A. M. Hartley and R. i. Asai, Anal. Chem., 3 5 , 1214 (1963). G. B. Jones and R . E. Underdown, Anal. Chem., 25, 806 (1953). A. D. Westiand and R. R. Langford, Anal. Chem., 28, 1996 (1956). "Standard Methods for the Examination of Water and Wastewater", 12th ed., American Public Health Association, inc., New York, N.Y., 1965, pp 195-205. I. M. Koithoff,W. E. Harris, and G. Matsuyama, J. Am. Chem. SOC.,66, 1782 (1944). J. W. Coliat and J. J. Lingane, J. Am. Chem. Soc., 76, 4214 (1954). M. G. Johnson and R. J. Robinson, Anal. Chem., 24,366 (1952). Ph. Mecheiynck and C. Mechelynck-David, Anal. Chim. Acta, 21, 432 (1959). R. E. Hamm and C. D. Withrow, Anal. Chem, 27, 1913 (1955). M. C. Rand and H. Heukeiekian,Anal. Chem., 25, 878 (1953). A. M. Hartiey and D. J. Curran, Anal. Chem., 3 5 , 686 (1963). R . J. Davenport and D. C. Johnson, Anal. Chem., 45, 1979 (1973).
(16) G. L. Lundquist, G. Washinger, and J. A. Cox, Anal. Chem., 47, 319 (1975). (17) D. R. Senn, P. W. Carr, and L. N. Klatt, Anal. Chem., 48, 954 (1976). (18) E. D. Wood, F. A. J. Armstrong, and F. A. Richards, J. Mar. Bid. Assoc. U.K., 47, 23 (1967). (19) A. D. GooisbyandD. T. Sawyer, Anal. Chem., 39,411 (1967). (20) F. J. Welcher, "The Analytical Uses of EthylenediaminetetraaceticAcid", D. Van Nostrand Co., New York, N.Y., 1960. (21) J. W. Clark, W. Viessman, Jr., and M. J. Hammer, "Water Supply and Pollution Control", 2d ed., international Textbook Co., London, England, 1971, p 234.
RECEIVEDfor review September 27,1976. Accepted December 6,1976. This work was supported by the National Science Foundation under Grant No. CHE 73-05204.
Normal Pulse Voltammetry in Electrochemically Poised Systems J.
Lee Morris, Jr.,
and Larry R. Faulkner"
Department of Chemistry, University of Illinois, Urbana, Ill. 6 180 1
Normal pulse polarography generally gives a severely distorted view of solutlon composition in a poised system, because the working electrode is active during the interval between pulses. Two ways for preventing the distortion were explored. In one scheme, the conventional depolarized reference electrode was replaced by a quasi-reference electrode (QRE), and the usual potential waveform was modifled so that the potential was held at 0 V vs. ORE between pulses. This method permits analysis In poised systems without distortion. The second scheme Involved the insertionof an electronic switch to free the working electrode from potentiostatlc control except during potential pulses. Restoration proceeds by dlffusion, but is Inhibited by coulostatic discharge of the double layer.
