V O L U M E 28, NO. 1, J A N U A R Y 1 9 5 6 phases. If the hydrocarbon is nearly all in the adsorbed phase, very large volumes of fluorochemical are required t o elute it. If, on the other hand, the hydrocarbon is nearly all in the liquid phase little interchange takes place and the hydrocarbon passes from the column without being significantly separated. Satisfactory results are obtained if the solubility of the paraffinic hydrocarbon in the fluorochemical is of the same order of magnitilde as that of the cycloparaffin hydrocarbon in the added component in the adsorbed phase. Temperatures of complete miscihility, which are useful in comparing solubilities of hydrorarbons in the fluorochemic,tlq and glycol ethers discussed here, AI r given by Nair ( 1 ) . While only paraffins and cvcloyaraffine have been included in the experiments reported here, there is little doubt that many other types of h\-drocaibons may be separated by this method. Other separations include those of aromatics and olefins from r x h other and their separation from paraffins and cycloparaffins. Since, however, aromaticq and oldins can be separated readily b) other methods including regular adsorption, the most iniportant use of the new method appears t o be for the separation of paraffins, particularly branched paraffins, from cycloparaffins. FI om the known solubilities of monocycloparaffins and dicycloparaffins in the usual polar solvents, and their estimated eoluhilities in fluorochemicals, it is predicted that another important
61 application \Till be for the separation of monocycloparaffins from dicyclopnraffins. ACKNOWLEDGRIENT
The authors gratefully acknowledge the awqtance of Michael D. Domenick in connection with some of the experiments reported here. LITERATURE CITED
(1) I I a i r , B. J., ASAL.CHEM.28, 52 (1956). and Rossini. F. D., I n d . Eng. ( 2 ) RIair, B. J., Westhaver, J. W,, Chem. 42, 1279 (1950). (3) lIair, B. J.. and White, J. D., J . Research T a t ? . Bur. Standards 15, 51 (1935). (4) Martin, A. J. P., and Synge, R. L. ll.,Bz’ochem. J . (London) 35, 135s (1941). (5) Rossini, F. D., lIair, B. J., and Streiff, -1.J., “Hydrocarbons from Petroleum,” Reinhold. X e w York. 1953. R E C E I V Efor D review June 10, 19.55. Accepted September 27, 1955. Division of Petroleum Chemistry, 128th Meeting, ACS, Minneapolis. Minn., September 1955. Investigation performed as part of work of American Petroleum Institute Research Project 6 in the Petroleum Research Laboratory a t Carnegie Institute of Technology. .\lost of the material is taken from a dissertation submitted t o Carnegie Institute of Technology i n partial fulfillment of requirements for degree of doctor of philosophy. M. J. Montjar was a graduate research fellow on the American Petroleum Institute Research Project A.
Determination of Cobalt by Anodic Electrodeposition Utilization of Isotope Dilution Method DARNELL SALYER and THOMAS R. SWEET McPherson Chemical Laboratory, The O h i o State University, Columbus 10, O h i o
Cobalt may be satisfactorily determined by an anodic electrodeposition-isotope dilution method. Since the isotope dilution technique requires that only a weighable portion of the pure oxide be deposited on the anode, the time of electrolj-sis may be short, and adherence difficulties that are encountered with cobalt oxide deposits are eliminated as a source of error. The w-eighing form of the deposit is cobaltic oxide trihydrate.
A
LMOST since the brginning of electroanalytical v, ork it has heen k n o ~ nthat cohalt and nickel may be deposited anodically as oxides ( 2 , 1 4 ) Several investigators have indicated the possibility of utilizing anodic behavior in the separation ( 1 , 2 1 ) and the determination ( 9 , 11) of cobalt. From the standpoint of their routine use in analysis. previous analytical methods involving the anodic deposition of cobalt as an oxide have not been satisfactory. Quantitative separations of cobalt from nickel mere obtained only by going to such extremes as using 10 platinum anodes interchangeably over a 20-hour period ( I ) , or by resorting t o double deposition of the oxide a t elevated temperatures and using separated electrode compartments ( I 1 ). I n addition there is a definite tendency for the deposits t o become nonadherent as electrolyeiq proceeds. I n the method given in the present paper, which involves both anodic electrodeposition and the isotope dilution technique ( I O ) , the duration of electrolysis can be decreased considerably. I n addition, nonadherence of the deposit is eliminated as a source of error, because the very thin deposits t h a t are used adhere well t o the electrode, and, even though some of the deposit is lost from the anode before the weighing and counting operations, no error is introduced.
