Coprecipitation Studies with Ferric Periodate Precipitated from Homogeneous Solution LOUIS GORDON and LEONARD GINSBURG' Deportment o f Chemistry, Syrocuse University, Syrocuse 10, N.
,A method for the precipitation o f ferric periodate from homogeneous solution is described. This compound was used CIS a carrier in studying the coprecipitation of aluminum, yttrium, and zinc. In the case o f aluminum, coprecipitation was negligible until at least 99.9% o f the iron was precipitated. With yttrium, coprecipitation was again slight until 99.9y0 of the iron hod been precipitated. The percentage o f yttrium coprecipitoted was independent of the initiol yttrium concentration within a fourfold change. The study o f the coprecipitation o f zinc, in comparison with observations of the aluminum and yttrium systems, showed that considerable coprecipitation occurred prior to the end of the precipitation. The extent of coprecipitotion o f zinc varied with the initiol zinc concentration: A good opproximotion to the Freundlich adsorption isotherm was observed with the data of the zinc study. The present studies again demonstrate that occlusion during the formotion of the precipitate can b e a negligible port o f the over-all coprecipitation process. The information obtained in these studies may b e useful in the preparation of relatively pure carriers.
P '
from homogeneoussolution has been used in recent investigations (6-9,11,19,93, g4) to study coprecipitation. Results obtained by Gordon, Teicher, and Rurtt (9) in their study of the coprecipitation of manganese with basic stannic sulfate as carrier indicated that most of the coprecipitation occurred toward the latter stages of the precipitation and that within rather wide limits the extent of coprecipitation was essentially independent of the initial concentration of manganese tracer. The present investigation is concerned with systems which exhibit characteristics similar in many respects. The carrier is ferric periodate which was precipitated from homogeneous solution. Separate coprecipitation studies were made with aluminum, yttrium, and zinc. RECIPITATION
REAGENTS AND APPARATUS
All reagents used were reagent grade. Present address, Research Department, Jones and Laughlin Steel Carp., Pittsburgh, Pa. 38
ANALYTICAL CHEMISTRY
Y.
Paraperiodic acid, reagent grade crystals, G. Frederick Smith Chemical Co., Columbus, Ohio. l,l0-Phenantbroline, G . Frederick Smith Chemical Co., Columbus, Ohio. A 0.1% solution was prepared by diss o l ~ n 1.OOgram g of thereagentina,hont 600 ml. of hot water and dilutinx to 1 liter (15). 8-Quinolinol, reagent grade, Baker & Adamson. Two erams was dissolved in
purification. Hydrazine hydrate, purified, 85% solution in water, Fairmount Chemical Co., Newark, N. J. No detectable amount of iron could be found in this reaeent. ~o IRON SOLUTION. Approximately 0.9 gram of electrolytic iron, Standard Sample Co., Ames, Iowa, was dissolved in about 5 ml. of hydrochloric acid. Five milliliters of sulfuric acid was added and all the chloride was driven off by evaporation to fumes of sulfur trioxide. The ferric sulfate was dksolved and the following were added: 54 ml. of (1 to 1) sulfuric acid, 2.425 grams of paraperiodic acid, and 80 grams of ammonium sulfate. The solution was then diluted to 2 liters. ALUMINUM SOLUTION. Approximately 4 grams of 99.99+% aluminum metal, obtained from the Aluminum Go. of America, was dissolved in hydrochloric acid and evaporated to dense fumes with sulfuric acid. The solution was diluted to 2 liters and standardized by precipitation of the oxinate according to the directions of Kolthoff and Sandell (14). The solution was found to contain 20.3 mg. of aluminum per 10 mI. By suitable dilution a solution was prepared that contained 1.015 mg. of aluminum per 20 ml. YTTRIUM-91 SOLUTION.Two millicuries of carrier-free yttrium was ohtained from Oak Ridge National Lahore tories as yttrium-91 chloride in hydrochloric acid solution. Radiochemical purity was more than 99%. The sohtion was evaporated to fumes of sulfur trioxide with sulfuric acid and diluted to 100 ml. ZINC-65 SOLUTION.One-tenth milliliter of a solution of zinc chloride in hydrochloric acid, containing 12.2mg. of zinc and with an activity of 4.8 millicuries, was obtained from Oak Ridge National Laboratories. This mas evaporated to dense fumes with sulfuric acid and diluted to volume in a 100-ml. volumetric flask, to give a zinc-65 stock solution. Zinc solution I was prepared by diluting 7.5 ml. of the stock solution to volume in a 50-ml. volumetric flask.
Zinc solution I1 was prepared by adding 25 mg. of stable zinc as the sulfate to another 7.5-ml. portion of the zinc-65 stock solution and diluting to volume in a 50-ml. volumetric flask. SPECTROPHOTOMETER. Absorbance measurements were made with a Beckman Model B spectrophotometer using I-em. cells.
Figure 1. Comparison of apparent volumes precipitated b y two different methods A. B.
3 hours 1 day
C. D.
3 days 1 week
pH MEASUREMENT. pH measurements were made on a Beckman Model H-2 pH meter which was calibrated with standard buffers (86). PRECIPITATION CELLSUSED IN CoPRECIPITATION STUDIES. Precipitations were carried out in a 50-ml. cell which
consisted of a cylindrical glass bottle fitted with a T 5/12 stopper, and w9.s about 90 mm. long. Upon completion of the precipitation process, the stopper was removed and the cell coupled to a filter assembly consisting of a sinteredglass frit and a receiver cup. The re ceiver cup was a cylindrical glass tube closed a t one end and approximately BO mm. in length. All the components ivere within an outside diameter of about 30 mm., so that the entire glass appsratus fitted into a 100-ml. brass Cornelltype centrifuge cup. The precipitate
Figure 2.
Agitator wheel
could thus be completely separated from the liquid phase by centrifugation.