The past several years have seen widespread exploitation of the decreased analysis times and increased signal-tobackground ratios offered by normal pulse polarography ( 1 , 2 ) , as compared to the conventional dc method. However, there is a generally unrecognized drawback to this pulse technique that severely affects its application to electronically poised systems, Le., systems which display a true equilibrium potential because both oxidized and reduced forms of a given redox couple are present. For the purpose of this discussion, we include among poised systems not only those fulfilling the strict sense of the term, but all others for which the voltammetric curves do not show a readily predictable region of zero shows faradaic current. For example, Fe(II1) in 1 M "03 zero current at the dropping mercury electrode only when the reduction of Fe(1II) is exactly compensated by the oxidation of Hg. The difficulty in these cases stems from the form of the potential program applied to the working electrode. That program involves the imposition of a base potential during the long waiting periods (0.5-4 s) between the potential pulses which stimulate the current that is actually sampled. The usual assumption is that negligible electrolysis occurs during the waiting period, so that the solution composition, as sampled in the pulse, is the same as that of the bulk. In a poised system, this assumption can hold only if the base potential coincides fortuitously with the electrode equilibrium potential. Otherwise, electrolysis during the waiting period modifies the solution composition near the electrode, and the pulse measurement gives a distorted view of the bulk composition. This problem was brought to our attention during an ex-
ploration of pulse methods as means for enhancing the sensitivity of the ferrioxalate actinometer ( 3 ) .The analysis step of interest is the measurement of a small concentration of Fe(I1) in the presence of a high concentration of Fe(II1). By itself, Fe(I1) can be measured via normal pulse polarography at a solid electrode carried through a positive-going scan. The base potential would be near 0.2 V vs. SCE. In the presence of Fe(III), this base value causes electroreduction which yields Fe(I1) directly and destroys the ability of the experiment to reflect the true bulk concentration of Fe(11). This particular analysis problem was considered recently by Parry and Anderson ( 4 ) ,who were able to apply pulse polarography to it by finding a supporting electrolyte (0.1 M sodium pyrophosphate) in which the Fe(II)/Fe(III) couple falls within the working range of the DME and also is sufficiently irreversible that a well-defined zero-current region divides the anodic and cathodic waves. They could determine Fe(1I) and Fe(II1) separately by setting the base potential in the zero-current region and sampling the currents stimulated by pulses to the plateaus of the two waves. We have been interested in approaches that can make pulse voltammetry more generally useful for poised systems. In this paper, we present and evaluate two approaches. One, featuring a polarizable reference electrode, proved useful. Another scheme, which depended upon electronic interruption of current at the working electrode, did not succeed in an analytically useful manner. Even so, we describe our experience with it because (a) the reasons for its failure are not obvious, (b) our results bear on reports in the literature, and (c) we include some theory that accounts for the failure and also applies directly to certain cases of normal pulse voltammetry at stationary electrodes.
EXPERIMENTAL Apparatus. All experiments were controlled and monitored via an interface to a Data General Nova 820 minicomputer. The potential program was generated by the computer, which acted through a 12-bit D/A converter driving an auxiliary input to a Princeton Applied Research Model 174 potentiostat. The current response of the electrochemical cell was monitored a t the potentiostat current-to-voltage ( I I E )converter (Princeton Applied Research Model 176) by means of an 8-bit A/D converter (Date1 Model ADC-EH1,4-fis conversion time). Sampled current voltammograms were displayed immediately after acquisition on an oscilloscope driven by parallel 8-bit D/A converters. Individual data sets could also be stored on magnetic tape cassettes for later reference. ANALYTICAL CHEMISTRY, VQL. 49, NO.
3, MARCH 1977
489
Figure 1. Logic-controlled analog switch LSB = logic input; Q1 = 2N5135; 0 2 = 2N4356; Q3 = TIS73 (five in parallel); D1 = signal diode; D2 = 1N4740A. Resistances in ohms, capacitances in fif
A
\ \ \ \
A
r\\,,
5
TI
I2 W
I-
O
.
0
.
0.
E
I
- 7 4
0
c z LE LE I
TIME
3 0
Figure 2. Potential program for gated pulse voltammetry t, = 1-200 ms; T = 5 ms-4 s; A € = (width of potential scan)/255
An electronic switch was inserted into the potentiostatic circuit between the working electrode and the input to the ZIE converter. The purpose of this device will be amplified later. Its effect was to remove the working electrode from potentiostatic control upon demand. The operation of the switch, as indicated in the circuit diagram in Figure 1,was controlled by a binary logic line from the computer interface. Application of a T T L logic “1”to the control input rendered the path from “A” to “B” conducting. A logic “0”, opened the path. Five parallel FET’s were required to increase the current handling ability of the switch to about 500 mA, and the back-to-back zener diodes were necessary to protect the device against the 100-V compliance of the potentiostat. In the conducting state, the resistance through the switch was 1 Q . Three main FORTRAN programs, together with attendant assembly language subroutines for interface handling, were developed in the course of this work. The first was an electrode pretreatment program designed to carry a working electrode through a series of potential steps of 0.1-s duration. Alternating step potentials of 1.0 and -1.0 V vs. the reference were employed, and the series always ended after a negative half-cycle. Pretreatments of this sort provided marked improvements in experimental reproducibility a t Pt electrodes in HCl media. The second program was a very versatile controller for pulse voltammetry. The important parameters for any experiment could be specified within wide ranges by direct entry a t the keyboard. The interval between pulses (or the drop time in experiments with a DME) was variable from 5 ms to 4 s, and the pulse duration could be set from 1to 200 ms. The initial and final potentials for the scan in pulse height could be specified arbitrarily within the range of -5 V to 5 V vs. the reference electrode. As usual, the initial value of this scan also served as the base potential in normal pulse voltammetric experiments. Each current sample, taken a t the end of a pulse, was represented by the 490
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
0
0
E, vs.
