LITERATURE (CITED
( 1 ) Back, R. M., Ling, S.-S., ANAL. CHEM.26, 1543-6 (1984). ( 2 ) Baldwin, D. E . , Loeblich, V. M., Lawrence, R . V., Jbzd., pp. 760-2 (1954). ( 3 ) Cassidy, H . C., “Fundamentals of Chromatography,” in “Technique of Organic Chemistry,” Vol. X, p . 283,
Interscience, Xew York-London, 1957. ( 4 ) Corcoran, G. B., ANAL. CHEM.28, 168-71 (1956). ( 5 ) Higuchi, T., Hill, N. C., Corcoran, G . B., Ibid., 24,491-3 (1952). ( 6 ) Isherwood, F. A., Biochem. J . 40, 688-95 (1946). ( 7 ) Marvel, C. J., Rands, R. TI.,J . Am. Chem. SOC.72,2642 (1950). ( 8 ) Ramsey, L. L., Patterson, W. I.,
J . Assoc. O$c. Agr. Chemists 31, 139-50 (1948). ( 9 ) Smith, A. I., ANAL.CHEM.31,1621-24 (1959). (10) Smith, E. D . , Ibid., 32, 1301-4 (1960). ( 1 1 ) Vandenheuvel, F. A., Hayes, E. R., Ibid., 24, 960-5 (1952). RECEIVEDfor review December 16, 1963. Accepted March 16, 1964.
Separation of Scandium from Yttrium, Lanthanum, and the Rare Earths by Cation Exchange Chromatography F. W.
E. STRELOW
and C. J. C. BOTHMA
National Chemical Research laboratory, South African Council for Scientific and lndustrial Research, Pretoria, South Africa
b Scandium is separated from yttrium, lanthanum, cerium, scimarium, erbium, and ytterbium b y cation exchange chromatography. The scandium is eluted from a column of A G 5 0 W - X 8 resin with 2N sulfuric: acid while the rare earths are retciined. The rare earths can then b e eluted with 4 N hydrochloric acid. Results of quantitative separations a r e reported for two different column size!;: 0.377 mg. of scandium could b e separated quantitatively from 4 4 3 . 6 mg. of lanthanum or from 2 8 8 . 4 mg. of ytterbium, and 150.8 mg. of scandium from 0.887 mg. of lanthanum. Flow rates a r e critical, especially .For the smaller columns. The procedure should also separate scandium from the rare earths not named above.
I
N RECENT YEARS, various procedures
applying ion exchange chromatography for the separaiion of scandium from the rare earth elements have been described. Radhakrishna (11) separated scandium from yttrium and lanthanum using a column of Amberlite IR-100H cation exchange resin and a citrate eluent. The wandium, which is eluted first, shows a strong tailing effect. Iya investigated the cation exchange separation of scandium from the rare earths in citrate ( 7 ) and E D T A (8) media, and Vickery (16) employed EDTA and hydrazinl. diacetic acid as eluents. Again strong tailing effects were observed and thfl separations were not quite quantitative. Fritz and Umbreit (3) separated scandium and yttrium using a Dowex 50-X4 cation exchange column. The ions were complexed with E D T A before adsorption and eluted with EDTA solutions of definite p H valueq. Quantitative separations of scar dium from the other rare earths are suggested by ion retention against pH curves given in their paper. Spedding et al. (IS) separated
scandium from the rare earths on cation exchange columns using HEDTA and Cu-EDTA as complesing agents. Minczewski and Dybczynski (10) investigated the anion exchange behavior of the EDTA complexes of scandium and the rare earths and found that separation of scandium from some rare earths was possible but from others, especially from thulium, ytterbium, and lutecium, was improbable. Rezanka et al. (12) separated scandium from lanthanum on a Katex S cation exchange resin using a lactate eluent. James, Powell, and Spedding (9) found that scandium is eluted before the rare earths from Amberlite IR-120 cation exchange resin using EDTA as eluent. KO results of separations are given. Hamagushi et al. (4, 5 ) investigated the separation of scandium from a number of elements using Diaion SK-1 cation exchange resin and HCI-NHa SCN mixtures as eluents. The separation of scandium from dysprosium and europium does not seem to be satisfactory