An Accurate Determination of Rare Earths and Scandium in South African Carbonatites Using Separation by Ion Exchange Chromatography SIR: Edge and Ahrens (1) have separated Sc(III), Y(III), Ce(III), Nd(III), La(III), Sr(II), and Ba(I1) from large amounts of the elements more abmdant in rocks by eluting the latter with 3N HC1 from a cation exchange column, while the rare earths and Sr(I1) and Ba(I1) remain on the column. The separation can be applied only when trace amounts of these elements are present. An early breakthrough occurs with larger amounts especially for Gd(III), Er(III), and Sr(I1) which have distribution coefficients of 11.4, 10.7, and 10.0, respectively. Fritz and Garralda ( 2 ) have separated the rare earths, d l ( I I I ) , and other trivalent cations from divalent cations such as Cu(II), Xg(II), etc., using Dowex 5OW-XS resin and 1.5N "03 as eluent. Hydrochloric acid was considered less suitable because an early break-through of aluminum occurred. Because elements such as Al(III), Fe(III), Be(II), U(VI), and Ti(1V) accompany the rare earths through a hydroxide precipitation step and interfere with the completeness of an oxalate precipitation by complex formation, an ion exchange separation of the rare earths from these elements would be valuable to the analyst. For the purpose of such a separation HC1 is a more favorable eluent because Al(III), Fe(TII), and Ti(1V) have lower distribution coefficients in HC1 than in the while the same concentrations of "03, distribution coefficients for the rare earths are approximately the same in the two acids. EXPERIMENTAL
Equilibrium Distribution Coefficients. Previously, only the cation exchange distribution coefficients of L a ( I I I ) , Ce(III), and Y(1II) in hydrochloric acid had been determined (3). For this work those of Sm(III), Gd(III), Er(III), Yb(III), and Sc(II1) also were investigated to get more complete information. The methods used were similar to those described previously (3, 6). The resin was the BioRad AG50W-X8 sulfonated polystyrene of 100- to 200-mesh particle size. The coefficients are presented in Table I. Determination in Carbonatites. From the known distribution coefficients it was decided that 1.75N HCl should be the most favorable eluent. Elution experiments showed that Fe(III), Ti(IV), Be(II), V(VI), Mg(II), Mn(II), and many other cations could be quantitatively eluted from a 20-gram (dry weight a t 105'
Table I. Equilibrium Distribution Coefficients in Hydrochloric Acid 2.ON 3.ON 4.0.~ 1.ON 0.2N 0.5N Cation O.lN 28.8 14.9 11.7 500 120 3200 Sc(II1) >io4 217 39.0 15.4 8.6 SmlIII) >lo4 >io4 1330 183 30 2 11 4 6 1 >io4 1220 GdiIIIj >io4 6 0 27 2 10 7 990 165 >io4 Er(II1) >lo4 27 4 12 2 7 4 960 153 >lo4 >io4 Yb(II1)
C.) column of AG50W-X8 resin 18 cm. in length and 2.0 cm. in diameter by 400 ml. of 1.75N HC1, a t a flow rate of 3.0 =t 0.2 ml. per minute, but 600 ml. of 1.75.V were necessary for the quantitative elution of Al(III), Ga(III), and Ca(II), because these elements are more strongly absorbed and show some tailing. From the above facts a method for the precise and accurate determination of the rare earths and scandium in South African carbonatites containing small amounts of these elements was elaborated and applied successfully. Procedure. About 5 grams of carbonatite were accurately weighed out, and the bulk of the sample was dissolved by boiling with about 4N HC1. Any insoluble residue was separated by filtration and the filtrate was kept. After ashing the filter paper the residue was treated with 1 t o 2 ml. of perchloric acid and a few milliliters of hydrofluoric acid in a platinum crucible. Fluoride and excess of perchloric acid were removed by evaporation almost to dryness, and the residue was then dissolved by heating with a small volume of about 4N HCl. This effected complete dissolution in most cases. When any insoluble residue remained it was separated by filtration, and the filtrate was added to the main filtrate. After ashing the filter paper, the residue was fused in a platinum crucible with a small amount of potassium bisulfate. The melt was cooled, dissolved in a small amount of about 1N HC1 and kept separately. The rare earths, scandium, ferric iron, etc. in the main filtrate were precipitated as the hydroxides a t a pH of about 8 from a large volume of solution by the addition of ammonium hydroxide. After boiling for a short time the precipitate was separated, washed with 2% ammonium chloride solution which had been adjusted to pH 8 with ammonium hydroxide, and dissolved in hot ca. 1N HC1. The filtrate from the hydroxide precipitation was slightly acidified, about 10 mg. of ferric iron were added as a chloride solution, and the iron and any remaining rare earth oxides were precipitated as the hydroxides at a pH of about 10.5 to 11 with a considerable excess of ammonium hy-
Table II. Results for Determination of Rare Earths plus Scandium in South African C a rbonatites"
Sample B B B B B B B
37 54 62 74 83 95 102
yo Rare earth oxides found
1.242 zk 0.001 0.644 f 0.001 1.735 f 0.001 0.052 f 0.001 0.562 f 0.001 0.348 f 0.001 0.781 f 0.001
Synthetic sample containing 1.018% R.E.* 1.017 f 0.001 Results are means of triplicate determinations with calculated standard deviations. A sample weight of 5.000 grams was assumed. The result is a mean of six determinations.