SCE
Figure 3. Depletion effects in pulse voltammetry (A) Normal pulse voltammetry: tp = 20 rns; T = 100 ms. (B-D) Gated pulse voltammetry: (B) t, = 20 rns; T = 100 ms. (C) tp = 20 ms; T = 500 ms.(D) t, = 20 ms; 7 = 4 s. Solution was 10 mM Fez+ in 1 M HCi and was not stirred. Upper trace in each frame is background
average of a set of 20 sequential voltage conversions (7 p s each). In addition, the electronic switch could be used to remove the working electrode from potentiostatic control during the entire pulse interval. The final program was identical to the controller just described, except that the base potential applied during the pulse interval was fixed a t 0 V vs. reference. This variation was designed for use with a polarizable reference electrode, as explained later. Reagents a n d Electrodes, All test solutions were prepared from reagent grade chemicals without further purification. Solutions containing Fe(I1) were made from ferrous ammonium sulfate, and those containing Fe(II1) were prepared from ferric chloride. The supporting electrolyte was either 0.1 M or 1M HC1.
Table I. Degree of Restoration after Time 7 Following Pulse Width t , T/t p 1 5 10
25 50 100
500 1000
2500 10000
C(0,T
Table 11. Effective Pulse Width under Various Conditionsa
+ t,)/C*
0.500 0.732 0.805 0.874 0.911 0.937 0.972 0.980 0.987 0.994
cm2served A polished Pt disk having a geometric ai-ea of 1.5 X as a working electrode. In most experiments,the reference electrode was a commercial SCE with an asbestos fiber salt bridge. In others, a Pt flag having an approximate area of 0.5 cm2 was used as a polarizable reference ( 5 ) ,or quasi-reference, electrode (QRE). A similar flag was used as the counter electrode in all experiments. RESULT-S A N D DISCUSSION We are concerned here with means for ensuring that the working electrode in a voltammetric experiment always faces a solution having the bulk composition whenever a potential pulse rises. If a DME is employed, the convection accompanying drop fall essentially erases the disturbing effects of the previous pulse; hence, one requires only a scheme for preventing faradaic disturbance of the composition during the succeeding pulse interval. For a solid electrode, the problem is more complicated, because the pulse interval must be actively utilized to restore the diffusion layer to the bulk composition after each pulse occurs. Gated Pulse Voltammetry. Our first attempt a t a solution to this problem arose from the work of Kirschner and Perone (6), who employed a gated working electrode in flash photoelectrochemical experiments. Their object was to follow the concentration of a transient species by sequential current measurement under potentiostatic conditions. To eliminate the effects of faradaic alteration of the solution composition, the working electrode was freed from potentiostatic control, except for the brief periods during which current sampling took place. We modified their method to allow sequential current measurements a t successively increasing potentials, as one does in normal pulse voltammetry. The potential program for this gated pulse technique is illustrated in Figure 2. At the end of each pulse, just after the current sample has been taken, the electronic switch is opened so that the working electrode is released from the potentiostat. It remains free throughout the pulse interval, T, and is reconnected a t the start of the next pulse. Isolating the working electrode during the pulse interval provides an opportunity for diffusion to refill the diffusion layer formed in the previous potential excursion. If the pulse interval is long enough compared to the pulse duration, the gated electrode method will allow the electrode to return to equilibrium, and it will permit the solution composition near the electrode to achieve the bulk value before the next pulse is executed. T o estimate the time required for restoration, Kirschner and Perone (6) alluded to an analogous problem in heat transfer (7). In effect, it describes the evolution of concentrations C(x,t) in a thin layer of solution displaying an initial profile C ( x , o ) , under the conditions that the boundary a t x = 0 has zero flux (i.e., no electrode reaction) and the boundary a t x = I is held at a fixed concentration C*, corresponding to the bulk value. They concluded that a waiting period, T,
( t p h f ,ms C*, M 10-2
t , = 1 ms 1.09
t , = 10 ms 10.3 13.1
t, = 100 ms
2.13 60.1 31.3 2400 10-5 2200 a (E112- E ) = 200 mV, c d = 25 bf/cm2, n = 1,D 10-3 10-4
101
109 213 3100 = 10-5 cm2/
S.