and the heavier rare earths are not separated at all. A separation of scandium from yttrium and the lanthanides using a strongly basic anipn exchange resin and 13N HCl as eluent has been described by Hamagushi, Kuroda, and Shimizu (6) and by Yoshimura, Takashima, and Waki (17’). The value of this separation appears to be somewhat doubtful because the distribution coefficient for scandium is very low, even at 135 HC1. Only small amounts of scandium can be absorbed ( < 3 mg. of Sc203) and an early break-through occurs. Furthermore, practical difficulties are encountered in handling the high HC1 concentration. Faris and Warton ( 2 ) investigated the ion exchange behavior of scandium, yttrium, and the rare earths in nitric acid-methanol mixtures using Dowex I and Amberlite CG-400 resin in the nitrate form. He presented distribution coefficients as a function of nitric acid
and of methanol concentrations, as well as separation factors relative t o gadolinium. The values of the distribution coefficients suggest that it should be possible to separate scandium from the rare earths by this procedure, provided that the exchange reaction rates are not slowed down too much by the presence of the organic reagent. KO data on actual separations are given in the paper. Faris ( I ) also investigated the absorption of scandium fluoride complexes on anion exchange resins. Most of the above procedures make use of organic complexing agents which introduce complications into the determinations or into further separations which may be carried out on the eluates. I n addition, quite a number of the procedures do not seem to provide satisfactorily quantitative separations for accurate analytical work. The introduction of a fairly simple eluent able to separate scandium quantitatively from the rare earths, therefore, seems to have its attractions. .4 s>sternatic survey (15 ) of equilibrium distribution coefficients of cations in sulfuric acid using Bio-Rad AG 50W-X8 resin was undertaken in this laboratory and suggested that it should be possible to separate scandium quantitatively from yttrium and the rare earths by elution with about 2A‘ H2S04. At 2.V acid, the distribution coefficient for scandium is 8.5 and the separation factors are cusc-~&= 32.3, ag,-y = 5.8, crgc-yb = 5.6. EXPERIMENTAL
Reagents and Apparatus. Analytical reagent grade chemicals were used whenever possible. Scandium oxide and rare earth oxides of 99.9% purity were obtained from L. Light and Co., England. T h e xylenol orange was supplied by E. Gurr, Ltd., London. Standard solutions containing 5 meq. of the cation per 20 ml. in 0.1S sulfuric acid were prepared and standardized titrimetrically with EDTA VOL. 36, NO. 7, JUNE 1964
1217
Figure 1. Elution curves for Sc(lll) with concentrations
HzS04 . of various
solid mixture with XaN03. For the titration with 0.001M EDTA, the volume of the solution before titration was reduced to about 5 ml. Thus, 1 drop of 0.001.V EDTA (3about 2 pg. of Sc) produced a distinct color change a t the end point. About 1.20 H S O , and KH4OH were employed to adjust the pH value. The experimental elution curves are presented in Figure 1. From the single element elution curoes of scandium and from the corresponding distribution coefficients of the various rare earth elements, it was concluded that 2 . 0 5 HzS04seemed to be the most favorable eluent for the separation. Composite elution curves for the ion pairs Sc(III)-Y~III), Sc(lII)-La(III)j Sc(111)-Ce(III), Sc(III)-Sm(III), Sc (111)-Gd( 111), Sc(111)-Er (111) , and Sc (111)-Yb(II1) were prepared using a 20-gram column and about 5 meq. of each of the two cations. The flow rate was kept a t 3.0 + 0.2 ml. per minute. Aliquots (25 ml.) were collected, and the amounts of scandium were determined as described above. Yttrium and the rare earths were determined by EDTA titration a t pH 5.8 using xylenol orange as indicator. For scandium, 0.OlM and 0.001M EDTA were used as described. The solutions were buffered by the addition of about 0.2 gram of hexamethylene tetramine. The elution curve for the Sc(II1)-Yb(II1) pair is given in Figure 2. Sc(II1)-Gd(II1) and Sc(II1)-Er (111) had a very similar elution curve, while yttrium and samarium appeared later in the eluate than ytterbium. No lanthanum or cerium could be detected in the first 800 ml. of eluate. The wide separation of the peaks of