I
PROPERTIES OF FERRIC PERIODATE PRECIPITATE
Initial experiments showed that an iron compound could be precipitated from homogeneous solution in a sulfuric acid or nitric acid medium containing ferric ion and periodic acid by hydrolysis of acetamide, at temperatures just below the boiling point of the solution. Chloride ion had to be excluded because of its reaction with periodate ion upon heating. In sulfuric acid, precipitation starts at pH 1.2 and is complete at ahout 2.5. I n nitric acid, the corresponding pH values are approximately 0.2 and 1.7. The difference is apparently due to degrees of complexation of ferric ion by sulfate and nitrate. There is considerable difference in the physical appearance of the precipitate formed from homogeneous solution and that obtained by a conventional technique. The latter precipitate is gelatinous, voluminous, and difficult to mash and filter. It is light yellow in color and settles very slowly. The precipitate obtained from homogeneous solution i s darker in color, dense, filterable, and easily washed. It settles comparatively rapidly and has an apparent volume equal to less than one twentieth that of the conventionally formed precipitate even after several days' settling. Figure 1 shows the striking difference in settling time and apparent volume of the precipitates formed by the two methods. However, there is no large discernible difference in the physical appearance of the precipitates as viewed under the microscope. An x-ray diffraction study made only with the dense precipitate obtained from homogeneous solution showed broad hands indicating a microcrystalline structure. It would thus
.-.- .-
..
100 r
.%'
/OK
Y
:/, 200
a,
400
600
,
800
TIME
Figure 3.
seem that the difference in the physical appearance of the two precipitates is due mainlv to the mode of Dackine of the crystais. An ironfIII) Deriodate of the formula Fe(IOSs has been postulated by Kimmins (IS). Fhychoudbury (16, f7) prepared a compound said to be FeHJ08. Smith (20) in an extensive survey of the literature of known periodates does not list any other Deriodates of iron(III). I n order to ascertain the formula of the precipitate a known quantity of iron was quantitatively precipitated, filtered, and dried. It was not possible to dry the precipitate to constant weight because i t lost weight slowly a t 110' C. From the initial weight of the iron taken and the approximate weight of the precipitate formed, the iron content of the precipitate was calculated to he 26 i 0.5%. The entire precipitate was then dissolved in hot sulfuric acid and the resulting solution cooled to room temperature. After the iron had been complexed with ammonium bifluoride solution, excess potassium iodide was added and the liberated iodine titrated in the usual manner with standard sodium thiosulfate solution to a starch end point. Assuming that all the iodine in the precipitate was present as periodate, the mole ratio of iron to iodine was determined to he 1.70 =t 0.02. The theoretical values for the compound Fes(IO&.7HZO are F e = 26.0% and Fe/I = 1.67. The compound is probably ferric paraperiodate or a basic ferric metaperiodate with the formula Fe(I0&.2Fe2O8.7H20,which would yield the same calculated values. The amount of water of hydration is probably variable.
,
1000
,
1200
,
,
1400
1600
IN MINUTES
Per cent iron precipitated
YE.
time
,
IS00
, 2000
PROCEDURES
USED IN COPREClPlTATlON STUDIES
Coprecipitation of Aluminum. To 50 ml. of the iron solution were added 20 ml. of the aluminum solution and 2.6 grams of acetamide. The p H mas adjusted to 1.0 and the solution mas diluted in a 100-ml. volumetric flask. A 20-ml. aliquot was transferred to each of four precipitation cells 18). I n each cell the following were thus present in the 20-ml. volume: 4.5 =t0.1 mg. of iron, 0.27 ml. of 1 to 1 sulfuric acid, 12.1 mg. of paraperiodic acid [lo% excess of the stoichiometric amount required for FedI03,l. 0.4 eram of ammonium sulface, 2%"; of a~nminum,and 0.52 gram of acetamide. Four stoppered cells were then inserted into brass cups, mounted on an agitator wheel, and immersed in a thermostated oil bath at 80' 0.8" C. for varying lengths of time.
*
The agitator wheel used was snggested by Stoenner (21) and is shown in Figure 2. The precipitation proceeded slowly; the per cent iron precipitated with time is shown in Figure 3. VOL. 29, NO. 1, JANUARY 1957
-
39
At the end of a desired reaction time, the cell was removed from the bath and the filtering sections were assembled as previously described. The filtrate was separated from the precipitate by centrifugation-finally a t 2200 r.p.m. The pH of the filtrate was determined and the solution was diluted in a 100ml. volumetric flask after addition of excess acid to prevent further precipitation. The precipitate was dissolved in the cell and frit with concentrated hydrochloric acid. The frit and cell were rinsed using suction. The r e s u b ing solution was finally diluted to volume. The filtrate and precipitate phases were then analyzed independently for iron and/or aluminum as described under analytical methods. Coprecipitation of Yttrium. For this study the experimental technique was identical to t h a t of the aluminum study, except for the substitution of 10 ml. of the yttrium solution for the aluminum. Usual techniques for handling radioactive isotopes were used. Coprecipitation of Zinc. For the study of the coprecipitation of zinc the following changes from the aluminum procedure were adopted.
