V O L U M E 25, NO. 4, A P R I L 1 9 5 3 within the molecule. Since the n-aliphatic disulfides that would be found in the motor gasoline boiling range have essentially the same diffusion current constant value, an average K value or the K value for n-butyl disulfide could be used to determine disulfide sulfur. If the gasoline contained tertiary disulfides, trisulfides, and polysulfides, this technique could not he used. Diphenyl disulfide is reduced a t a more positive voltage than aliphatic disulfides and these tn-o types of sulfur compounds do not interfere with each other in the polarographic method of analysis. The polarograph has been found a valuable analytical tool in studying the chemical reactions that take place in gasoline
561
sweetening processes, and also for the examination of crude oils for naturally occurring sulfur compounds. LITERATURE CITED
Brdicka, R., Collection Czech. Chern. Conamuns., 5 , 238 ( 1 9 3 3 ) . ( 2 ) Gerber, M.I., J . A n a l . Chern. (C.S.S.R.), 5 , 262 (1950). (3) Hall, M. E., ANAL.CHEM.,22, 1137 (1950). (4) Kolthoff, I. M.,and Barnum, C . , J . 4rn. Chon. Soc., 63. 520 (1)
(1941). (5) Proske, G., Angew. Cltem., 59, 121 (1947). RECEIVED for review June 17, 1952. Accepted January 21, 1953. Presented before t h e Division of Refining a t the meeting of the American Petroleum Institute, San Francisco, Calif., May 1952.
Polarography of Copper in Ethylenediaminetetraacetic Acid Solutions ROBERT L. PECSOK Crniversity of California, Los Angeles, Calif. The reniarkabk sequestering ability of ethylenedianiinetetraacetic acid (Versene) for metal ions has been of great value in many fields. Its use as a supporting electrolyte for polarographic procedures has recently been investigated. Except for the ironVersene system, systematic polarographic data are not yet available. Both the reduction of the copperVersenate complex and the oxidation of copper from a dropping amalgam electrode into a Versene electrolyte are nearly reversible and yield well defined polarographic waves. The diffusion current is proportional to concentration of copper from 10-5 to 10’2 M. For 0.25 M \-ersene, pH 6 to 9, the diffusion current constant is 2.85 and the diffusion coefficient
THE
remarkable ability of ethylenediaminetetraacetic acid (the sodium salts of ethylenediaminetetraacetic acid are referred to here as Versene, with the corresponding Versenate ions) to form stable complexes with nearly every metal has been known for several years. I t has found increasing use in many applications \\here a sequestering agent is desirable (1). The large values for the stability constants ( 1 2 ) suggest its use as a supporting electrolyte for polarographic determinations. Only a few papers have appeared concerning the polarographic behavior of these complexes; and, except for the iron-Versene system ( 5 ) ,complete systematic data have not been published. Furness, Crawhaw, and Davies (4)utilized the polarographic reduction of the copper-Versene complex to determine Versene. They reported a reversible wave with a half-wave potential of -0.444 volt us. the saturated calomel electrode in a supporting electrolyte of ammonium nitrate a t pH 6.5 to 7.0. An earlier paper of the present author ( I O ) summarized the results of a survey of a number of common metal-Versene complexes. This paper presents the results of a polarographic investigation of the copper-Versene system. A number of variables, such as temperature, concentration, pH, and drop time, have been tested for their effect on the polarographic wave. The system has been tested for reversibility and the stability constant of the complex has been redetermined. EXPERIMENTAL
Polarograms were recorded with a calibrated, undamped Sargent iModel X X I polarograph. The modified H-cell, with external Eaturated calomel electrode, was placed in a water bath
is 5.47 X 10-6 sq. cm. per second. Equations for the
electrode reactions are proposed and confirmed from the dependence of half-wave potential on pH. Systematic data are presented concerning the polarographic behavior of the copper-Versenate system. The effect of variables such as temperature, drop time, pressure of mercury, concentrations of copper and Versene, and the pH on the half-wave potential and diffusion current have been determined and interpreted. The stability constant of the complex in agreement has been calculated to be with Schwarzenbach’svalue. The shift of the interfering iron wave in mixed Versene-fluoride supporting electrolytes is discussed.