or gravimetrically as the oxide after oxalate precipitation. Less concentrated solutions were prepared when required by dilution of the above solutions. Borosilicate glass tubes with fused-in glass sinters of No. 2 porosity and a buret t a p at the bottom were used as columns. The columns were fitted with 500-ml. dropping funnels connected by ground-glass joints. Two different sizes of columns were used. The larger ones were about 50 cm. long and 2.0 cm. in diameter; the smaller ones were 35 cm. long and 1.15cm. in diameter. The larger columns were loaded with 20.0 f 0.1 grams (dry weight at 105' C.) of Bio-Rad .4G 50W-X8 resin of 100- to 200-mesh particle size. The resin was in the hydrogen form. For the smaller columns, 5 grams of the same resin were used. The resulting resin beds were 16 to 17 and 12 cm. long, respectively. Procedure. Single element elution curves for scandium were prepared using a 20-gram resin column and 1.0, 1.5, 2.0, and 2.5N HzSOa as eluent. About 5 meq. of scandium in a solution containing less than 0.3N free acid were absorbed on the column. The scandium was eluted with sulfuric acid of the designated concentration at a flow rate of 3.0 i 0.2 ml. per minute, and 25-ml. aliquots of the eluate were collected with an automatic fractionator. After the excess of acid was removed by evaporation, the amount of scandium in the aliquots was determined by titration with either 0.01 or 0.001M EDTA (for small amounts) a t a pH of 3.5 to 4.0 using xylenol orange as indicator. The indicator was used as a 0.5%
$
Figure 2.
Elution curve for Sc(lll)-Yb(lll)
Sc(11I) and Yb(II1) suggests that it should be possible to use smaller columns, especially when the total amount of scandium plus rare earths to be separated is 50 mg. or less. An elution curve for the Sc(II1)-Yb(II1) pair was therefore prepared using a 5gram resin column, 1 meq. of each of the cations, and 2.l' H2S04as eluent. The flow rate was 2.0 0.2 mi. per minute. KO satisfactory separation was obtained. The elution curve is given in Figure 3. Since tailing caused by a slow exchange reaction rate seemed to be the main reason for the unsuccessful separation, the experiment was repeated using AG 5OU'-8 resin of 200to 400-mesh particle size and reducing the flow rate to 0.6 f 0.1 ml. per minute.. Under these conditions the separation was quite satisfactory. The elution curve 14 given in Figure 4. Analysis of Synthetic Mixtures. As a result of the foregoing work, a method for the separation of scandium from yttrium a n d rare earth metals was elaborated and applied to the analysis of synthetic solutions containing scandium and one rare earth element. Appropriate amounts of standardized solutions of the cations were measured out with a pipet and mixed. The cationq were absorbed on a column containing either 20 grams of 100- to 200-mesh resin or 5 grams of 200- to 400-mesh resin. Scandium was eluted from the larger rolumns with 450 ml. of 2W H q S 0 4 at a flow rate of 3.0 f 0.2 ml. per minute. The sulfuric acid was removed from the column by washing it with 50 ml. of 0,LV HC1, and the rare earths were eluted with 500
*
20
B a 10
s
3
a
250
Figure 3.
1218
-
$L
Y 50
100 ML ELUENT.
IS0
Elution curve for Sc(lll)-Yb(lll)
ANALYTICAL CHEMISTRY
200
ML
Figure 4.