For the initial zinc concentration of 10-sM, 5.00 ml. of zinc .solution I was substituted for the aluminum solution. To obtain an initial zinc concentration of 4 X 10-4M, 5.00 ml. of zinc solution I1 was used. For the coprecipitation with an initial zinc concentration of 3.5 X 10-3 M zinc, solution I was used with the addition of 22.5 mg. of stable zinc as the sulfate prior to the addition of the acetamide. In all other respects the procedure was identical with that used for the aluminum and yttrium studies. ANALYTICAL METHODS FOR PRECIPITATE AND FILTRATE PHASES
Iron. Precipitate and filtrate aliquots were analyzed for iron using the 1,lO-phenanthroline methods of Fortune and Mcllon @) and Mehlig and Hulett (16). One milliliter of hydrazine hydrate was used as the reducing agent instead of the hydroxylamine recommended by previous authors. The solutions were adjusted to pH 2.5 and allowed to stand overnight before the ahsorbance was determined a t 505 myr. A reagent blank was prepared for each run and was found to be negligible. Total iron present in each run was determined by combining the filtrate and precipitate analyses. This gave a mean value of 4.38 =z! 0.04 mg. of iron for 23 precipitations in the aluminum study. For cases where more than 90% of the iron was precipitated, better analytical results could be obtained by difference; thus i t was not necessary to determine iron in the 40
ANALYTKAL CHEMISTRY
precipitate. In these cases the determination of the per cent iron precipitated was based on the difference between the value for iron remaining in the filtrate and the experimentally obtained value for total iron. Because the entire filtrate was used for iron determination, accurate values for per cent iron precipitated were obtained in the range of very high fraetions of iron precipitated. This procedure also permitted the use of the entire precipitate for the determin% tion of aluminum, thus increasing the accuracy of this determination. I n the yttrium study, the material balance for iron gave a value which was somewhat greater than that obtained in the aluminum study, probably because of differences in pipetting techniques. The value ohtaincd was 4.56 i 0.03 mg. of iron as computed from the data of 16 precipitation NUS. For the i n c study the material halance for iron again gave a very reproducible value. For 16 precipitations in the zinc study the total iron amounted to 4.49 i 0.03 mg. for a new stock solution. For the precipitations with an initial zinc concentration of the order of lO-3M the quantity of zinc remaining in the filtrate interfered (#) with the method for iron. Therefore, in these cases the per cent iron precipitated was determined from the amount of iron found in the precipitate and the value for total iron. Aluminum. The method used for aluminum determination was adapted from that of Xassner and Ozier (18).
tate was thus analyzed for aluminum, the amount of aluminum was still so small as to cause a large relative error. Consideration of the various errors involved showed au estimated absolute error in aluminum of about 1 to 2%. Yttrium. Radiometric assay of yttrium-91 was used. It was found t h a t beta counting was much more reproducible if the liquids wcre counted in open vials than if they were evaporated to dryness, even though the counting efficiency was much diminished. Under the experimental conditions the variations in amounts
Effect of Different Fractions Table I. Precipitated on Counting Rate of Yttrium-90 Soh. Counts A Added, per M1. Minute* 0 2
6 12 16 20
Over-all average
7484 7355 7386 7550 7257 7307 7389 f 84
* E a h value is an average of three de-
terminations.
The solution to be analyzed was evaporated to about 25 ml. Two-tenths gram of tartaric acid was added, followed by 15 d.of a saturated solution of sodium sulfite and 1.5 grams of potassium cyanide. The solution was then heated until the brown color was completely discharged, which indicated that all the iron present was now in the form of ferrocyanide. The solution was then cooled, the pH adjusted to 8.9 i 0.3, as determined with pH paper, and the solntion extracted with four 5-ml. portions of the 8-quiuolinol solution. The combined chloroform layer was transferred to a 50-ml. volumetric flask and diluted to the mark with chloroform. About 3 grams of anhydrous sodium sulfate was added to dry the chloroform solution. The absorbance was determined a t 390 mp using 1-em. Corex cells. After the iron precipitate was dissolved, the solution was diluted in a 110ml. cassia volumetric flask. A 5-ml. aliquot was withdrawn for iron determinationand the volume of the solution remaining in the flask was readjusted to 100 ml., all of which was taken for aluminum determination. Aluminum waa not determined in the filtrate which usually contained almost all the aluminum initially added. Although virtually the entire precipi-
Figure 4. Apparatus for detection of gamma radiation
of other substances present did not have a n appreciable effect on the counting rate, as is shon-n by the data of Table I. Solution A was prepared so that 20 ml. contained the same quantities of all substances, except yttrium. which were present in a precipitation run. A constant amount of tracer was added to each of six 100-ml. volumetric flasks along with the indicated amount of solution A (cf. Table I). The flasks were diluted to rolunie and three 2-ml. aliquots withdrawn and counted under standard conditions. The average counting rate is given for each solution.
Table II. Total Counts per Minute of Zinc-65 a t Various Fractions of Iron Precipitated
Iron Pptd.,
72
5 4:3
45 68 70 95 99
Average
Counts per Minute” 126,150 126,120 125,660 123,230 125,010 123 250 124;350 124,360 d= 1,olio
Sum of counting rates of precipitate and filtrate corrected for coincidence and decay. Q
Evaporation of the liquid caused the counting rate to decrease slowly upon repeated counting. Therefore, the standard procedure consisted of pipetting three 2-ml. aliquots into vials directly before counting and then counting each vial for 6.4 minutes. The solution in each vial was counted only once and then discarded. In many respects the analytical problem with yttrium was similar to that encountered with aluminum, because the coprecipitation of yttrium was very slight over most of the range of iron precipitated. Thus, when less than 10% of the yttrium was coprecipitated, the precipitate phase m s diluted to 100 ml. and analyzed for yttrium, whereas residual yttrium was not determined in the filtrate. In these cases the amount of yttrium found in the precipitate was conipared with that in a standard which was prepared by diluting 10 nil. of the yttrium-91 solution to 500 ml. In the range of very high fractions of iron precipitated it was desirable to determine yttrium in both phases a s well as to determine iron in as large an aliquot as possible of the filtrate. Again, use was made here of the 110-ml. cassia volumetric flasks. The filtrate was diluted to 110 ml. and three 2-ml.
aliquots were withdiaivn for radiometric analysis. The volume of the solution remaining in the flask was readjusted to 100 ml. and this ivas used for iron determination. The yttrium determination was in these cases based on 21110 of the filtrate, while the iron determination was based on 10’11 of the filtrate. The vial containing the liquid was reproducibly positioned under a GeigerRIiiller tube and counted by a conventional method employing preset time. The dead time of the tube was determined by the conventional split-card method to be 49 microminutes. Corresponding coincidence corrections w r e made. -4material balance could be computed for six runs in which the yttrium xas determined in both the filtrate and the precipitate. The average corrected for both coincidence and decay was 11860 1 1 5 0 counts per minute. As in the case of aluminum, it !vas necessary to determine very small quantities of tracer in the precipitate. In these cases the very high relative error incurred by the excessively small counting rate constituted only a small absolute quantity and therefore had very little effect on the general nature of the results obtained. Zinc. Radiometric assay of zinc-63 was used. Zinc-65 is a gamma emitter that decays with a half life of 250 days. Scintillation counting mas used as the detection method, using standard amplification and scaling equipment provided by the Atomic Instrument Co., Cambridge, Mass. A unit was devised whereby a 100-ml. volumetric flask could be reproducibly positioned n-ith reference t o the scintillation crystal. This device is shou-n in Figure 4.
