.E.
-0.8
-
I
Figure 2.
I
0 1
VOLTS vs. S.C.E.
Derivative stripping voltammetry
Figure 3. Derivative and conventional stripping voltammetry for 6 X
2 X 10% Cd+2, 10-minute electrolysis time 2 X 1 Oe8M Cd+z, 2-minute electrolysis time - - _ Experimental blanks - Photographic reproductions of original d a t a
A. B.
the derivative stripping method compai'es favorably with that obtained by conventional stripping, particularly for less dilute solutions,. The accuracy is not so good in ekher case ( 3 ~ 4 . 2to 4.57,)) but this is not unusual or unacceptable in cases where concentration is varied over several orders of magnitude. I n addition, temperature fluctuations, slight changes in cell geometry, equilibration losses, and leaching of vessels can all contribute to this error. The lowest concentration investigated, 6 x 10-ll.l.f Cd+2,is included in Table 1 with some reservation. I t was, of course, extremely difficult to prepare a solution of this concentration. Repeated efforts led to only limited success-that is, solutions of approximately t,he desired concentration could be prel)ared, but they were not stable. At 6 x 10-"Jf, the solutions deteriorated by about 10 to 20% per hour. Thus, the results of these studies are included not so much to indicate what is obtained from a 6 X 10-llJf C d f 2 solution, but rather to show that derivative signals corresponding to that low a concentration can he measured readily. Figure 3 shows the type of derivative stripping curve obtained a t this concentration for only 60 minutes' time. Comparison to conventional stripping is made. Obviously, the presence of Cd could not even be detected a t this concentration using conventional stripping \-oltammetry and a 60-minute electrolysis time. Use of Short 13lectrolysis Times. The increased sensitivity of the derivative stripping method can be used advantageously t o minimize electrolysis times. T o illustrate this point a very dilute solution of Cd A 2 (8 X M) was investigated by both derivative and conventional stripping voltam+*
+----&y \
-0.8
-0.4 L
lO-11M
metry. Electrolysis tinies were varied from 7 5 minutes down to 1 minute, and the data obtained are summarized in Table 111. Typical stripping curves are shown in Figure 4. Obviously, meaningful analytical data can be obtained below 10 minutes' electrolysis time only with the derivative technique. The precision decrease. with the shorter electrolysib times, but it is possible to determine 8 X 10-l" ?;I Cd+2 n i t h an electroly4s time of 2 minutes and a precision of *267,. If one extrapolates these results to concentrations of 2 X l O - 9 M and above, the conclusion is that determinations
Concn., C, moles/l.
x x 2 x 2 x 2 x 4 x
a
b
10-5
10-6 10-7
3 3
3 8 10-9 25 10-10 90 (6 X 60 Six replicate determinations at each Only three replicate determinations.
Concn.,
x x 2 x 2 x 2 x 4 x
2 2
with derivative stripping might be made with electrolysis times of no more than 1 minute; conventional voltammetric stripping would require at least 5 and probably 10 minutes for comparable sensitivity and precision. Thus, where rapid trace analysis is crucial and relative errors of the order of 20 to 30y0 are tolerable, the derivative stripping technique offers a very definite advantage over the con-
Derivative peak height, In', amp./sec. 8 . 46 46 X 9 12 X 9.12 x lo-' 10-7 7 68 X lo-*
2 40 X lo-* 7 i o x 10-9 5 49 x 10-9
5 8 X
Av. dev., I,'/Ct 0 141 0.141 0 1.52 152 0 138 0 150 0 142 0 153 0 16
n
/O
**Il . O 0 0 9 1 7
1 4 2 3 2 0 10
concentration.
Determination of Cd(ll) in KCI Solution by Conventional Stripping Voltammetry"
C, moles/l.
a
Electrolysis time, t , min.
10-8
Table II.
_-
Determination of Cd(ll) in KCI Solution by Derivative Stripping Voltammetry"
Table I.