maintained a t 25.0’ i 0.1’ C. The capillary had a rate of flow of 1.811 mg. per second and a drop time of 4.65 seconds at an applied e.m.f. of -0.5 volt; ( m 2 / 3 t 1 ’=6 1.920 mg.2/3sec.-l 2 ) . Solutions were deaerated Rith purified nitrogen. 811 potentials were measured and are reported us. the saturated calomel electrode unless specifically noted otherwise. Half-8 ave potentials are accurate to within & 2 mv. ,411 diffusion currents were properly corrected for residual current. S o maximum suppressors were necessary and none mas used. A Beckman Model G pH meter was used to measure the pH of solutions; the same instrument v a s used as a potentiometer for some of the experiments. Stock solutions of copper were prepared from either copper sulfate pentahydrate or copper \\ire, and mere standardized iodometrically. Stock solutions of T’ersene were prepared from the disodium salt (Eastman Kodak Co.) after purification by recrystallization through the insoluble acid. Other chemicals used were of reagent grade and were not further purified. A dilute copper amalgam n-as prepared by electrolysis of a copper sulfate solution between a platinum anode and a mercury cathode. Known amounts of copper and mercury mere used to yield a 0.001% amalgam. -4second capillary was used for the dropping amalgam electrode. RESULTS AND DISCUSSION
The copper(I1)-Versenate complex is found to give only a single m x e , reduction proceeding directly t o the metal. I n no case is there any evidence of steprtise reduction through the + l state, which indicates that the f l complex (if any) is unstable with respect to disproportionation to the +2 complex and the metal. Reversibility. The equation of the wave was determined at+ suming the usual form ( 7 ) :
562
ANALYTICAL CHEMISTRY Table I.
Effect of Temperature
[0.4m.Vf Cn(I1) in 0.25 'it. T'ersene, p H 5.301
A O.4-niM solution of copper in 0.25 M Versene, pH 5.30, Fas polarographed and a graph was made of the wave according to Equation 1. The points yield a straight line of slope -0.0384 instead of the expected slope of -0.0295 for a two-electron reduction. While this might be interpreted as evidence of an irreversible reduction, thus invalidating any thermodynamic significance of bhe data, this slope is relatively constant over wide variations of conditions. Furthermore, as seen below, many of the other criteria of reversibility are fulfilled by this system. In most cases, reversibility was checked by noting the diffvr znce Ea14 - E l / , ( 8 ) .
Temp.,
C.
Ei/ 2, Volt
Zd, pa.
The proper corrections for back pressure of mercurj- have been made. The constancy of the ratio of the diffusion current to the square root of the effective height of the column is proof that this electrode reaction is controlled entirely by diffusion. Proportionality of Current to Concentration. The diffusion current was memured for sixteen copper concentrations from 1 X 10-6 1 ' 1 to 0.0145 V in 0 25 ill Versene, pH 6.1. The diffusion current is porportional to concentration within the experimental error, ca. 1 to 2%. Under these conditions the diffusion current constant, I , equals 2.85 0.05.
*
Table 11.
f
I
I
I
10 ga.
I
I!
-,( ,I\
r
I
0
I -0.2
E,
Effect of Pressure of Mercury
[0.2 n1.M Cri(I1) in 0.26 .M Versene, pH 4.501 Height of Column 40 cm. 62 cm. 80 cm. 38.5 60.5 78.6 2.746 1.148 1.811 7.32 4.65 3.54 0.942 1.184 1.360 0.162 0.152 0.153
volts
ii
The same experiments were repeated a t pH 8.70, where B value of I , equal to 2.81 0.03, is obtained. It is apparent that the diffusion current is not a function of pH in this p H region. The diffusion coefficient, D, of the copper-Versenate ion is calculated to be 5.47 X 10-6 per second.
*
I
I
I
-0.4
vs. S.C.E.