cemented in place in such a fashion that the rectangle they formed was just large enough t o circumscribe the volumetric flasks. The flasks mere marked and could thus be rotated t o a reproducible position. In this manner, either the filtrate or the precipitate phase could be diluted t o volume in a counting flask and the counting rates directly compared. Each 100-ml. flask contained 50 ml. of concentrated hydrochloric acid as part of the diluent in order to give solutions of approximately the same density. The solutions were counted for a preset time interval of 10 minutes. The resolving time of the setup was found to be 1.6 microminutes and the appropriate coincidence corrections were applied to all counting rates. The counting flasks were calibrated to an arbitrarily chosen standard by counting the same standard solution in each. The counting rate was found to be essentially independent of the amount of iron and other constituents present in a flask.
-1Lucite shelf was mounted on the tietrction cell and four Lucite posts r e r e
Coprecipitation of Aluminum. The resiilts of the study with aluminum as
Table I1 shows the constancy of the total counting rate with the indicated variation of iron in the t r o phases. A material balance for zinc gave an average value of 124,580 =!= 850 counts per minute for 12 precipitations. As in the yttrium study, the low counting rates encountered where the fraction of iron precipitated was small caused high relative errors in the zinc determination. However, as no material was consumed in the zinc determination, amounts of filtrate and precipitate could be used to give the maximum attainable accuracy in the iron analysis. The large relative error in the zinc determination was a small absolute quantity and did not affect the over-all picture of the observed coprecipitation. RESULTS A N D DISCUSSION
3 40 0
w
30W 0
a
.!
J 3
a IRON PRECIPITATED (%)
2
EXPAND ED SECTION
..
IO
20
.-a
a- .
30 40 50 60 IRON PRECIPITATED
Figure 5.
70
80
90
I(
(Yo)
Coprecipitation of aluminum VOL. 2 9 , NO. 1, JANUARY 1957
41
tracer are shown in Table I11 and plotted in Figure 5 . The present system seems t o be similar to the manganese-tin system as studied by Gordon, Teicher, and Burtt (9). They found that, in cases where the initial manganese concentration was of the order of 10-+M, coprecipitation was very slight a t low fractions of tin precipitated. The precision of the manganese-tin work was somewhat limited by the analytical methods used. The investigation was restricted to measurements of residual solution concentrations only. Such measurements are admittedly poor in the region of low coprecipitation. I n the present study, equipment and methods permitted direct determination of trace contaminants in the precipitate. The aluminum-iron system appears to be an extreme example of the systems typified by the manganese-tin system. Coprecipitation of aluminum is small until 99.9+% of the carrier has precipitated, after which considerable coprecipitation occurs. Although no precipitations were carried out in which the final pH exceeded 3.3, it mas necessary to determine if the tracer was precipitating a t this pH independently of the presence of the carrier. A solution was prepared which contained all the reactants usually present in a run, with the exception of the iron. This solution was allowed to react in the usual manner until a final pH of 3.7 was reached. It was filtered in the usual way and the “precipitate” was analyzed for aluminum. The aluminum found was equivalent to 0.4 and 1.2% of the aluminum present, which was within the experimental error for such aluminum determinations. Thus, the present system represents a case of true coprecipitation, in that the tracer is normally soluble in the absence of carrier, whereas it coprecipitates with iron under similar experimental conditions. The folloving observations should be noted with respect to the observed coprecipitation. Initially, the ratio of the molar concentrations of iron and aluminum is of the order of 10. The ratio has changed to however, when 99.9% of the iron has been precipitated. Only a t this point, where the aluminum ions greatly outnumber the iron ions in the solution, is aluminum coprecipitated. Coprecipitation of Yttrium. In order to determine whether any coprecipitation phenomenon was being masked by the large amount of total aluminum present, it was desirable to extend this study to very much smaller concentrations of tracer ion. However, because the limit of sensitivity for the determination of aluminum had already been reached, it mas decided t o use a radioactive tracer. 42
ANALYTICAL CHEMISTRY
Coprecipitation of Aluminum with Ferric Periodate Iron Aluminum
Table 111.
Time,
Final PH
Min. 205 515 515 300 325 370 390 450 540 960 960 1145 1145 1175 1475
*
Ppt’d.,
%
D‘
A‘
3.73
2.8
55.3 59.4 71.1 83.8 93.5 94.2 97.98 97.92 99.38 99.38 99.95 99,98 99.98 99.98
1.1 2.5 4.1 1.7 2.9 2.1 2.9 2.9 2.5 3.9 5,O 3.9 16.1 15.7 22.4 21.4 39.8
0.74 0.10 0.10 0,005 0.028 0,009 0.018 0.017 0.003 0.002 0,001 0,001 0,001 0.0002 0.0003 0.00003 0,00001 0.00004 0.00004
0.75 0.10 0.12 0.006 0.037 0.014 0.028 0.034 0,009 0.011 0.007 0.008 0.008 0.005 0,008 0,007 0.005 0.021 0.020
%
1.6 0 0
1.7 1.7 .7
1585 2580 2580 4635 4635 4735 a
Pptd.,
a a
a
b b b
3.3 3.3 0
c
c
E
c
E
c
Not measured. S o detectable amount of iron could be found in filtrate. Not calculable.