droxide. After a few minutes' boiling the precipitate was separated, washed, and dissolved in hot ca. 1N HC1 as described above. The solution was added to that of the first hydroxide precipitate and the whole precipitation process was repeated. The hydrochloric acid concentration in the final solution was adjusted to about 0.5N by dilution with distilled water. This solution was then percolated through a 20-gram resin column as described above, and the sample was washed onto the resin with about O.1N HC1. Ferric iron, Al(III), Ti(IV), U(VI), Be(II), etc. were then eluted with 600 ml. of 1.75N HCl, followed by the elution of scandium plus the rare earths with 800 ml. of 3.00-V HC1. A flow rate of 3.0 + 0.2 ml. per minute was maintained throughout. The rare earth containing eluate was taken almost to dryness and the rare earths plus scandium were precipitated as the oxalates from a small volume (10 to 50 ml. depending on amount of rare earths present) of ca. 0.05N HCl by the addition of excess of oxalic acid. After standing over-night the precipitates were separated by filtration, ignited to the oxides a t 900' C., and weighed. The analysis results for 7 carbonatites are presented in Table 11. VOL. 38, NO. 1, JANUARY 1966
127
RESULTS A N D DISCUSSION
The results show an extremely high degree of precision as can be seen from Table 11. The deviation from the mean for all the single determinations was not larger than the average weighing error; but because no suitable standard samples were available and because spectrochemical methods and x-ray fluorescence spectrometry are not sufficiently accurate, it was not possible to establish whether the accuracy of the method was of the same order as the precision. A synthetic carbonatite sample was therefore simulated by mixing appropriate amounts of standard solutions and carrying out the above separation procedure on the mixture, which contained 1718 mg. Ca(II), 60.8mg. Mg(II), 2.8mg. Sr(II), 12.7mg. Ba(II), 45.6 mg. Fe(III), 57.8 mg. Ti
(IV), 89.8 mg. Al(III), 12.3 mg. U(VI), 5.6 mg. Be(II), 3.4 mg. Nb(V), and 15.1 mg. Mn(I1) in about 0.5N HCl, as well as 50.92 mg. rare earths consisting of 22.08 mg. La(II1) and 28.84 mg. Yb (111). The results of the determinations of the amounts of the rare earths are included in Table 11. They show that the accuracy as well as the precision of the method is very high. It is therefore well suited as an accurate reference method for the determination of the rare earths and scandium in carbonatite samples or minerals of similar composition; but it should be pointed out that for the procedure as described above the total amount of rare earths present should not be higher than about 2 to 2.5 meq. If a separation of scandium from the rare earths also is required, either
cation (5) or anion exchange chromatography (4) in sulfuric acid media can be employed using a solution of the final rare earth oxides in this acid. LITERATURE CITED
(1) Edge, R. A., Ahrens, L. H., Anal. Chim. Acta 26, 355 (1962). ( 2 ) Fritz, J. S., Garralda, B. B., Talanta 10, 91 1196.1) ' - \----,.
(3) Str,elow, F. W. E., ANAL. CHEM.32, 1185 (19t30). (4) Strelow, F. W. E., J. S. African Chem.