greater than 24 times the pulse width, t,, ought to restore the bulk concentration. However, their experiments showed evidence for cumulative depletion effects, even when ?Itp = 38. In our gated pulse technique, we observe similar problems. Figure 3A shows a normal pulse voltammogram of a solution of Fe(I1). Since this system is chemically reversible, Fe(II1) created during a pulse is reduced back to Fe(I1) by potentiostatic electrolysis during the following waiting period. Thus each pulse begins with the working electrode facing a solution of bulk composition. These measurements do not suffer from cumulative electrolytic distortion; hence, the response of Figure 3A is that which ought to be seen in the corresponding gated pulse experiment if T is long enough to effect restoration of the diffusion layer. Figures 3B-D show gated pulse voltammograms for t , = 20 ms and 7 = 0.1,0.5, and 4 s. Depletion effects are clearly reduced as T increases, but even for 7 / t , = 200, the wave height in the gated pulse voltammogram is significantly below that in the normal pulse experiment. Failure to restore the diffusion layer in these long periods seems surprising in light of the predictions of Kirschner and Perone, but the underlying reasons are discernible. One aspect is that enormous waiting periods are required for complete (99% or 99.9%) restoration, even though the 90% level is reached rather quickly. This point is demonstrated by a rigorous solution to the problem in which electrolysis takes place for 0 5 t 5 t,, but is prevented for t 2 t,. See the Appendix for details. If the initial electrolysis is diffusion controlled, the surface concentration C ( o , t ) is zero before t,. Afterwards, it grows back to the bulk value C*, and we can regard its progress as a measure of restoration. We find where St ,(t) is the step function that rises from zero to unity at t = t,. The variable 7 is t - t,. Table I lists C(o,t, T ) / C *for selected values of Tit,. One sees that the Kirschner-Perone criterion, 7/tp = 24, restores the surface concentration only to 0.87 C*. Complete restoration requires ?Itpin excess of 2500. For E in the rising portion of the wave, restoration for a given r / t , would probably be more complete; however, a precise treatment would be very complex and limited in utility. Even though the departure from complete restoration is small in the convenient operational range of 50 < r / t p < 1000, one must realize that the electrolysis/restoration cycle is repeated many tens of times, and the departures in each cycle tend to accumulate. This cumulative effect is essentially the basis for the fall in the sampled current after the peak in Figure 3B. In real systems, a serious complication compounds the restoration problem when the concentration C* is low. The derivation of Equation 1was founded on the assumption that the surface flux of the electroreactant is zero for t L t,. This will not hold until the faradaic process removes enough capacitive charge to shift the electrode potential to a farada-
+
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
491
A
p--(
lropa
I
E , VS. QRE
+ W z
a a 3
B
0
T
I
I
C
C
0.6
0.4
0.2
1
E , V S . SCE
Figure 4.