2 2
Cd+
Electrolysis times, 60 minutes A. Derivative B. Conventional Arrow points out region where voltammetric peak should a p p e a r _ - _ Experimental blank Photographic reproductions of originol d a t a
10-5 10-6
lo-'
Electrolysis time, t , min. 3 3
10-8 10-9
3 10 45
10-1"
120
Peak height, z p , amp. 2.62 x 10-5 2.60 X 2 42 x 10-7 8 60 X 4.07 X
1.87 X Six replicate determinations at each concentration.
Av. dev., i,/Ct 0.437 0.436 0.403 0 430
5% &1.3
0.452
0.389
VOL. 37, NO. 1, JANUARY 1965
1.7
1.1
3 7
2.8 3 2
0
11
Table 111.
Conventional and Derivative Stripping Voltammetry with Short Electrolysis Times
8 X
ElecDerivative trolysis pk. height, time Ip’,amp./sec. t, min. X 75 ... 45 30 10 5 2 1
6.29 3.7 0.9 0.5 0.3 0.1
M Cd+z in KC1 solutions
Av. dev.,
Ip‘/t X 10’0
%
...
2i.8
1.38 1.2 0.9 1.0 1.5 1
4 10 21 26 100
Voltammetric peak current, i,, amp. X 10-8 2.10 ... 0.64 0.15 0.1
... ...
ip/t, 1010 2.80 ... 2.2 1.5 2
x
... ...
Av. dev.,
% zt3.1
‘ii
25 100
...
...
, -1.2
Six replicate determinations at each electrolysis time. Predicted values ( X 1010): I p ’ / t = 1.2; i p / t = 3.4
ventional voltammetric stripping method. Furthermore, if better precision is required, the derivative technique may be used with shorter electrolysis times than required by conventional stripping voltammetry to attain comparable precision. If one can be allowed to extrapolate the data from Table I11 in the other direction, a limit of detection is predicted of about 10-llM, considering electrolysis times no greater than 100 minutes.
Derivative Stripping Voltammetry of Mixtures. One of the distinct advantages of the derivative technique
I
-1.0
I
I
-0.6
-0.2
VOLTS vs. S.C.E.
Figure 4. Derivative and conventional voltammetry with short electrolysis time for 8 X IO-LoM Cd+2 Electrolysis time, 10 minutes A. Conventional 8. Derivative Arrow points out CdC2 peak. - --Assumed blank Photographic reproductions of original data
-
12
ANALYTICAL CHEMISTRY
is the improved resolution obtainable. To illustrate this characteristic, the anodic stripping curves for 2 X lO-*M Cd+2, Zn+*, P b f 2 , and TI+ ions in 0.002M KC1 were obtained by both derivative and conventional voltammetry. Figure 5 demonstrates the different stripping curves obtainable. Obviously, the qualitative detection of T l + ion in particular is improved by the derivative measurement. Quantitative measuremenb is still a problem, however, since accuracy depends on how good a base
>
!
,
v,
-0.4 VOLTS vs. S.C.E. -0.8
1
0.0
Figure 5. Derivative and conventional stripping voltammetry for mixture of metal ions Electrolysis time, 10 minutes. Concentrations of all ions 2 X 10 A. Conventional 8. Derivative Both curves a r e photographic reproductions of original data
line may be obtained. It seems reasonable to predict better accuracy with the derivative measurement, simply because of better resolution, but no data have been obtained yet to verify this. LITERATURE CITED
(1) De Mars, R. D., Shain, I., ANAL. CHEM.29, 1825 (1957). ( 2 ) Kemula, W., Kuhlik, Z., d nal. Chim. Acta 28, 104 (1958). (3) Meites, L., Ibid., 18, 364 ( 1958). (4) Nikellv, J. G.. Cooke. W. D., ANAL. ( 5 ) Perone,’S. P.,~Mueller, T. R., ANAL. CHEM.37, 2 (1965). ( 6 ) Perone, S. P., Oyster, T. J., Ibid., 36, 235 (1964).
(7) Shain, I., “Treatise on Analytical Chemistry,’’ I. 11. Kolthoff and P. J. Elving, eds., Part I, Section D-2, Cham 50. Interscience. New York.
RECEIVED for review September 17, 1964. Accepted October 22, 1964. 12th Anachem Conference, Detroit, Mich., October 1964. Investigation supported in part by Public Health Service Research Grant Xo. CA-07773-01 from the National Cancer Institute.