Znst. 17, 114 (1964). ( 5 ) Strelow, F. W. E., Bothma, C. J. C., ANAL. CHEM.36, 1217 (1964). ~
(6) Strelow. F. W. E.. Rethemever. R.. Bothma, C. J. C., Zbid., 38, 106"(1965):
F. W. E. STRELOW
Kational Chemical Research Laboratory South African Council for Scientific and Industrial Research Pretoria, South Africa
Stripping and Voltammetric Determination of Manganese via Manganese Dioxide SIR: Stripping analysis is becoming a well established analytical technique. The hanging mercury drop as well as platinum and gold electrodes have been used for determination of metals by anodic stripping using voltage scan coulometry, voltage step coulometry or fast sweep voltammetry (1, 5 , 7). The corresponding technique of cathodic stripping of oxidized species has been applied in the determination of halides (6, 8) by the formation and analytical reduction of silver and mercury halide films. Other examples might be the coulometric determination of metal surface oxides and sulfides via reduction (3).
Stripping analysis by reduction of an insoluble metal oxide deposited on a noble metal electrode has not previously been reported. In the work reported here the analysis of manganous ions by stripping of anodically deposited manganese dioxide on a platinum electrode was investigated. The voltammetric determination of manganous ion by oxidation to manganese dioxide a t a platinum electrode is also reported. EXPERIMENTAL
Apparatus. A conventional rotating platinum electrode (about 1/2 X 5 mm.) was used in an H-cell. The counter electrode was a mercury-mercurous sulfate electrode in saturated potassium sulfate. Voltages were applied from a simple battery powered adjustable source. Manual currenttime integrations were done by obtaining current us. time curves with a microampere measuring potentiometric recorder. A4reasunder the current-time 128
ANALYTICAL CHEMISTRY
curve were estimated by counting rectangles alternatively. An electronic integrator utilizing a d.c. analog amplifier coupled to a digital voltmeter was used for direct readout of stripping coulombs. Procedure. The procedure for electrode pretreatment consists of preanodizing the electrode a t 1.4 volts us. S.C.E. for 30 seconds in O.1M sodium perchlorate after which the potential is changed t o 1.0 volt and is maintained a t this value through all subsequent steps. The residual current is allowed to come t o a steady value. This value is about 0.1 pa. and is usually established in one or two minutes. The manganous solution is then added to the cell to begin the deposition step. The deposition step is terminated by draining the sample solution from the cell. The cathodic stripping step consists of coulometric measurements when 0.1M perchloric acid is placed in the cell. For manual integration the coulombs flowing until the current is constant or zero are included, while for electronic integration a measured stripping time is used. RESULTS A N D DISCUSSION
Stripping Analysis. The data for analysis of lO-8M-lO-5M manganous solutions is shown on Table I. The precision for replicate determinations is reasonably acceptable for these concentrations. The number of microcoulombs divided by the product of concentration and deposition time (&/et) column shows that this quantity drops sharply for the most concentrated solution in each series. This result may be related to the fact that not all of the manganese dioxide deposited during the deposition step is reduced during the
stripping as evidenced by the fact that for 10-5Ji solutions a brown coloration of the electrode is visible after several minutes of stripping at which time the stripping current has decayed to negligible values. The &/et values for concentrations less than 10-5.M are reasonably constant. The practical consequence is that for electrodes of the dimensions used here a t least, the method is best suited to concentrations of 10-6M or less. When sodium perchlorate was replaced by acetate buffer in the deposition step nonreproducible amounts of manganese dioxide were deposited. This was not corrected by deposition a t more positive potential and may be attributed to formation of some manganic species. Because these measurements are made a t relatively positive potentials, the reproducibility of the electrode surface conditions is very important. When, for example, the determination was attempted without first anodizing the electrode a t 1.4 volts virtually no manganese dioxide deposited. It was also found that the purity of the distilled water used during the electrode pretreatment was very important. Only when water redistilled from oxidizing solution was used was it possible to obtain a reproducible manganese dioxide deposit. This high dependence on electrode preconditioning corresponds to the results found by Fleischmann (2) indicating that manganese dioxide deposition currents are dependent on preformation of nucleii which are formed a t preferred sites on the platinum surface. The lower sensitivity limit is established by the current pulse obtained for