Because a radioactive isotope of aluminum was not available, yttrium was then selected, since it is a trivalent cation and is readily available as a carrier-free beta emitter. The tracer was used in carrier-free form and its concentration was estimated to be approximately 10-1OM. The results of the yttrium study are shown in Table IV and plotted in Figure 6. A few precipitations viere carried out with carrier-free yttrium-90, prepared according to the directions of Salutsky and Kirby ( l a ) , and are reported in the last part of Table IV.
Because the yttrium-91 concentration does not remain constant due to decay, a control experiment using approximately 3.5 times the usual quantity of yttrium-91 was performed. This gave essentially the same coprecipitation results. Two such points are noted in Table IV. As the entire study n-as carried out within one half life of the yttrium-91, this effect cannot be attributed t o changes in concentration due to decay of yttrium-91. The results obtained in this study are almost identical with those of the aluminum study. The same similarities
7C c
60 n
2 -
50
a 0 40
I
w
(r
a
30
5 -t
+
>
.-
20 IO
.-. 94.00 95.00
IO
20
i
,i ‘
96.00 97.00 98.00 99.00 100.00 IRON PRECIPITATED (%)
,-,&EXPANDED
-
SECTION
*-._.-..-.-..-.*’.
.-a_.
0
I
30 40 50 60 I RON PRECIPITATED
Figure 6.
70 (%)
Coprecipitation of yttrium
80
90
100
Table IV.
Time, RIin. 220 295 325 235 350a 345 385 335 420 465 600 795 970 1070 1940" 1455 1800 2010 2320 2400 2320b 31.V .~. 390" 290" 50OC 43OC 58OC
Final P1-l
Coprecipitation of Yttrium with Ferric Periodate
Iron Pptd., c-
/C
Yttrium Pptd., c
/O
D'
A'
0 12 0 12 1.6 22. i0.019 0.024 1.65 29.1 0.015 0,018 1.6 32.4 0.013 0,015 1.8 32.1 0.013 0.015 1.7 55.0 0,007 0,010 1.75 0.9 0,005 0 009 63 3 1.7 0.005 25 5 0.9 0 008 0 004 0 008 1.8 12 7 1.0 0 007 1.1 1.85 78 9 0 003 94.93 0 004 2.0 1.6 0 0008 2.2 0 008 2,4 0 0003 98.63 99.06 2.3 3.3 0.0003 0 007 2.35 0.011 6.0 0.0002 99 62 2.65 99.78 12.6 0.0002 0.022 0 0001 0 017 2 55 99 89 11 1 2 9 99 92 37 9 0 0005 0 067 3 1 99 92 47 8 0 0007 0 091 0 12 3 2 99 95 61 0 0 0008 99.96 53.9 0.0005 0,099 3.2 3.25 99.96 54.9 0.0005 0.102 1 8 $4 1.1 0 004 0 008 . - x 1.9 88.5 1.4 0,002 0.006 1.8 89.8 1.5 0.002 0,007 2.0 95.8 1.9 0.0009 0.006 2.1 98.7 3.0 0,0004 0.007 2 3 99.5 6.2 0 0003 0 012 Experiments with approximately 3.5 times standard amount of yttrium-91. * Standard amount of yttrium-91 added after 99.97, of carrier had been precipitated. e Experiments performed with varying amounts of yttrium-90 used as tracer. 1 6
3 1,;
0 4 0.6 0.6 0.6 0.6 0.8
~~
Q
to the manganese-tin study can be observed. The curve (Figure 6) appears somewhat smoother in the lower regions of iron precipitated, probably because the analytical method used for yttrium is more sensitive. The extent of coprecipitation of yttrium was slight until 99.9% of the iron was precipitated, after which coprecipitation of yttrium became large. An experiment similar to that with aluminum was performed where the precipitation was essentially carried to completion (pH = 3.2) with all components present except iron. Approximately 0.3% of the total amount of yttrium was apparently precipitated. This value is negligible within experimetal error. T o confirm the fact that occlusion during precipitation is negligible, the following experiment was also performed. ST7ithoutyttrium present, precipitation was carried out to an extent of approximately 99.9%. The usual amount of yttrium-91 was then added to the precipitation cell and the reaction was permitted to proceed until 99.96% of the iron had precipitated. I n this experiment 54.9% of the yttrium coprecipitated; this can be compared to similar runs where the yttrium was present throughout the precipitation and yttrium coprecipitated to about the same extent. I n order to obtain a qualitative nieasure of the rate a t which coprecipitated yttrium would redissolve in an appropriate mother liquor the following experiment was performed.
TKOprecipitations were cai ried out in the usual manner and the solid and liquid phases werB separated By analysis of the filtrates it was determined that 40 and 64%, respectively, of the yttrium had coprecipitated. The filtrates were discarded and the solid phases were reserved for later use. Two solutions containing all the usual components of a precipitation run with the exception of yttrium were prepared, The pH of the solutions was adjusted with ammonia to 1.7 and 1.9, respectively, and the solids thus formed were separated from the liquid phase. Approximately 50 and 90% of the iron precipitated a t these p H values. The solids obtained were discarded and the filtrates which contained iron, but no yttrium, were thermostated a t 80" C. Each of the precipitates, prepared previously, which contained yttrium, was added to a filtrate with vigorous stirring. The yttrium concentration in the filtrate relative to the total amount of yttrium present was then determined a t various times. The results are shown in Table V At the end of 4 and 6 hours, 17 and 32%, respectively, of the coprecipitated yttrium went into the liquid phase. It is thus apparent that it takes considerable time for a precipitate containing yttrium t o equilibrate with fresh filtrate. On the basis of the general nature of the observed coprecipitation along with the rate of equilibration just cited, the conventional method of precipitation by means of droprvise addition of a precipitate may be compared with precipitation from homogeneous solution (4, 5, 66). I n the case of drop-
wise addition, a particle of precipitate is rapidly formed a t the point of contact of the precipitant with the solution. The small volume of solution immediately surrounding the solid is temporarily completed devoid of carrier ions. The situation here is then very similar to the solution situation a t the point of high coprecipitation in the systems under investigation (cf. Figures 5 and 6). A considerable amount of coprecipitation nould be expected with each small particle of precipitate. Since i t takes considerable time for equilibration to be complete, the coprecipitated impurity on a particle of freshly formed precipitate would not be expected to dissolve again immediately after the local concentration gradient has disappeared. In this manner the net eoprecipitation in a conventional precipitation n-ould be the sum of the coprecipitation occurring n ith the formation of each particle of precipitate. On the other hand. if a precipitate is formed from homogeneous solution, the effects of local concentration gradients are minimized. The depletion of solution concentration of carrier does not occur until almost all of the precipitate has been formed and the total amount of coprecipitation is much less. I n addition to this effect, precipitation from homogeneous solution causes the precipitate to be formed a t a slow rate, thereby virtually eliminating other effects such as inclusion and trapping, which cause additional coprecipitation in the conventional precipitation techniques. In addition to the indicated possibility of preparation of carrier relatively free of trace material, this study demonstrates that for these systems occlusion
Table V. Approximate Rate of Resolution of Coprecipitated Yttrium with Filtrates of Low pH
Elapsed' Time, Min.
