V O L U M E 25, NO. 5, M A Y 1 9 5 3 needed to complete the oxidation of methanol and whether large excesses of reagent would cause further oxidation of the formic acid produced during the reaction. Data from these experiments are given in Table 11. It is apparent that an appreciable excess of the reagent is desirable for the analysis and that an excess of 20 ml. of 0.1 iY ceric reagent does not adversely affect the results. Interferences. Most easily oxidized organic compounds such as aldehydes and ketones might be expected to interfere with this procedure. The higher aliphatic alcohols are also oxidized under the conditions used and further investigations of these reactions are now in progress. Formic and acetic acids are not oxidized under the conditions used. Results. Table I11 shows typical recoveries of methanol obtained by the proposed procedure. The methanol used for
823 these experiments was dried by distillation from a magnesium metholate solution, and considerable care was taken in making up the solutions for analysis to avoid losses of the alcohol by volatilization. The average per cent error of these results m s 0.47,. LITERATURE CITED
(1) Blott, A. H., and Gilman, Henry, “Organic Syntheses,” Collective Yol. 1, 2nd ed., p. 220, New York, John TViley 8i Sons, Inc., 1941. ( 2 ) Kolthoff, I. M,,and Sandell, E. B., ”Textbook of Quantitative Inorganic Analysis,” 3rd ed., pp. 593, 583, 475, New York,
3Iacmillan Co., 1952. RECEIVED for review April 15, 1952.
Accepted January 13, 1953.
Analytical Chemistry of Niobium and Tantalum Separation of Iron and Manganese f r o m the Earth Acids C . F. HISKEY AND A. L. BATIK1 D e p a r t m e n t of C h e m i s t r y , Polytechnic I n s t i t u t e of Brooklyn, Brooklyn 2, S. Y. REvIous
reports (1, 4 ) have described a chlorination procedure
Pfor the earth acids and other analytically related oxides using
octachloropropane as the chlorinating reagent and working a t atmospheric pressures. Quantitative distillation of titanium( IV) and tin(IS-) chlorides may be made from their mixtures with niobium(V) and tantalum(T’) chlorides. Before attempting chlorination, hon-ever, it is most important to separate iron. Failure to do so results in a catalytic decomposition of the reagent by the iron( 111) chloride formed before the remaining oxides can be chlorinated. hnother reason for eliminating the iron a t an early stage is its interference with the spectrophotometric determination of tantalum. I n 100% sulfuric acid a t a wave length of 285 mw, iron as the sulfate complex has an absorptivity 22 times greater than the peroxytantalate complex which peaks there (6). BIODIFIED SCHOELLER SCHEME
The Schoeller ( 7 ) scheme of analysis provides for iron and manganese renioval as the divalent sulfides from ammoniacal tartrate solution. This separation occurs, following precipitation of the major fraction of the earth oxides by acid hydrolysis from tartrate solution and prior to the subsequent precipitation with tannin of the minor fraction from a n ammoniacal tartrate medium. ThuP, two separate steps are required to gather the earth oxides for chlorination. I n an effort to eliminate one of these steps it was first thought that it might be possible to remove the iron and manganese as sulfides from ammoniacal tartrate. T o do this the pyrosulfate fusions of niobites or tantalites were leached into acid tartrate media and the iron was reduced by passing hydrogen sulfide into the solutions. .ifter the yellow color disappeared, the solutions were made ammoniacal and either hydrogen sulfide was bubbled in or ammonium sulfide, was added. The recovery of iron was satisfactory, but the amount of manganese that precipitated varied widely. T o illustrate the recovery of combined iron and manganese osides obtained in this way, data have been assembled in Table I for a group of synthetic oxide mixtures having the following approsimate conipoqition: iron plus manganese oxide, 20%; titania, 57,; zirconia, 1%; stannic oxide, 5%. The remaining i O % v a s divided between niobia and tantala, their ratios varying from 10 to 0.1. The results reported are totals for the iron and manganwe osides, which were weighed as Fe,Oa and lInaO,. 1 Present address, Repearch Analytical Laboratories, Nathieson Chemical Co.. Xianara Fall*. S . T.
Recovery of the combined oxides is good. Vnfortunately, however, this is due to the fact that the iron-manganese ratios in these samples were large. I n all cases they were in excess of 15 and sometimes as high as 30. Although no systematic quantitative analyses of these precipitates were made, it was established that when the percentages were low most of the loss was due to incomplete manganese precipitation.
Table I. Recovery of Iron and RIanganese by Sulfide Precipitation from Ammoniacal Tartrate Solution % Taken
% Found
Diff.
%Taken
%Found
20.92 21.57 20.49 20.65 20.50
20.48 21.01 19.70 20.73 20.38
-0.44 -0.56 -0.78 tO.08 -0.12
20.82 25.29 20.61 21.28
21.06 25.23 19.86 21.32
Diff.
+o.