ELUENT
200
Elution curve for Sc(lll)-Yb(lll)
ml. of 4 5 HCl at thf. same flow rate. From the smaller columns, scandium was eluted with 70 ml. of 2 S H2S04 a t a flow rate of 0.4 i 0.1 ml. per minute. The washing with 0.1S HC1 was omitted, and the rare earths were eluted with 150 ml. of 4 S HC1 a t the same flow rate. The excess of acid was removed by evaporation, and scandium and the rare earths were determined by titration with EDTA% as described above. When large amounts of rare earths were present, these were determined gravimetrically as the oxides after precipitation of the oxalates from 0.1S "Os, and the ignition of the precipitates to constant weight at 1000" C. Instead of removing the sulfuric acid by evaporation, the scandium was precipitated as the hydroxide by neutralizing with ammonia to pH 7.5. Beryllium, which does not interfere with the E D T A titration, was added :is a carrier when only small amounts of scandium are present. The precipitate was separated by filtration and dissolved in a minimum of hot dilute nitric actd. The acid was removed by evaporation and the scandium determined by EDTA titration. This variation was useful when the large columns were used because the evaporation of larger amounts of sulfuric acid was time-consuming. The results of the determinations are presented in Table I (large columns) and Table I1 (small columns). DISCUSSION
The described method provides a simple means of separating scandium from yttrium and the rare earth elements. It has the advantage that it does not introduce into the sample solution large amounts of organic complexing agents which complirate later separations and determinations. Furthermore, the columns are used in the hydrogen form and the addition of inorganic ions, like sodium or ammonium, is omitted. The scandium shows some tailing, but the elution peaks are E O widely separated that the elution of scandium can be prolonged until the recovery has become quantitative (>99.9yo) before the rare earths appear in the eluate. Experiments showed that ori eluting a 5-gram column with 70 ml. of 21V HzS04 at a flow rate of 0.6 i 0.1 ml. per minute, less than 10 pg. of scandium, out of a total of 15.08 mg., remained on the column. With the larger columns, 5 meq. of scandium can be separated quantitatively from 5 meq. of Y(III), La(III), Ce(III), Sm(III), Gd(III), Er(III), or Yb(II1). Since rare earths from the cerium as well as from the terbium and yttrium groups are included in this number, it seems reasonable to assume that the separation procedure can be applied to separate scandium from all the rare earth elerrents. When lanthanum and cerium are predominantly present, up to 10 mec. of rare earths can be separated from 10 meq. of scandium.
Table
I.
sc 75.4 75.4 75.4 75.4 75.4 75.4 75.4 7.54 0.377 0.189 150.8 7.54 0,377 0.189 150.8
Table
II.
sc 15.08 0.377 0.189 15,08 0.377 0.189
Results of Quantitative Separations Using 2 0 - G r a m Resin Column
Taken, mg. Other La(III), 221.8 Ce(HI), 228.7 Sm( III), 249.8 Gd(III), 261.6 EriIII). 279.1 Yt(III), 288 4 Y(III), 146 7 La(III), 443 6 La(III), 443 6 La(III), 443 6 LalIII). 0 887 Y t i m j ; 288.4 Yt(III), 288.4 Yt(III), 288.4 Yt(III), 1 . 4 4
Found, mg. sc
Other
75.5 f 0 . 1 75.4 f 0 . 1 75.5 f 0 . 2 75.3 f 0 . 1 75.4 f 0 . 1 75.3 f 0 . 1 75.3 0.1 7.55 f 0.01 0.377 f 0.002 0.188 f 0.002 50.9 f 0 . 2 7.54 f 0.01 0.376 f 0.003 0.188 f 0 . 0 2 50.7 f 0 . 1
*
221.7 228.9 249.8 261.5 279.0 288.2 146.6 443.5 443.5 443.7 0.887 288.4 288.5 288.5 1.43
f0.1 f0.2 f0.2 f0.1 f0.2 f0.3 i0.1 f 0.2 f0.3 f0.2 f 0,009 f0.2 f0.2 f0.2 f 0.01
Results of Quantitative Separations Using 5 - G r a m Resin Column
Taken, mg. Other La(III), 4 4 . 5 La(III), 8 8 . 9 La(III), 177.8 Yt(III), 5 7 . 7 Yt(III), 5 7 . 7 Yt(III), 5 7 . 7
With the smaller columns, u p to 1 meq. of scandium can be separated from 1 meq. of rare earths. With the larger columns, 0.2 mg. of scandium can be separated from more than 400 mg. of lanthanum and from 200 mg. of ytterbium. The separations with the smaller column have the important advantage that much less sulfuric acid is used and, therefore, its evaporation takes much less time. Furthermore] the amount of trace impurities, like iron and aluminum, which can interfere with the E D T A titration by blocking the xylenol orange indicator] is much smaller. Using the 5-gram column, a separation and determination of scandium in a mixture with rare earths can be completed in about 5 hours. Because only a small fraction of this represents actual working time and because quite a number of separations can be carried out simultaneously, the method is suited for routine work. When optimum conditions are employed] the separations are sharp and the method is suitable for highly accurate reference analysis work. Flow rates are critical because the tailing of the scandium increases with increase of flow rates, especially with the smaller columns (Figures 3 and 4 ) . The tailing effect must be ascribed to the fact that s d d i u m has a slow exchange reaction rate. This is to be expected because all cations with high fundamental cation exchange affinities have slow cation exchange reaction rates [Th(IV), Zr (IV), rare earths(II1) 1.
sc
Found, mg.