Coprecipitated Yttrium Returned to Elapsed FiltrateTime, after
- %
Synthetic Filtrate Adjusted to pH 1.7; Precipitate with 64,% of Added Yttrium Coprecipitated Used 3 14 34 160 380
9 16 20 28 32
Synthetic Filtrate Bdjusted to pH 1.9; Precipitat,e with 40% of Added Yttrium Coprecipitated Used 4 4 17 32 74 120 240
6 7 9 11 17
VOL. 29, NO. 1, JANUARY 1957
43
plays a negligible part in the over-all coprecipitation mechanism. Coprecipitation of Zinc. The results of the coprecipitation study using zinc a s the tracer are shown in Table V I and in Figure 7 . This system does not seem t o be as extreme as the aluminum and yttrium systems, as considerable coprecipitation is observed much before 99% of t h e iron is precipitated. The extent of coprecipitation of zinc Tras found t o vary with t h e initial zinc concentration (cf. Figure 7 ) . Larger initial zinc concentrations resulted in coprecipitation of smaller fractions of t h e zinc. An experiment similar to that for the aluminum and yttrium studies was performed, wherein all components nere taken, with the exception of iron, and made to react in the usual viay. The initial zinc concentration n-as 7 x 10-3M and the final pH was determined t o be 3.5. Approximately 0.4% of the initial zinc was found in the “solid phase.” Thus, the tracer is normally soluble under the experimental conditions in the absence of the carrier. DISTRIBUTION LAWS
The general shape of the curves obtained in the studies with aluminum and yttrium (cf. Figures 5 and 6) suggests the possibility that these systems might conform to either the homogeneous or logarithmic distribution lair (22). The
Table VI.
80 70
-
c3
w
%
k ‘z 0:
a
-
INITIAL ZINC CONCENTRATION
60-
40
N
IO
r c
1
0
IO
20
40
30
conventional forms of these equations are shown in Table VII. I n both equations T denotes tracer and C denotes carrier. These equations can be dcrived for the simple case where the tracer and carrier form very similar compounds-e.g., radium sulfate and barium sulfate-in a manner similar to that shon-n by Gordon ( 5 ) . If the dw-
Coprecipitation of Zinc with Ferric Periodate
Initial Zinc Concentration 1.4 X l O - 5 X 1 55 1.65 1.7 1.7 2.0 2.7 3.2
.
4.96 43.2 45.6 70.5 94.2 99.75 99.99
1.0 6.5 7.2 12.3 24.3 61.2 82.9
0.20 0.11 0.11 0,088 0.054 0.029 0.031
0.26 0.17 0.18 0.17 0.20 0.48 0.83
Initial Zinc Concentration 4 X 10-4111 189 250 300 360 470 500 780 1640 12960
1.5 1.6 1.65 1.7 1.85 1.9 2.15 2.7 3.7
190 475 705 1525 12960
1.55 1.9 2.0 2.1 3.6
1.83 13.8 39.7 47.9 70.2 79.9 97.24 99.71 0
0.48 1.1 2.6 3.3 6.2 6.9 11.2 22.1 i0.5
0.26 0,073 0.048 0.046 0.042 0.032 0.012 0.0057
0.37 0.10 0.074 0,075 0,084 0.075 0.074 0. i27
b
Initial Zinc Concentration 3.5 X 10-3111 8c 93“ 96 99e (I
Iron could not be detected in filtrate. X o t calculable.
Analysis based on precipitate analysis only.
44
ANALYTICAL CHEMISTRY
60
70
80
90
I00
Figure 7. Coprecipitation of zinc
A”’
195 240 315 330 520 1310 2780
50
PERCENT IRON PRECIPITATED
0.40 2.7 4.4 7.2 6.5
0,031 0.0051 0.0056 0,036 6
0 078 0.021 0.030 0.041 b
ivation is adapted to other systems (3), the mathematical form remains essentially unchanged for the aluminum or yttrium tracers with iron carrier. The expressions obtained in the derivation are not much affected by the fact that simple one-to-one compounds are not used, so long as all the cations involved are of the same valence. On the other hand, if the expressions are derived using a divalent cation such as zinc as the tracer with a trivalent cation such as iron, changes in the mathematical forms of the equations result. Table T’II s h o m the mathematical expressions for the constants of the various systems as derived in a manner siniilar to that cited above ( 5 ) . The appropriate equations of Table VI1 were used with the data of Tables 111, IV, and V I to calculate values of the distribution coefficients, D’, A’, D’”, and A”’. In fire experiments (cf. Tables I11 and VI) the amount of iron in the filtrate r a s so small that it could not be detected; thus, it was not possible to calculate the distribution coefficients in these experiments. However, in all the other cases, the calculated values (3) of the distribution coefficients (especially the D values) varied with fraction of carrier precipitated. A consideration of the experimental errors shom-ed that these could not account for the variations observed in the coefficients. However, the variations might possibly be due, though it does not appear likely, to the unknovn chemical nature of the coprecipitating impurity. The distribution expressions for aluminum and yttrium were derived 11y assuming that the impurities coprecipitated as A15(10& and Y6(IOa)3,
The quantities, log [Zn+2]so~ution and Table
VII.