Figure 1. Polarograms A . 0.2 m.W copper in 0.25 M Versene with dropping mercury electrode B . 0.2 m M copper in 0.25 M Versene with a copper amalgam (0.001%) electrode C. 0.25 M Versene with amalgam electrode
Typical polarograms indicating the degree of irreversibility are shown in Figure 1. In the composite wave (curve B), obtained with a dropping amalgam electrode in a Versene solution, there is no discernible break between the cathodic and anodic portions, This smooth curve is evidence of reversibility. The half-xave potentials of the three waves are: A , -0.320 volt; B , -0.300 volt; and C, -0.290 volt vs. the saturated calomel electrode. Effect of Temperature. In the region 15" to 35' C., the temperature dependence of both the diffusion current and the halfwave potential is linear, as shown in Table I. The temperature coefficient of the half-wave potential, -0.5 mv. per degree, is of the proper sign and magnitude for that of a reversible electrode process ( 9 ) . The temperature coefficient of the diffusion current, 1.56% per degree, is precisely that to be associated with a diffusionicontrolled electrode process ( 6 ) . Effect of Drop Time. The data in Table I1 were obtained by recording polarograms of a 0.2-mM copper solution in 0.25 M Versene, pH 4.50, a t various heights of the mercury column.
020
3
I
I
I
I
I
I
I
I
I
4
5
6
7
B
9
10
,I
I2
13
PH
Figure 2.
Half-Wave Potential of 0.4 m M Copper in 0.25 F Total Versene at Various pH Values
Effect of pH. The pK values of ethylenediaminetetraacetic acid are 2.00, 2.67, 6.16, and 10.26 (11); thus, there are five different Versene species (the un-ionized acid, the tri-, di-, and monohydrogen Versenate, and the Versenate ion), each present as a major constituent consecutively over the pH range 0 to 14 (see 10, Figure 1, for a graph of the relative concentrations of
563
V O L U M E 2 5 , NO. 4, A P R I L 1 9 5 3 each species as a function of pH). Since the predominant Versene species in the solution, as well as the formula of the copper complex, are a function of pH, the half-wave potential is likewise expected to be a function of pH. The dependence should be related to the number of hydrogen or hydroxyl ions in the electrode reactions. This dependence was determined by polarographing a solution of 0.4-m.lf copper in 0.25 F total Versene a t various pH values The lower limit of pH is restricted by the solubility of ethylenedianiinetetraacetic acid, ea. pH 3 a t this concentration. As the pH is increased above 12, the reduction becomes so irreversible that measurements are difficult and meaningless. The results are plotted in Figure 2, where the pK's of ethylenediaminetetraacetic acid have been marked across the graph The graph is divided into several straight-line portions, corresponding roughly t o the regions of pH values between successive p K values of ethylenediaminetetraacetic acid. At each pK value, the solution contains equal concentrations of two Versene species, but between two pK values the solution contains only one predominant species. Therefore, the electrode reactions can be formulated as follows: pH
< 6,
+ HzY-
CuY"
+ 2H+ + 2e
+ H + + 2e = Cu + HY'
= Cu
pH 6
- 9,
CuY'
pH 9
-
+ 2e = Cu + Y= CuYOH' + 2e = Cu + Y= + OH-
pH
(3)
11.3, CuY-
> 11.3,
(2)
(4)
to 250 mJT. Three series of polarograms were taken a t pH 4.50, 6.20, and 8.80. The pH varied slightly with additions of Versene and the half-wave potentials were corrected to the above pH values by use of the previously determined dependence on pH. Plots of half-wave potential us. log concentration of Versene yield straight lines with slopes of 12, 22.5, and 15 mv. per tenfold change in Versene concentration. These results do not confirm the proposed electrode reactions, hut rather tend to indicate less than one Versene molecule per copper in the complex. The latter explanation is untenable, leaving the slight irreversibility of the polarographic waves as the only plausible explanation. Determinationof Stability Constant. The standard potentials of the several proposed electrode reactions are related to each other through the ionization constants of ethylenediaminetetraacetic acid and water, and are related to the stability constant of the complex by use of the electrode reaction: Cu++
= Cu
(6)
For example, subtracting Reaction 2 from Reaction 6, one obtains:
+ HSY-
CU++
= CUY-
+ 2H+
(7)
Subtracting the third and fourth ionizations of ethylenediaminetetraacetic acid from 7 , one obtains:
(5) Cu++
where HIY represents ethylenediaminetetraacetic acid. From the Nernst equation, the standard reduction potentials and the related half-wave potentials for the reactions should depend upon pH as listed in Table 111, where the calculated and observed values are compared.
+ 2e
+ Y= = CuY-
(8)
The equilibrium constant of Reaction 8 is the stability constant of the complex. Combining the standard e.m.f.'s and equilibrium constants in a similar manner, one obtains: log K
+ pK3 + PK,
~=~ 2 ~E," 0.059
Table 111. Effect of pH [0.4 m M Cu(1I) in 0.25 F total Versene]
Reaction No."