12 -0.06
-0.76
$0.04
It is concluded that moderate amounts of iron may be quantitatively removed in niobite and tantalite mineral analyqis, permitting the subsequent chlorination of the remaining earth oxides. Manganese recovery in such a case is likely to be very faulty and its determination had best be ignored if this separation is used. The remaining manganese will be recovered for the moqt part in the filtrate from the earth oxide precipitation step. A trace ail1 be found with the earth oxides. It represents only a very small fraction of 1% of the earth oxides and so will not interfere either in the chlorination step or in subqequent colorimetric reactions contemplated for the niobium and tantalum determinations. It cannot, of course, follow the titanium because this element is distilled out as a volatile chloride. Procedure. A half gram of the fincly powdered minrral(80- to 100-mesh) is fused with 20 times its weight of potassium pyrosulfate, first a t Ion. heat and then st high heat until a clwr melt results. After cooling, the cakr is tapped out of the crucible into about 200 ml. of a 10% solution of ammoniacal tartrate. The remaining melt is leached out by immersing it in the same solution. The tartrate is then heated to boiling to speed the dissolution of the melt. After the crucible has tieen rinsed and removed, the solution is made ahout 0.1 JT with hydrochloric acid-Le., 10 ml. of 2 -11acid are added. Hydrogen sulfide is now bubbled in to reduce the iron to the ferrous state. If the ferric sulfide or hrdrouide is precipitated in the ammoniacal solution, large amounts of the earth oxides (Loprecipitate. Stannous sulfide precipitating at this point is now filtered out. T o the filtrate remaining, excess of ammonium hydroxide is added along 1% ith ammonium sulfide to precipitnte the iron and the manganese
ANALYTICAL CHEMISTRY
824
sulfides. These are filtered off, leaving the earth oxides for subsequent separation and chlorination. CHELATION OF IRON AND MANGANESE
An alternative and somewhat niore satisfactory approach involves a precipitation of the earth oxides while retaining the iron and manganese in solution. The operations attending the separation of the iron and manganese, if conditions warrant, may be avoided altogether. Particularly with richer ores and minerals, the principal values are in the earth oxide fraction and thus the analysis of that fraction may be the only one of interest. Agents like ethylenediaminetetraacetic acid complex strongly both the oxidized and reduced states of iron ( 2 , 3,6). It seemed desirable to apply such complesing agents to this problem as well, a n d what follows is a description of some experimental findings in this connection. Materials Used. Ethylenedianiinetetraacetic acid (Versene, Sequestrene). il5$, by weight solution of the disodium salt was prepared and used unless otherivise noted. Chel 242 and Chel 153. These are commercial preparations of the Slrose Chemical Co., Providence, R. I., and bear their trade name. They are chelating agents of the imino diacetic type b u t are not otherwise described. Chel 242 was received in liquid form as a 43yo by neight solution and was used as received. T h e Chel 153 came as a white powder and like the Versene was made as a 5y0 solution. rl synthetic oxide mixture was prepared consisting of 50% tantala, 33.5% niobia, and 16.570 titania. These oxides u-ere the high purity materials commonly used in these investigations. This particular mixture was chosen because it represents one of the more difficult ones to handle analytically. General Procedure. A 0.2-gram sample of the ovide mixture was fused with 2.5 grams of potassium pyrosuliate in a porcelain crucible. After a clear melt was obtained, the crucible was cooled sufficiently to allow the addition of a few milliliters of concentrated sulfuric and then rewarmed slightly to produce a clear liquid solution of the oxides. The amount of sulfuric added at this point should be kept to the minimum which will give a melt liquid a t room temperatures. After this was cooled, it was diluted with about 40 ml. of ice-cold distilled water t o a clear solution. At this point known amounts of standard solution of either (or both) manganese(I1) sulfate or iron(I1) ammonium sulfate were added t o complete the mixture of elements that was to be separated. The chelating agents m-ere added in appropriate amounts and finally enough guanidine carbonate to make the p H of the re-
sulting solution about 12. After a brief period of boiling to ensure completeness of earth oxide hydrolysis, the solution rvas filtered through a Whatman KO.40 paper and thoroughly nashed nith water to free it of potassium salts. The precipitate n a s then ignited and weighed. This allowed an estimation of the earth oxide recovery. T o establish the completeness of iron and manganese retention in the filtrate, the oxide precipitates were analyzed directly for these elements when the contamination wis present in trace amounts. Othern-ise it was merely determined by the difference between the weights of earth oxide taken and the final aeight of precipitate recovered. The procedure emploj ed consisted of a pyrosulfate fusion of the precipitate, follon-ed by leaching into dilute sulfuric. To determine iron the colorimetric thiocj-anate procedure was employed. For manganese, periodate oxidation to permanganate was employed. The permanganate was also determined colorimetrically. 4 blank correction x a s made for the trace quantity of iron present in the 2.5 grams of p j roeulfate used. EXPERIMENT4L RESULTS
Bfter it had been established that neither the guanidine caibonate nor any of the chelating agents to be used interfered with the quantitative hydrolysis and recovery of earth acids in the oxide mivture taken, experimental studies of the separation of iron and manganese nere started. The hydrolyzed oxide piecipitates obtained Tvith guanidine had much smaller flocs than the precipitates formed n i t h ammonia hydrolysis This tended to slow the filtration somen-hat, but did not otherwise aflect the analytical process. T‘arious quantities and various combinations of chelating agents were used to Eee how successful the retention of iron and manganese in the filtrate could be made. Some typical data ohtained for the individual ions are assembled in Table 11.