Forms“ of Distribution Laws (3)
log
-4luminum or
Conventional Yttrium* Honiogeneons Distribution
Zinc
Logarithmic Distribution
Square brackets denote solution concentrations. 2’ denotes tracer and C denotes carrier. i and f refer to initial and final solution concentrations. In the present iron-zinc system, where the initial concentration of iron is constant, the above equations were modified to use the data as given in Table VI. B,‘, = D,,[Fe+++2 1 , 3 = 7 ‘ Zn pptd. X (fraction Fe in s o l u t i ~ n ) ~ ’ ~ I Fe pptd. X (fraction Zn in solution) [Fe-+-]%1/3= log __ (fraction Zn in solution) A”’ = [(fraction iron in solution)’ - 11 D‘ and A ’ for present periodate systems are equal numerically to D1’sand 1’’s(3). The conventional homogeneous distribution equation is that used by Henderson and Iiracek (10). The conventional logarithmic distribution equation is that derived by Doerner and Hoskins (1). Equations used for aluminum, yttrium, and zinc were derived a s described (S, 5 ) .
respectively. Although these compounds have not been reported in the literature, the formulas were nevertheless arbitrarily selected to conform t o the iron compound. I n the case of zinc, Zn(IOs)* was considered to be the coprecipitated species because this compound has been reported in the literature (20). HOWever, the modified coefficients, as noted in Table VI1 by D”’ and A’”, were not a t all constant when calculated using the data of Table TI. iilthough there appeared to be no theoretical grounds for doing so, the data of the zinc study were used with the conventional form of the distribution equations to determine whether either coeffi-
Table VIII.
cient would remain constant. Again no constancy of the distribution coefficients could be observed: The variation was in fact greater. I n any event, the fact that the position of the coprecipitation curves (cf. Figure 7 ) shifted Kith the initial zinc concentration was also contrary to the theoretical expectations of the equations. I n an effort to interpret the results of zinc coprecipitation study, calculations were made to see if the data would fit the Freundlich equation. The form of the equation used was:
r/20 1111. 18 51s 4500 18 518 4500 18 518 4500
Fe
Pptd., %b 50 50 50 70 70 70 90 90 90
Zinc Pptd.,O
120g[Zn-+I
lo
7.5 3.4 1.7 11.8 6.0 2.2 20.7 8.3 2.7
-3.206 -2.107 -1.469 -3.176 -2.006 -1.503 -3.039 -1.975 -1.523
Cso~n.
1.220 2.699 3.646 1.201 2.687 3.644 1.553 2.677 3.641
Value interpolated from Figure 7. calculated as: sozinc precipitated X initial zinc concentration % iron DreciDitated X initial iron concentration solid - [Zn++].,l,. calculated as: [Initial zinc concentration x (1 - fraction zinc precipitated)]
(E)
solid
were calculated as in-
dicated in Table VIII. Khen plotted, the data show a straight-line relationship for a relatively large change in initial zinc concentration. The range of the initial zinc concentration is such that it varies by a factor of approximately 225. Because the solution concentration of zinc as well as the quantity of the solid phase is continually changing throughout the course of a precipitation leaction, it is somewhat surprising that the data follow the simple form of the Freundlich expression. I n contrast to the observations with the trivalent tracers, the relative coprec7ipitation of zinc appears to be somewhat greater prior to the 99% cairier preripitated point than is the case nith aluminum or yttrium. The extent, of coprecipitation also varies with the initial zinc concentration. The data appear to follow the Freundlich isotherm t o a moderate degree for large variations in solution concentrations of zinc ion. The process whereby zinc coprecipitates with ferric periodate appears more complex than in the case of the trivalent tracers. ACKNOWLEDGMENT
This work was supported in part by the Atomic Energy Commission. The authors wish t o thank B. P. Burtt of Syracuse University for his very helpful advice concerning the radiochemical aspects of the investiga t’ion. LITERATURE CITED
(1) Doerner, H., Hoskins, W., J . A m . Chem. SOC.47, 662 (1925). (2) Fortune, W. B., Mellon, M. G., IND. ESG. CHEY., ASAL. ED. 10, 60 (1938). ( 3 ) Ginsburg, L., Ph.D. dissertation, Syracuse University, 1955. (4) Gordon, L., ASAL. CHEM.24, 459 (1952). (5) Zbid., 27, 1704 (1955). (6) Gordon, L., Peterson, J. I., Burtt, B. P., Zbzd., 27, 1770 (1955). (7) Gordon, L., Reimer, C. C., Burtt, B, P., Zbzd., 26, 842 (1954). (8) Gordon, L., Rowlev, K., Zbid., in
Calculations for Freundlich Plot of Data of Zinc Coprecipitation Study
Initial Zinc Concn.,
):(
press.
(9) Gordon, L., Teicher, H., Burtt, B. P., Zbid., 26, 992 (1954). !lo) Henderson, L., Kracek, F., J . , 4 n ~ Chenz. SOC.49, 738 (1927). (11) Kall, H., Ph.D. dissertation, Syracuse University, 1955. (12) Nassner, J. L., Ozier, AT. A, ANAL. CHEW23, 1453 (1951). 113) Kimmins. 111. A,.’ J . Chem. SOC.5 5 . 148 (1889). (14) Kolthoff, I. M., Sandell, , E . B., .