15.07 f 0 . 0 1 0.378 f 0.002 0.190 f 0.002 15.08 f 0 . 0 2 0.377 f 0.002 0.190 f 0.003
Other 44.6 f 0.1 88.9 f 0 . 1 177.7 f 0 . 2 57.7 f 0 . 1 57.8 f 0.1 57.8 f 0 . 1
Table 111. Equilibrium Distribution Coefficients in 1N HCI and 1N HzS04
Cation
Kd in In' HC1
Kd in 1 S HZSO,
La( 111) CelIIT) Sm(IIi) Gd(II1) Yt( 111) Y(II1) Sc(II1)
265 265 217 183 153 145 120
326 3 18 269 246 249 253 35
The equilibrium distribution coefficients indicate considerable complexing action of the sulfuric acid on scandium. This is demonstrated by the coefficients for scandium and some rare earths in 1N HC1 and 1 S HzS04, as shown in Table 111. The coefficients were determined according to a method described before (14). The coefficients of yttrium and the rare earths are higher in sulfuric acid than in hydrochloric acid, as one would expect from the effective hydrogen ion concentrations. For scandium] the case is reversed. From the depression of the distribution coefficient, the overall complexing action of the sulfuric acid on scandium can be estimated as being comparable to that on uranium. It therefore seems possible that scandium will be absorbed from sulfate solutions by anion exchange resins in the sulfate form. VOL. 36, NO. 7 , JUNE 1964
1219
( 7 ) Iya, V. K., J . Rech. Centre .\-atl. Rech. Sci. Lab. Bellevue (Paris) 35, 91
LITERATURE CITED
(1) Faris, J . P., ANAL. CHEM.32, 520 (1960). ( 2 ) Faris, J. P., Warton, J. W.,Ibid., 34, 1077 (1962). (3) Fritz, J . S., Umbreit, G . R., Anal. Chim.Acta 19, 509 (1958). (4) Hamagushi, H., Kuroda, R., Aoki, K., Sueishita. R.. Onuma. X.. Talanta 10. 1 6 (1963). ’ ( 5 ) Hamagushi. H., Kuroda. R.. Onuma. N., Ibid, p. 120. ’ (6) Hamagushi, H., Kuroda, R., Shimizu, T., Anal. Chim.Acta 28, 61 (1963). I
,
I
,
’
(1956). (8) Iya, V . K., Loriers, J., Compt. Rend. 237, 1413 (1953). (9) James, D. B., Powell, J. E., Spedding, F. H., J . Inorg. Sucl. Chem. 19, 133 (I961 \ \ - - - - I
(10) Minczewski, J., Dybczynski, R., J . Chromatog. 7,568 (1961). (11) Radhakrishna, P., Anal. Chzm. Acta 8, 140 (1953). (12) Rezanka, I., Frana, J., Vobecky, M., Mastalka, A,, J . Znorg. .Vucl. Chem. 18, 13 (1961). (13) Spedding, G. H., Powell, J. E.,
Daane, A . H., Hiller, M. A,, ildams, W . H., J . Electrochem. SOC. 105, 683 (1958) \..__,
(14) Strelow, F. W. E., AKAL. CHEM.32, 1185 (1960). (15) Strelow, F. W. E., unpublished data, Sational Chemical Research Laboratory, 1961-63. (16) x’ickery, R. C., J . Chem. Soc. 1955, p. 245. (17) Yoshimura, J., Takashima, Y., Waki, H., Xippon Kagaku Zawhi 79, 1169 (1958). RECEIVED for review Xovember 29, 1963. Accepted February 4, 1964.
Some Factors Affecting Mobility Measurements in Paper Electrophoresis R. H. HACKMAN and MARY GOLDBERG Division of Entomology, Commonwealth Scientific and Industrial Research Organization, Canberra, A.C.T., Australia
F The mobilities of serum albumin, glucose, and maltose, but not of glutamic acid and lysine, increased with the wetness of the paper. Both the applied potential gradient and the wetness of the paper influence the resolution of serum proteins in barbital buffer. In carefully controlled experiments it is possible to achieve a high degree of reproducibility as well as uniformity of conditions throughout a paper strip.