0.4
0.2
E, VS. SCE
Voltammetry of a poised system
A mixture of 1 mM Fez+ and 10 mM Fe3+ in 0.1 M HCI, with stirring. (A) Gated pulse voltammetry: tp = 20 ms; T = 500 ms. (B) Normal pulse voltammetry: t, = 20 ms; T = 500 ms
ically inactive region. In essence, a coulostatic discharge lengthens the effective width oft, (8-10). An idea of the size of this effect can be gained by considering an experiment in which a step is made a t t = 0 to potential E , where electrolysis is diffusion limited. The step is applied for a period t,, then the electrode is disengaged from potentiostatic control. Electrolysis proceeds coulostatically. We will assume that the diffusion-limited flux is consumed until the potential falls back to Ellz. The time at which E reaches E112 is designated the effective pulse width (tp)eff, and after that time the electrolysis rate, and the surface flux, are regarded as zero. The charge that must be removed from t, to (t,),ff is Cd(E1/2 - E ) A , where C d is the differential double layer capacitance, which we take as independent of potential from E to Ell2. This charge is leaked as the integrated faradaic current required for the Cottrell flux, hence
or (3)
Table I1 shows the effective pulse width as a function oft, and C* for ( E - El/*)= 200 mV, D = 10-5 cm2/s, n = 1,and c d = 25 pf/cm2. Given the effective width, the degree of restoration during a waiting period T can be approximated by substituting (tp)eff/T into Equation 1 in place of tp/T. Perhaps the most striking feature of Table I1 is the extremely strong dependence of (tp)eff on C*. At the 10 mM level, the Cottrell flux after t , is sufficiently great to discharge the double-layer rather quickly, and ( tp)eff differs little from t,. At the loF5M level, discharge requires periods comparable 492
0.6
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
Figure 5. Voltammetry of a poised system A mixture of 1 mM Fez+ and 1 mM Fe3+ in 1 M HCI, without stirring. (A) Modified pulse voltammetry with a ORE. (B) Normal pulse voltammetry with an SCE, positive scan (€i = 0 V). (C) Normal pulse voltammetry with an SCE, negative scan (E, = 0.8 V). In all cases t, = 20 ms; T = 100 ms
to the longest restoration times we have used. In such an instance, restoration does not even begin before the next pulse is applied, and we actually converge to a limiting mass transfer problem that is identical to staircase voltammetry. These results show that the gated pulse method offers no utility for analysis in quiescent solutions. On the other hand, convective transfer greatly increases the flux of electroreactant into the depletion region and produces a considerable shortening of the necessary restoration period. We found that simply stirring the solution magnetically a t a moderate rate could remove depletion effects altogether for C* 2 10-3 M. Mixtures of oxidation states could then be determined without electrolytic distortion, as shown in Figure 4A. The voltammogram displayed there was recorded for a solution having a 1mM concentration of Fe(I1) and a 10 mM concentration of Fe(II1). The anodic and cathodic components indeed show a tenfold difference in height. The normal pulse voltammogram in Figure 4B is distorted enormously by preelectrolysis of Fe(II1) during the waiting period T , and the two wave heights are virtually equal. Unfortunately, depletion effects become evident below 1 mM, even with stirring, apparently because (t,),tt grows to unmanageable levels. One could possibly extend the range a bit further by using a rotated electrode ( I I ) , but the impetus for doing so is small. We note in closing this section that Equation 1 also describes the progress of restoration in a normal pulse experiment involving an irreversible electrode reaction at a stationary electrode. Our results suggest that the use of conventional t , and T values will not permit complete restoration, and distortion will be seen in the voltammograms. Moderate