Region of P H
AEi/z/ApH
Calcd.. mv./pH (2) (3) (4)
(5) a
I1 3
Obsd..
my. / 11 H
- 59
- 0-1
0 -29.5
- 34
-29.5
-26
-4
See text.
The agreement confirms the proposed reactions. I t is obvious that for analytical applications, the pH of the supporting electrolvte must be recorded and considered. Concentration of Versene. For the reduction of a metal complex ion t o the metal, the half-wave potential must also be a function of the concentration of the complexing species. From a n inspection of the proposed electrode reactions and consideration of the Nernst equation, one finds that for all regions of pH this dependence should be -29.5 mv. per tenfold increase in the concentration of Versene. This is true even for low pH values where the concentration of the complexing species, the completely ionized Versenate ion, is low relative to copper, but nevertheless is maintained a t a constant value a t the electrode surface through the rapid ionization of other Versene species present a t relatively high concentration. A supporting electrolyte of 0.1 iM sodium nitrate was used for these experiments with a constant concentration of copper of 0 . 2 mW, and increasing concentrations of Versene from 1.25
where K c ~ Yis- the stability constant, Ka and K , are the third and fourth ionization constants of ethplcnedianiinetetraacetic acid, and ET0is the standard e.m.f. of reaction 7 . Since both Reactions 2 and 6 involve reductions to a dilute copper amalgam a t the dropping mercury elect rod^, terms involving the copper metal activity and the e.m.f. of the reference electrode cancel out in the subtraction process, and the use of half-wave potentials for obtaining Ei" is justified. The latter is obtained by extrapolating the half -wave potential for reaction 2 to 1.V Versene and pH 0, and subtracting the extrapolated value from the half-wave potential of Reaction 6. When these calculations are performed, one obtains:
log KruY- =
2 X 0.070 -
o;y9- f 6.16
+ 10.26 = 18.8
(10)
This compares favorably with the value 18.38 (in 0.1 X potassium chloride, 20" C.) reported by Schwarzenbach ( I d ) , obtained by pH measurements, and involving the use of a second complex of intermediate stability. As an independent check, a potentiometric titration mas attempted. Versene solutions were titrated with a standard copper solution a t a copper wire indicator electrode. The titration reaction is identical to reaction 7 above, while the electrode reaction determining the e.m.f. is similar to reaction 2, except that reduction is to the pure metal rather than to a dilute amalgam. After repeated efforts a t various p H values and concen-
564
ANALYTICAL CHEMISTRY
trations of reacting species, and different pretreatment of the copper electrode, it was impossible to prevent a drift in the measured e.m.f. It is suspected that the copper metal which in the presence of Versene reduces hydrogen ion as the drift was substantially repressed by vigorous stirring. Under these conditions it is possible to compute standard e.m.f.’s for Reactions 7 and 8, and the stability constant. Values obtained are in general agreement with previous values but exhibit a definite trend over the course of the titration and are not regarded as significant. Determination of Copper in the Presence of Iron. Since the reduction of +3 iron normally precedes that of copper, it is desirable to have available a simple method of preventing the prior reduction of iron in determinations involving unfavorable ratios of iron to copper. Faucherre and Souchay (S) claimed that this can be accomplished by the addition of fluoride. This was verified in this study, provided that the high concentration of flouride stated by the earlier authors, 14 M , is used. At lower concentrations of fluoride, ca. 1 M, an interesting situation prevails. Consider the following stability constants ( 2 , 6): Fe(II1) Y - , 1026; Fe(I1) Y’, Fe(II1) F2+, and Fe(I1) F+, < 30. Of the Fe(II1) complexes, the Versenate is the more stable. Yet the ferric-Versenate complex iF far more easily reduced than the ferric-fluoride complex (the half-wave potential is about 1 volt less negative) because of the existence of a stable ferrous-Versenate complex and a very weak ferrous-fluoride com-
plex. The stability constants quoted indicate this qualitative behavior. Therefore] in order to form the ferric-fluoride complex and prevent the prior reduction of iron in mixtures of copper. iron, Versene, and fluoride, it is necessary to use a fluoride concentration of the order of 50 times that of the Versene concentration. LITERATURE CITED (1) Alrose Chemical Co., Providence, R. I., Technical Bulletins, May and October, 1951. (2) Dodgen, H. W.,and Rollefson, G K., J . Am. Chem. Soc., 7 1 , 2600 (1949). (3) Faucherre, J., and Souchay, P I Bull. soc. chtm. Fmnce, 1949, 722. (4) Furness, W., Crawshaw, P., and Davies, W. C., Analyst, 74, 629 (1949). (5) Kolthoff, I. M.,andiluerbach, C., J . Am. Chem. Soc., 74, 1462 (1962). (6) Kolthoff, I. M., and Lingane, J. J., “Polarography,” p. 92, Xew York, Interscience Publishers, 1952. (7) Ibid., p. 213. ( 8 ) Ibid., p. 228. (9) Ibid., p. 234. (10) Pecsok, R. L., J . Chem. Educ., 29, 597 (1952). (11) Schwarsenbach. G.. and -4ckermann. H., Helu. Chim. Acta. 30, 1798 (1947). (12) Schwarzenbaoh, G., and Freitag, E., Ibid., 34, 1492 (1951). RECEIVED for review August 4, 1952. Accepted December 29, 1952. Presented before the Division of Analytical Chemistry at the 132nd 3Ieeting of the . 4 v E R I c a m CHEMICAL SOCIETY, Atlantic City, N. J.