I
N PAPER ELECTROPHORESIS the dis-
tance travelled by a charged molecule depends on a number of factors and these have been discussed by McDonald (5) and Ribeiro, Mitidieri, and Affonso (7). Very little is known about the effects of variations in the ratio of water to paper-i.e., wetness-and wetness is certainly the one factor over which least control has been exercised. The apparatus described by Bailey and Hackman (1) has been modified and used to study the effects of paper wetness and, in the case of proteins, of potential gradient on mobility. The degree of wetness of the paper is controlled by the pressure applied to the paper. EXPERIMENTAL
The apparatus of Bailey and Hackman ( 1 ) was modified to give an increased flow rate of water and hence an increased rate of cooling. Half-inch inlet and outlet tubes were located on the center line of the lower surface of the bladder near each end and passed through the base of the apparatus. Experiment showed that with a flow rate of 4.5 liters per minute or less the pressure throughout the bladder was uniform. The flow rate was maintained a t 4.5 liters per minute by inserting a suitable choke in the outlet tube. 1220
ANALYTICAL CHEMISTRY
Water, thermostatically controlled a t 15’ f 0.1’ C., circulated through the overflow device and bladder. A strip of Whatman No. 3 filter paper (57 X 12.7 cm.) was wetted with buffer solution and the two ends (for a length of 7.5 cm.) pressed between filter paper. Surplus liquid was removed from the central 42 cm. by compressing the strip between two pieces of cotton cloth for 1 minute a t a pressure 4 mm. of mercury greater than that a t which the experiment was to be performed. The paper strip was removed and compounds ( 5 - p l . spots) to be investigated were placed along a line a t the center of the strip. Temperature rise in and evaporation from the exposed ends of the strip-i.e., those parts not under pressure-were prevented by the use of wicks made from Whatman No. 3 filter paper. Two pieces 14 X 12.7 cm. and two pieces 12.7 x 12.7 cm. were wetted with buffer solution and pressed between filter paper. A long and a short piece of the wick were placed above and below each end of the strip, the shorter pieces being outermost and the extreme ends of the strip and wick being in line. The wicks, including the ends of the strip, were completely enclosed in a polythene sleeve 14 cm. long. This arrangement gave a stepped effect to the wick, which, when the paper was put under pressure, minimized distortion of the bladder. Pressure was applied to the paper strip and the ends of the wicks were dipped into buffer solution in the electrode compartments. Each compartment contained 1 liter of buffer solution and the apparatus was levelled to prevent syphoning of liquid through the paper. The strip was allowed to come into equilibrium with regard to buffer content (and temperature) and experiments showed that the time required was 1.5 hours with a pressure of 24.3 mm. of mercury (0.47 p s i . ) and 1 hour at a pressure of 254.5 mm. of mercury (4.92 p.s.i.). A potential difference was applied between
the electrodes in the buffer compartments so that the potential gradient along the central 25 cm. of the strip was 20 volts per cm. (or 5 volts per cm.) as determined in separate experiments by the use of platinum electrodes placed on the paper. At the conclusion of the experiment the paper strip was removed after the wicks had been torn off and discarded. The strip was dried and the positions of the spots were determined with an appropriate reagent. This procedure gave a linear relationship between migration distance and time and between migration distance and potential gradient for all the substances under investigation. No experiment was considered satisfactory unless the potential difference and current were constant throughout the entire experiment. At a constant potential difference, constancy of current is an indication of steady conditions both of temperature and of wetness of paper. The temperature of the paper, as measured in separate experiments by very small thermocouples enclosed in very thin glass tubing placed between the paper and the glass plate, was constant during an experiment, was uniform over the area of paper being used for mobility measurements, and was reproducible. The temperature recorded on top of the paper is not necessarily that of the paper itself, which is probably cooler. The ratio of water to paper for each buffer a t each pressure was measured by conducting experiments in which that part of the paper strip between the wicks was enclosed in Mylar film (0.0005 inch). Aipreviously marked area was cut from this central part (both film and paper), wrapped in Mylar film, weighed, and dried to constant weight a t 110’ C. Experimental conditions are given in the tables. Serum proteins were also separated under a potential gradient of 5 volts per em. for 8 hours. Amino acids were detected with ninhydrin, sugars with benzidine (8), proteins with bromphenol blue (9), and dextran with