stirring during the recording of the voltammograms probably would eliminate the problem. Use of a Polarizable Reference. An alternative approach to the problem of restoration is to return the potential after each pulse to the equilibrium value. If the electrode reaction is chemically reversible, potentiostatic electrolysis will then actively serve to restore the bulk conditions. If reversibility does not prevail, diffusion must accomplish the restoration, as in the gated electrode method. In either case, the potentiostat removes the double-layer charge immediately; hence, there is no coulostatic discharge to lengthen the effective pulse width. Fisher, Belew, and Kelley (5)have shown that a polarizable electrode, such as a Pt flag, can substitute in a three-electrode cell for the usual strongly poised reference. Such a polarizable reference is usually called a quasi-reference electrode (QRE), because its potential is not constant on a thermodynamic scale but instead depends on solution conditions. It eventually adopts the potential of zero net current; thus one can maintain a working electrode at its own equilibrium potential, without foreknowledge of that potential, simply by holding it a t 0 V vs. a QRE made of the same material. We have exploited this fact by developing a new voltammetric method expressly intended for use with a QRE. I t essentially is a normal pulse technique, but the base potential is always 0.0 V vs. QRE. Step potentials are scanned from negative to positive values (or vice versa), so that both anodic and cathodic currents are recorded. Figure 5 shows a comparison of voltammograms for a solution containing l mM Fe(I1) and l mM Fe(II1). In frame A one sees the modified pulse voltammogram for a scan from -0.3 V to +0.3 V vs. QRE. The anodic and cathodic components are essentially equal in height, as one expects from the concentration ratio. Figures 5B and 5C display normal pulse voltammograms corresponding to potential scans of 0.0 to 0.8 V and 0.8 V to 0.0 V vs. SCE, respectively. The base potentials for these two experiments were 0.0 V and 0.8 V vs. SCE; thus Fe(II1) was reduced continuously during the pulse interval in the former experiment, and Fe(I1) was oxidized throughout that period in the latter. Both voltammograms show gross distortions arising from this pre-electrolysis. In particular, the substance which is active at the base potential hardly registers a t all in the corresponding voltammogram. The utility of the QRE for preventing this effect is strikingly apparent. We have found that with no attention whatsoever to cleanliness of the solution or to the area or surface condition of the reference electrode, the modified pulse method can be utilized effectively with the Fe(II)/Fe(III)couple to the lop4 M level. Surface processes on Pt interfere significantly with analysis a t lower concentrations. Of course, similar problems are common to nearly all techniques involving solid electrodes. With care and favorable surface properties, one could probably extend the operational range of the QRE to l O - S - l O - ~ M. The lower limit would be determined by the buffering capacity of the system with respect to potential (12).As the concentration of electroreactants falls, the experimental precision with which the equilibrium potential is defined becomes poorer, and the time required to achieve it becomes longer. These factors are likely to restrict the QRE to the range above 10-6 M.
APPENDIX The problem leading to Equation 1 was stated just before the presentation of that result. We assume semi-infinite linear diffusion. I t is well known that Laplace transformation on t of Fick's second law yields an ordinary differential equation which, under the initial condition C ( x , o ) = C*and the semi-
-
infinite condition Lim ( x m ) C ( x , t ) = C* has the solution C ( x , s ) = C*/s A ( s ) e - ( S / D ) " 2 x (A-1) where s is the transform variable, c ( x , s ) is the transform of C ( x , t ) , and A ( s )is an undetermined function. The remaining conditions are
+
C(o,t)= 0
for 0
< t < t,
(A-2)
fort L t ,
J(o,t) = 0
(A-3)
where J ( o , t )is the surface flux of the electroreactant and t , is the pulse width. The solution to this problem has the form C(x,t) = C'(x,t)
+ St,(t)C'I(x,t - t,)
(A-4)
where St,(t) is the step function that rises from zero to unity a t t = t,. Condition A-2 can be expressed C ( o , t ) = St,(t)C"(o,t -
- tP)
(-4-5)
C(o,s) = e-7sGI(o,s)
(A-6)
Comparison of Equations A-1 and A-6 shows that -C*
+
A(s)= S e-TsCII(o,s) Substitution of Equation A-7 into A-1 permits an evaluation of the transformed flux. Since J(o,s) = D[dE(x,s)/dx],=o,
J(o,s) = JI(o,s) + e-rsJII(o,s)
which rearranges to,
- e-rs[s1/2D1/2~11(o,s)+ J'I(o,s)]
=0
(A-9)
This relation must hold for a range of s, hence each bracketed term must be zero. The first yields the Cottrell result, D l/ZC* J'(o,t) = (A-10) T 1 / 2 t 1/2 The second gives, (A-11) Condition A-3 can be written,
+
J1(o,s) e-rsJ1l(o,s) = gI(o,s) - L(S,,(t)J'(o,t)) (A-12) which shows that e-T,TP(o,s) = -L{St,(t)J'(o,t)) (A-13) Substitution of Equation A-13 into A-11 and inverse transformation by convolution gives
S,,(t)C'(o,t - t,) = C ( o , t )
Substitution from Equation A-10 and evaluation of the integral gives Equation 1.