Determination of Aluminum in Zirconium Based on Separation by Ion Exchange HARRY FREUND AND FREND JOHN MIRiER Oregon State College, Corz;allis, Ore.
No entirely satisfactory method for the determination of small amounts of aluminum in zirconium has been available, owing to the lack of a suitable separation method. This paper describes an ion exchange separation whereby the zirconium in a solution 0.06 M in hydrochloric acid and 0.8 M in hydrofluoricacid is exchanged on a column of Dowex1, while the aluminum passes through the column. The aluminum may be determined in the eluate by conventional gravimetric or colorimetric procedures. Synthetic samples containing amounts of aluminum varying from 0.001 to 3.8% have been analyzed, As a consequence, a satisfactory method is now available for the separation and determination of small amounts of aluminum in zirconium.
N
0 E S T I R E L Y satisfactory method is knolm for the deter-
mination of small amounts of aluminum in zirconium. Employment of a caustic solution to precipitate the zirconium is not feasible, as small amounts of aluminum will inevitably be coprecipitated with the gelatinous zirconium hydrous oxide. Separations based on the precipitation of zirconium from acid solution with cupferron ( 8 ) ,arsonic acids ( d ) , mandelic acid and its derivatives (S, 6), phthalic acid ( I I ) , and m-nitrobenzoic acid ( I O ) have been suggested. I n each case, although aluminum is not precipitated by the reagent, the precipitation of the major constituent creates other serious problems. The direct precipitation of aluminum is thus far not feasible for the lack of a reagent that will precipitate aluminum but not zirconium. Ion exchange offers a superior technique for the separation of
these metals. Under suitable conditions the negatively changed zirconium fluoride complex is exchanged while the aluminum complex passes through the resin column. The fluoride complexes are obtained when the metal is dissolved in hydrofluoric acid according t o the customary solution procedures. EQUILIBRIUM EXPERIMENTS
Iiraus and Xoore ( 5 ) have reported exchange characteristics for the zirconium fluoride complex. A somewhat more extensive study of the zirconium complex and the hafnium complex and a comparable investigation of the aluminum complex are required to determine suitable conditions for the analytical separation. The partition of the metal fluoride complex between the solution and the resin is determined by equilibrating a weighed quantity of the resin with a solution containing a known total concentration of the metal. The distribution coefficient is defined as the ratio of the concentration of the metal in the resin phase to the concentration of the metal in the solution. At constant weight of resin and volume of solution a modified coefficient may be computed equal to the ratio of the millimoles of metal in the resin to the millimoles of metal in the solution. I t has been shown that the greater the separation between the distribution coefficientsof two ions, the more easily they may be separated (9). Reagents. Dowex-1. The resin used was 200- to 400-mesh. It is available from hlicrochemical Specialties Co., Berkeley, Calif. Zirconium metal. Metal turnings of production-run zirconium were furnished by the Bureau of Mines, Blbany, Ore. Hafnium metal. Metal turnings were furnished by the Bureau of Mines, Albany, Ore. Aluminum metal. Reagent grade metal foil was used.