Cobalt-60, a beta-gamma emitter having a half life of 5.3 years, served as the radioactive tracer for the isotope dilution. Highly pure radioactive cobalt can be obtained from Oak Ridge at a low cost. AIETHOD
-4series of solutions containing various known weights of pur e cobalt as sulfate was prepared. A fixed volume of an active cobalt solution was added to each, the solution was buffered, and deposits viere plated on sandblasted platinum disk anodes, using a modified tower electrolysis cell (10). Each deposit was dried, neighed and counted, and its specific activity, S.d.,calculated in counts per minute per milligram. If TV) the weight of inactive cobalt taken, is plotted against l/S.A.>a linear standard ciirve is obtained which obeys the equation
where A is the total activity of the volume of active cobalt used for the isotope dilution, and w is the weight of cohalt in the added active cobalt solution. K h e n cobalt occurs in solution with other metal ions, many of them precipitate as hydroxides, hydrous oxides, or sulfates as the solution is being buffered to p H 7.8 for the anodic deposition. If the amount of the other metal is not too large, this precipitate can simply be filtered off and cobalt deposited as usual from the filtrate. The quantity of metal or metals to be separated in this manner is limited only by the fact that sufficient cobalt must remain unadsorbed in the solution t o form a weighable deposit on the anode during subsequent electrodeposition. If, on the other hand, substances are present which are not precipitated below a p H of 7.5 to 8.0, they may or may not interfere with deposition. Those expected to interfere include: cations
62
ANALYTICAL CHEMISTRY
known t o deposit anodically-e.g., manganese, nickel, lead, and silver-substances strongly adsorbed on hydrous oxides-e.g., arsenious acid (15)-and reducing agents such as organic anions ( 4 ) and acids (8, 9 ) , halides, hydrogen peroxide, etc. APPARATUS
MODIFIED TOWER ELECTROLYSIS CELLS,50 ml. ( I O ) PLATIKUM DISK ANODES, inch in diameter x 0.005 inch
thick, with one surface uniformly sandblasted SAMPLECHANGER, Tracerlab SC-9D shielded, nianiial SCALER,Potter predetermined decade scaler, Model 341 GEIGERTUBE, Tracerlab TGC-2 ELECTROANALYZER, Eberbach, rotating electrode LIICROBALbNCE DRYINGOVEN,adjustable t o 40' C. ALUMISUM ABSORBER, 70 mg. per sq. cm. PH METER,Beckman Model H2 CLINICALCENTRIFUGE,50-ml. borosilicate glass centrifuge tubes REAGENTS
Unless otherwise stated all chemicals were of the analyzed reagent type. Double distilled water was used in preparing all solutions. Radioactive Cobalt Solution. Cobalt-60 was purchased from S. Atomic Energy Commission, Isotopes Division, Oak the tT. Ridge, Tenn. ( O R S L ) . Approximately 1 me. of this solution with 200 mg. of carrier was diluted to 2 liters. Standard Cobalt Solution. I n the minimum amount of concentrated sulfuric acid 1.000 gram of spectrographically pure cobalt sponge was dissolved, and then diluted t o 1 liter. Rlatthey
Table I .
Preparation of Standard Curve
S A . of Oxide Deposit, Counts Per Min./Mg. 2755 2655 2042 1973
w t . of
co,
Mg.
5 000
7 500 10 00
1558 1542 1552 1226
12.50
2705
1:s '4., Mg./Counts 1 Min. 3 697 X 10-6
2001
4 q08 X 10-4
1551
6 448
x
10-4
1220
8.196
x
10-1
x
10-4
AV.
1.5 00 io49 790 812
20 00
801
12 5
__
-
Table 11. Interference Study l/S.A.
Initial Composition of S o h .
A.
Noninterfering substances IO. 00 mg. of Co 1 0 . 0 0 mg. of Co 10.00 mg. of Co 1 0 . 0 mg. of Fe 10.00 mg. of Co 1 0 . 0 mg. of Cu 1 0 . 0 mg. of Fe 10.0 mg. Cu 1 0 . 0 0 mg. of Co 10.00mg. of Co 1 0 . 0 mg. of Cr as CrrOi-1 0 . 0 mg. of Zn 10.00 mg. of Co 10.00 mg. of Co 1 0 . 0 mg. of Bi 10.00mg. of Co 10.0 mg. of Ag 10.00 mg. of Co 10.0 mg. of AI 10.00 mg. of Co 10.0 mg. each of Ba, Ca. Sr, M g 10.00mg. of Co 10 0 mg. of Pb 10.00 mg. of Co 10.0mg. of Cd
++ ++ ++ ++ + ++
El.