LITERATURE CITED Z.Anal. Chern., 173, 79 (1960). (2) E. P. Parry and R . A . Osteryoung, Anal. Chern., 36, 1366 (1964). (3) C . G. Hatchard and C . A . Parker, Roc. R. SOC.London, Ser. A, 235, 518 ( 1956). (1) G. C. Barker and A. W. Gardner, Fresenius'
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
493
(4) E. P. Parry and D. P. Anderson, Anal. Chem., 45, 458 (1973). (5) D. J. Fisher. W. E. Belew. and M. T. Kellev in “Polaroaraohv 1964”. G. J. Hills, Ed., Interscience, New York, 1966,‘pp 1043-1659 (6) G. L. Kirschner and S. P. Perone, Anal. Chem., 44, 443 (1972). ( 7 ) H. S. Carslaw and J. C. Jaeger, “Conduction of Heat in Solids”, 2nd ed., Clarendon Press, Oxford, 1959, p 100. (8)W. H. Reinmuth and C. E. Wilson, Anal. Chem., 34, 1159 (1962). (9) P. Delahay, Anal. Chem., 34, 1161 (1962). (IO) J. Bonastre, M. Astruc, and J. L. Bentata, Chim. Anal. (Paris), 50, 113 (1968).
(11) D. J. Myers, R. A. Osteryoung, and J. Osteryoung, Anal. Chem., 46, 2089 (1974). (12) P. Delahay, “New Instrumental Methods in Electrochemistry”, Interscience, New York, 1954, p 43ff.
RECEIVEDfor review October 12,1976. Accepted December 13,1976’ We are grateful to the Science Foundation for supporting this work through Grant MPS-75-05361.
Coprecipitation and Electrodeposition of Polonium from Sea Water J. P. Cowen, V. F. Hodge,” and T. R. Folsom Scripps Institution of Oceanography, L a Jolla, Calif.
92093
Polonium can be electrodeposited onto carbon rods dlrectly from acidified sea water, stripped from the rods, and autoplated onto silver counting disks with an overall recovery of tracer of 40 f 2% for an electrodepositiontime of 16 h, 61 f 6% in 24 h, and 85 f 4% In 48 h. Coprecipitation of polonium from sea water, previously acidified then neutrallred, results In recoveries of 80 f 8 % (including autoplating) when 6 mL 1 N NaOH/L are used in the precipitation. Absolute *loPo activities in sea water by the two methods agree. Filtration of coastal sea water through a glass wool plug removes 32% of the *loPo activity.