I
“Textbook of Quantitative Inorganic Analysis,” 3rd ed., p. 321, hlacmillan, New York, 1952. (15) Nehlig, J. P., Hulett, H. R., IND. ENG.CHEX., ANAL. ED. 14, 869 (1912). (16) Raychoudbury, P. C., J . Indian Chem. SOC.16, 269 (1939). VOL. 29, NO. 1, JANUARY 1957
45
(17) Ibid., 18, 576 (1941). (18) Salutsky, RI. L., Kirby, H. W., AN~L. CHEW27, 567 (1955). (19) Salutsky, h l . L., Stites, J. G., Martin, Ibid., 2 5 , 1677 (1953). A. W., (20) Smith, G. F., “Periodic Acid and Iodic Acid and Their Salts,” p. 5, G. F. Smith Chemical Co., Columbus, Ohio, 1950.
private communica(21) Stoenner, R. W,, tion. (22) Wahl, 4.C., Bonner, N. A., “Radio. activity Applied to Chemistry,” DD. 106 ff.. Wilev, Sew York, 1951. (23) JTeaver, B., ANAL. CHEX. 26, 477 (1954). (24) Ibid., p. 479. (25) Willard, H. H., Ibid.,22, 1372 (1950). -
(26) Willard, H. H., hlerritt, L. L., Dean, J. A., “1,ytrumental hfethods of Analysis, 2nd ed., p. 182, Van Xostrand, New York, 1951.
1
RECEIVED for review January 13, 1956. Accepted August 30, 1956. Division of Analytical Chemistry, 128th Meeting, ACS, Minneapolis, Ninn,, September, 1955
Separation of Iron from Aluminum by Precipitation of Ferric Periodate from Homogeneous Solution LEONARD GINSBURG, KAY MILLAR, and LOUIS GORDON Department o f Chemistry, Syracuse University, Syracuse 7 0, N. Y Iron(ll1) can b e quantitatively precipitated from homogeneous solution as the periodate in very dilute nitric or sulfuric acid. The precipitate is dense, easily filtered and washed, and can be quantitatively converted to ferric oxide by slow ignition. Iron can b e separated from aluminum by the method described, more effectively in nitric than in sulfuric acid. By a single precipitation in nitric acid, 85 mg. of iron can b e separated from 10 mg. of aluminum; with twostage precipitation, as much as 100 rng. of aluminum can be present. IRON PRECIPITATES have poor analytical properties because they are voluminous and difficult to filter and wash. One notable euception is basic ferric fo1,mate precipitated by the urea method of Willard and Sheldon (4). These investigators proposed the use of this dense precipitate for the removal of iron preparatory to the determination of bivalent metals. The separation of iron from aluminum by this method was not satisfactory ( 2 ) . Gordon and Ginsburg (1)have shoIvn that ferric periodate can be precipitated from homogeneous solution. Aluminum alone does not precipitate under the conditions usrd to precipitate iron as the periodate. Thcse procedures ( f ) have been modified for the separation of iron and aluminum. The present study describes experimental results obtained by precipitating iron as the periodate in the presence of aluminum and determining the efficiency of the separation by determining aluminum in the precipitate and iron in the filtrate.
MAX’
REAGENTS A N D APPARATUS
Unless otherwise specified, all cheniicals were reagent grade. Ferric Sulfate Solution. Approximately 8 grams of electrolytic iron (Standard Sample Co., Ames, Iowa) was dissolved in about 100 ml. of 46
0
ANALYTICAL CHEMISTRY
1 to 1 hydrochloric acid b y heating. The resulting solution was evaporated t o dense fumes with sulfuric acid and the residue was taken u p with distilled water. This solution, free of chloride and ferrous ion, was filtered and diluted to approximately 6 liters. The resulting solution was standardized by precipitation of hydrous ferric oxide, followed by ignition to the oxide. The solution mas found to contain 35.4 mg. of iron per 25 ml. Ferric Nitrate Solution. Approximately 17 grams of the electrolytic iron was dissolved in approximately 100 ml. of hot 1 t o 4 nitric acid. The resulting solution was boiled t o expel ovides of nitrogen. After filtration, the solution was diluted t o approximately 2 liters and standardized as above. The ferric nitrate solution was found t o contain 85.4 mg. of iron per 10 ml. All other solutions were prepared as previously described (1). All pH measurements were made on a Beckman Model H-2 DHmeter.
copious evolution of nitrogen during the reduction of the periodate. The solution was cooled and then analyzed for iron (1). Aluminum in Precipitates. Aluminum was determined in the ferric periodate and in the ignited residues of ferric oxide. For the determination of aluminum in ferric periodate concentrated hydrochloric acid was added t o the filtering crucible containing the precipitate, which readily dissolves. Ignited residues had t o be warmed gently for several minutes with a few milliliters of concentrated hydrochloric acid to effect solution. In both cases the resulting solution was diluted to volume in a 50-ml. volumetric flask and then a suitable aliquot was withdrawn for aluminum determination Usually 20 ml. of the hydrochloric acid solution was taken and neutralized to a pH of approximately 1. Aluminum was then determined (1). PRECIPITATION OF IRON I N SULFURIC ACID SOLUTION
ANALYTICAL METHODS
I r o n in Filtrates.
The filtrates and washings obtained from each precipitation experiment were evaporated t o about 50 ml. The resulting solution was acidified with a few milliliters of sulfuric acid and then treated cautiously with 85% hydrazine solution, which was added dropiyisc until the iodine color produced n’as completely discharged. The hydrazine m a added slowly because of the
Table I.
Find PH
1.9
1.95 2.05 2.1 2.2
Procedure. I n a n unscratched 400ml. beaker place a solution containing no more than 50 mg. of iron(II1) and 5 ml. of 1 to 1 sulfuric acid. Add 2.5 grams of paraperiodic acid. Dilute t o about 100 ml., dissolve all solids, add 9.5 grams of acetamide, dilute t h e solution t o 150 ml., and adjust t h e p H to 0.8. Cover the beaker n i t h a watch glass and maintain the solution a t 95” C. for approximately 3.5 hours. Then filter
Iron Remaining in Solution a t Various pH Values after Precipitation from Sulfuric Acid Solutions Iron in Filtrate Iron in Filtrate
and Washings,
Final
hfg.
PH
and Washings, Mg.
0.6
2.3 2.35 2.4 2.5 2.6
0.1,0.03,0.2 0.1 0.1 0.05, 0.07