6.375 X 6,520 X 6.470 X 6.520 X 6.635 x 6.560 X 6.580 X
+
Interfering substances 10.00 mg. of Co 10.0mg. 10.00mg. of Co 1 0 . 0 mg. 10.00 mg. of Co 1 0 . 0 mg. 1 0 . 0 0 mg. of Co 10.0 mg. 1 0 . 0 0 mg. of Co 10.0mg. 10.00 mg. of Co 10.0 mg.
++ ++ ++
6 940 x 10-4 7 190 X 10-4 6 4 , ~ x 10-4 9.970 X lo-:
of Xi of Nia of M n of As of H g (acetate) of H g (oxide)
7 455
9 40
C . Separation of Co and hfn by cobaltinitrite pptn. 10 00 mg. Co 1 0 . 0 mg. of M n (1 pptn.) 10 00 mg. Co 10.0 mg. of M n (2 pptns.)
++
_
_
_
_
_
10X 10.4
6 490 X 10-4
+
6 700 X 10-4
At p H of 5 and at 90° C., according t o Torrance _
x
6 589 X 10-4
D . Separation of Co and H g by use of Cu 10.00 mg. Co 10 0 mg. of H g (oxide) .
10-4
10-4 10-4 10-4 10-1 10-4 10-4 6 . 4 5 0 x 10-4 6 . 5 8 0 X 10-4 6.455 X 10-6 6.720 X 10-4 6 550 x 10-1 6 , 5 0 5 x 10-4
_
(11). _
~
-
~
cobalt sponge of Johnson, Matthey and Co., Ltd., London, was used. Boric Acid-Potassium Sulfate Buffering Solution. One liter of a solution that was 0.1OM in boric acid and 0.05.V in potassium sulfate was prepared. T o adjust the p H to the desired value 0.1 and 0.5%' sodium hydroxide solutions were used. Solutions of metal ions for interference and separation studies were prepared from the nitrates, sulfates, or oxides. Each solution contained 2 mg. per ml. of metal. Oxides were initially dissolved in the minimum amount of sulfuric acid and then diluted. PROCEDURE
Preparation of Standard Curve. Pipet 10-ml. portions of the radioactive cobalt solution into various volumes of the standard inactive cobalt solution. Mix thoroughly. Add 20 ml. of boric acid-potassium sulfate solution, and add 0.1M sodiuni hydroxide dropwise with stirring until the p H is 7.6 t o 7.8, as determined with a p H meter. Dilute t o 50 ml. with double distilled water. If a small precipitate of basic salt forms, filter the solution and discard the precipitate. Clean the sandblasted platinum disk anode (using acid iodide or oxalate to remove oxide deposits), dry for 15 minutes a t 40" C., and weigh t o the nearest 0.002 mg. With the disk in place in the cell, introduce the activt. buffered solution. Electroplate a t I .5 t o 1.8 volts for 40 minutes. Remove the still active solution through the cell's glass side arm, rinse out the cell with double distilled water, and remove the disk which now contains the oxide deposit. Wash the deposit with double distilled water and remove adhering water droplets with a piece of filter paper. Allow the disk t o dry in air until no water is visible, and place in a 40" C. oven for 2 t o 2.5 hour?. Weigh as before. Determine the activity of the deposit by placing the disk in a planchet in position in the sample changer with ii 70-mg. per sq. em. aluminum absorber in place. Make dead time, background, and efficiency corrections on the observed activity. and calculate the specific activity. Data for a standard curve are given in Table I. At TF valuer: less than 6 mg., the small amount of deposit formed in 40-minute deposition leads t o some uncertainty. Similarly, above about 15 mg., the activity of deposits obtained in the usual electrodeposition time is low; this also leads to uncertainty or incoilvenience. Consequently, the portion of the curve between 5 anti 15 mg. is best for most purposes and should be used if the approximate per cent cobalt in an unknown is known. Counting only the gamma radiation from deposits removes the possibility of errors due to self-absorption of the beta radiation. The 70-mg. per sq. cm. aluminum absorber is suitable t o screeii out the beta rays. .4 standard gamma source t h a t was counted daily and used in maintaining constant counting efficiency consisted of a sample of active metallic cobalt deposited on a .. disk electrode like those useti for the oxide deposits. 