Similar affinity of polonium and plutonium for marine surfaces implies that studies of the more easily measured polonium might be valuable in predicting some consequences of plutonium disposal in the oceans (1-4). Rates at which 239,240Pu and 210Podeposit out of sea water onto surfaces of giant brown algae and “inert” surfaces, such as glass and cellulose, suggest that both nuclides are associated in coastal sea water with colloidal sized species having diffusivities of about cm2/s. This parallel behavior possibly represents an 3X initial step in the incorporation of both a-radioactive heavy elements into marine food webs and/or their transport to coastal sediments. The study of zlOPois facilitated by the greater activity concentrations found on marine surfaces and in sea water, about 200 times that of plutonium. Such studies require many determinations of the natural a-radioactive zlOPoin sea water. Thus a simple, high recovery procedure was sought. Tsunogai and Nozaki analyzed Pacific Ocean surface water by consecutive coprecipitations of polonium with calcium carbonate and bismuth oxychloride after addition of lead and bismuth carriers to acidified sea water samples ( 5 ) . After concentration, polonium was spontaneously deposited onto silver planchets. Quantitative recoveries of polonium were assumed a t the extraction steps and plating step. Shannon, Cherry, and Orren, who analyzed surface water from the Atlantic Ocean near the tip of South Africa, extracted polonium from acidified samples as the ammonium pyrrolidine dithiocarbamate complex into methyl isobutyl ketone (6). They also autoplated polonium onto silver counting disks. An average efficiency of 92% was assigned to their procedure after calibration with 210Po-210Pbtracer experiments. Our experience with autoplating polonium from several thousand biological and sea water samples is that yields are often about 85% but it is not uncommon to get yields of from 50-100%. Likewise it appears risky to assume that removal of 494
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
polonium from sea water by coprecipitation will always be nearly quantitative because of its high adsorbability. Two simple procedures for concentrating 210Po from sea water are compared here: coprecipitation upon partial precipitation of the natural calcium and magnesium with sodium hydroxide (7) and a new method, electrodeposition of polonium directly from acidified sea water onto carbon rods. Polonium thus concentrated, was autoplated onto silver counting disks held in spinning Teflon holders. Recoveries of zlOPowere monitored by the addition of zOsPo.
EXPERIMENTAL Prior to collection, enough reagent grade 12 M HCl was added to acid-washed polyethylene sample bottles to bring the acidity of the sea water to about 0.5 M. Four to 7 L of sea water were found to give a 1u counting error of 10%or less with counting times of 1000 to 2500 min. The time between collection and analysis was usually less than 1week, preferably 1or 2 days. In the laboratory, samples were accurately weighed into 11-L polyethylene buckets (remembering to subtract the weight of the added HC1) and about 0.5 pCi of *08Po (z08Potracer solution, E.R.D.A. Health and Safety Laboratory, New York, N.Y. 10014) was mixed into the sample by means of a magnetic stirrer. Partial Precipitation Procedure. The sample was neutralized with 50% (20 M) NaOH (Baker “Analyzed Regent”) until a slightly cloudy precipitate persisted. Then 12 M HC1 (Baker “Analyzed Regent”) was added dropwise until the precipitate just dissolved. (The addition of a few drops of a 0.04% solution of bromothymol blue facilitated observation of the disappearance of the precipitate.) Six mL of 1 M NaOH/L of sea water was then added, giving a milky precipitate. After 2 h of stirring, the precipitate was allowed to settle overnight. The liquid was siphoned off and the remaining precipitate transferred with several small deionized water rinses to a 1-L polyethylene bottle whose top had been cut off and centrifuged. The precipitate was dissolved with 1 2 M HC1 and the resulting neutral solution (about 40 mL) adjusted to 0.5 M with 12 M HC1. This solution was transferred to a 100-mL glass beaker with three 0.5 M HCl rinses, bringing the final volume to about 50 mL. Polonium was autoplated onto silver. Electrodeposition. Carbon rods (Ultra Carbon, P.O. Box 747, Bay City, Mich.) appropriately wired to a constant voltage power supply were inserted into 6 to 7 L of sea water which had been acidified to 0.5 M with 1 2 M HCl and spiked with zOsPo. One set of four 30 cm X 0.30 cm rods, held securely in a polyethylene holder a t the corners of a 1.5-cm square, acted as the cathode and another functioned as the anode. The anode and cathode holders were held approximately 7 cm apart. Polonium deposited onto the cathode rods. Stirring was accomplished by means of a 6-cm Teflon-coated magnetic bar. At a constant voltage of 2.3 V, vigorous stirring, and an acid concentration of 0.5 M, the set-up was allowed to plate for 16-48 h. (The whole apparatus should be located in a fume hood to remove corrosive gases.) After plating, the cathode rods were withdrawn from the sea water with the voltage still applied, and broken off above their entrance into the water (about 15 cm exposed). The rods were further broken into