7' Error w,M g . Separations and Interference Study. T o a mixture of 10.00 -2.8 9,720 -0.2 9.080 ml. of standard cobalt solution, 9.9iO -0.3 10 ml. of radioactive cobalt, -0.2 9.980 and 20 ml. of buffer, add 10.0 +I.5 10.15 40.3 10.03 mg. of the metal to be tested. f0.7 10 07 Add 0.5M sodium hydroxide to -1.4 0 860 a p H sufficient to precipitate +o.i 10.07 -1.2 9.880 the added metal as hydroxide f3.0 10.30 (or to 7.6 t o 7.8 if no precipi+0.2 10.02 tate forms). Filter off the prei n 03 +o 3 cipitate on a coarQefilter paper Adjust to p H 7.6 t o 7.8, filter-6 R 10 6R ing again if necessary. Proceed c10.0 11 OR with the electrodeposition in > 1000 ca. 109 the described manner. Find 15.91 59 1.5 6 11 S f i the specific activity as before, 48 14 8 and read the weight of cobalt from the standard curve. During the buffering process, iron is 4 1 0 10.10 9 61 -0.6 precipitated as ferric hydroside a t a pH of 3.5. Cupric, zinc, aluminum, and cadmium ione t2 8 10.28 are removed by filtration after precipitation a t pH 6.5. Lead, alkaline earths, and some ~
V O L U M E 28, N O . 1, J A N U A R Y 1 9 5 6 d v e r ions precipitate as sulfntes and are filtered off.
63
Table 111. Comparison of Plating Methods" Behavior
Description
Cath-
Table 11, A, s h o w thiit during ode Of Plating Deposit Deposit Soln Vked Voltage wveral metals, when initially No change None Very thin, red-brown 0 05 g . KzCriOi 2 3 present in an amount equal to S o h . changed from deep Slight Sonadherent, dark 40 ml. 3 0 7 4 a O H 0 5 that of the cobalt, do not inblue t o , gray: black brown to black suspension terfere Tyith the determination. 0 5 g. I(K4HiOs 2.0 Good and uniform, jet Soln. turned yellow- S o n e TILble 11, 13, shows results for black green t o t a n 5 ml. 30% NaOH ions from which deposition Sonuniform, brown N o change decomposed Yes 30 ml. 30% NaOH 1.5 on standing 1 ml. glycerol .*(:paration may not be made N o change Yes 0 . 5 ml. HCzHsO: u p t o 5 T o deposit :%rid which must be separated dark NaC~H30zt o p H 5 from cobalt prior to deposition Sone Black forms slowly N o change 0 . 0 5 g. K?Cr?OI 2.3 0 . 3 g. K2SOr t)y some method other than Pink precipitate forms S o n e 20 ml. 0 . 1 s S a O I I u p t o 2 lionuniform hydroxide or sulfate precipitain s o h . 20 i d . 0,l.U KaHxPO, tion. It was already known Uniform. brown-black No change 3 i n l . 0.1.\- S a O H 1-2 None plate fairly adherent 25 ml. 0 , l J f IIaB08 in that nickel codeposits to a n.03.u IGSO~: P H = small extent (11). , Cations codepositing were XaHCOi t o neutral. 0.5 U p t o 2 Good black deposit, Soln. turned light green, None very adherent no ppt. g. excess manganese and nickel, mangita Following factors affecting type and quality of deposits were held constant throughout s t u d y ; temperature, being far the worse' 2 j e C.; deposition time, 20 minutes: amount of cobalt used, 10 mg. as sulfate; volume of plating bath was 50 ml.: 'The cobaltinitrite method of' smooth platinum anodes were used. precipitation of cobalt can he used to effect a separwTable IV. Reproducible Deposits tion of these metals prior to (Deposited from borate buffer a t p H 7.6, 1.7 volts, and 1 t o 2 ma. per sq. om modic deposition. Arsenite or arsenious acid strongly inhibits dried 2.5 hours a t 40' C.) deposition probably because of the fact that it is adsorbed S.A. %co = wt. of Activity c.P.M.; S . A . Deposit on the surface of the deposit that does form. Because ferric iron, Deposit, Net Mg. S . A . Metal I\.Ig C.P.M. precipitated a t p H 3 from a hot solution, strongly adsorbs the 3 294 5438 1651 53.9 itrsenic, the arsenic can be removed from cobalt-arsenic mixtures 3.283 5426 1652 53.9 2.690 4373 1626 53.1 leaving most of the cobalt in an arsenic-free filtrate. Mercury 2,301 3867 1680 54.8 ran be removed by adding finely divided copper to the acidic 2.020 3316 1642 53.6 2 645 4331 1638 53.5 mixture, shaking intermittently for 15 minutes. and decanting 1 841 299 1 1625 53.0 (Table 11, D). S . A . of metal = 3064 c.p.m./mg. Av. 7c CO = 5 3 . 6 8 For the determinations of Table 11, A, C, and D, the 9tandard deviation of a single measurement is 1.48% .i)
Thus, the reproducibility is satisfactory and well within t h e limitations of the isotope dilution method.
REPRODUCIBLE DEPOSITION
Hefore undertaking the development of the analytical method, a study \vas made t o determine the conditions under which a reproducible weighing form of the oxide could be obtained. The results are shown in Table 111. Plates obtained in experiments three, eight, and nine appeared to be best. Deposition from dolutions buffered with borate was chosen for subsequent experiments. Two means of quantitatively studying reproducibility were used: the specific activity of anode deposits obtained under the Pame set of conditions, and the per cent cobalt in the deposits. The per cent cobalt was calculated by the following equation:
yo C o
=
specific activity of oxide specific activity of metal
x
100
The same active cohalt was used in the preparation of both metal and oxide deposits on the same type of disk electrodes. Identical gpometry conditions were used in counting. The following procedures for drying the oxides were tested: a t room temperature in a desiccator over anhydrous calcium sulfate; at room temperature in a vacuum desiccator over barium (ixide; a t several temperatures from 30" t o 175" C. in an oven: arid finally, ignition a t 300", 500", and 1000" C. Drying for 2.5 hours at 40' C. gave the best reproducibility. Ignition to (dmlt,o-cobalticoxide, as suggested by Smith ( 9 ) ,led to deposits which were very difficult t o strip. The results of activity measurements for a series of deposits prepared from solutions buffered with boric acid-borate and dried a t 40" C. are shown in Table IV. The absolute standard deviation for these experimental value$ of the per cent cobalt i p 0 . 4 1 ~ o .
COiMPOSITION OF DRIED DEPOSITS
The experimental value of 53.68% cobalt (Table IV) that was ohtained from activity measurements indicates that the composition of the dried deposits is cobaltic oxide trihydrate (theoretical % cobalt = 53.60). This is in agreement with other findings ( 1 , 6, 7 ) . ACKNOWLEDGMENT
The authors gratefully acknowledge aid to Darnel1 Salyer in the form of a graduate fellowship from the Cincinnati Chemical Works. LITERATURE CITED
(1) Coehn. A., and Glaser, M., Z . anorg. Chem. 33, 9 (1903). ( 2 ) F'iucher, S . W.. Kastn. Arch. 16, 219 (1820); Pharrn. Cenlr. 109-10 (1830). (3) Grube. G.. and Feucht, O., 2 . Elektrochem. 25, 568 (1922). (4) Haissinsky, ll.,J . c h i m phys. 29, 469 (1932). ( 5 ) Huttig, G. F., and Kassler. R.. Z . anorg. Cliein. 184, 254 (1929) (6) Muller, E., and Spitzer, F., Ibid.,5 0 , 326-7 (1906). (7) Root, J . E., J . Phys. Chem. 9, 1 (1905). (8) Scholl, G . P., J . A n i . Chem. Sac. 25, 1049 (1903). (9) Smith, E. F.. Trans. Am. Electrochem. Soc. 27, 30--4 (1915); "Electroanalysis." 6th ed.. pp. 137, 146, Blakiston, P h i l a delphia, 1918. ( I O ) Theurer, K., and Sweet, T.R.. . ~ N A L . CHEM.2 5 , 119 (1953). (1 I ) Torrance, S.,Analyst 64, 109 (1939). (12) Tubandt, C., 2. anorg. Chem. 45, 368 (1905). (13) Weiser, H. B., "Inorganic Colloid Chemistry," vol. 2, p . 36. Wiley, Kew York, 1936. (14) Wernicke, W., Pogg. Ann. 141, 109 (1870). R K E I V E Dfor review June 29. 1955.
Accepted October 13. 1955,