Table 111.
Separation of Iron and Manganese
(Reagent was 35 ml. of Versene, 20 m!. of a 5% solution of Chel 153, and 1 mi. of Chel 242 as supplied in about 250 ml. of solution) Gram of Contaminant as Fen01 1InaOl
....
0.0203 0.0203 0.0203 0.0203
....
.... .... ....
.... .... ....
.... .... 0.0206 0.0206 0.0206 0.0206
E a r t h Oxide Mixture, Gram
Fen03 or h h o a in P p t . , 11g.
0.1062 0.0938 0.0626 0.0404 0.000 0.1019 0.0603 0.0389 0.000
0.1 0.2 0.18 0.19 nil 0.3 0.3 0.3 nil
Table 11. Separation of Iron and Manganese from Earth Oxides (Weight of iron and manganese present in these samples was sufficient t o give 0.0203 gram of FelOs and 0.0198 gram of lInsOn, respectively) Oxide E a r t h Oxides Precipitate Contaminant Chelating Agent Taken, Recovered, Present, RI1. Gram Gram Gram
_-
Iron Contamination 12 6 35 40 8 35 8 4 4 35 6 1
35
Chel 153 Chel 242 Versene Versene Chel 153 Versene Chel 153 Chel 242 Chel 242 Versene Chel 153 Chel 242 Yersene
0.2043 0.2151 0.1944 0.1939 0.2365
0.2236 0.2339 0.2138 0.2134 0.2532
0.0193 0.0188 0.0194 0.0195 0.0167
0.2034
0.2051
0.0017
0.2065
0.2080
0.0004
0.1970
0.1969
0.00017
Manganese Contamination 12 6 35 40 8 35 8 4 4 35 6
1 35
Chel 153 Che! 242 T’ersene Versene Chel 153 Versene Chel I53 Chel 242 Chel 242 T‘ersene Chel 153 Chel 242 T’ersene
0.1991 0.2044 0.2049 0,2025 0.1982
0.2189 0 2242 0 2243 0 2217 0.2013
0.0198 0.0198 0.0194 0.0192 0 0031
0.2174
0.2356
0.0182
0.2169
0.2139
0.0003
0 1982
0.1Di9
0.0003
Inspection of these data reveals first of all that the individual chelating agents are almost without effect on the separative process. This result is most surprising. When the chelating agents were mixed together, however, their effectiveness was enhanced markedlyand themixture containing all three of them gavequantitative retention for both the iron and the manganese. I n this mixture the precipitate contained less than a 0.1% of iron and about the same amount of manganese, although they were about 10% of the precipitate initially. Although it was surprising to observe the failure of the individual reagents to retain the iron and manganese in solution, this only testifies to the great adsorptive powers of these hydrolyzed oxides. Having found a combination of agents n-hich were useful in this separation, the authors decided to vary the percentages of the contaminants. Accordingly, this variable was studied and some data obtained are given in Table 111. Here it, can he seen that the effectiveness of the coniplexing mixture is independent of the quantity of earth oxides mixed nith the fixed amount of the iron or manganese contaminants. IVliile the iron and manganese have thus varied from 0 to 1007, in these sample,?, the total amount has not varied. Presumahly if larger amounts of iron or iiiarigaiiese are taken, the amount of the chelating mixture niu.Gt lie increased proportionately. - 4 the ~
V O L U M E 2 5 , NO. 5, M A Y 1 9 5 3 complexation constants for these elements ale enormous, no large excess of unreacted chelating agent ia required for the reaction to go to completion. The excess reagent probably functions here b y displacing absorbed complexes into solution, thus achieving the excellent Qepnrationobserved. EFFECT OF TIY AYD ZIRCO\IURI
I n separate qtudies the effect of tin and zirconium on this qeparation n as investigated. The results of those experiments are considered important because these elements are commonly a'sociated with the earth oxide minerals, although usually in small amountq. K h e n added to the solution prior to the hydiolysis step, neither of these elemerits interfered x i t h the letention of the iron and manganese b j the chelating agents. Moreover, the tin apparenty a a - retained quantitatively in the filtrate I n any separation that might be planned, it would thus go n i t h the iron and manganeqe and not appear in the titanium distillate. Zirconium, on the other hand, divided in thi. operation. The amount going with the iron and manganese vai ied considerably, but ucually was betTveen 60 and 907,. I n concentrates where ziiconia is a major component, this separation cannot be used without further investigation. On the other hand, the zirconia is often a very minor constituent, and its determination is then not too important. It is easily recovered from the filtrate by making
a25
the solution aliout 6 JI n i t h sulfuric acid and then precipitating it with phosphoric acid. ACK3-OK LEDG3f E S T
The authors ~ o u l dlike to acknon-ledge the help of Russell H. Atkinson for some direct experimental assistance in the course of this investigation and the advice of Joseph Steigman and Sorman Adler on several important points. LITERATURE CITED
(1) dtkinson, R. H., Steigmati, J., and Hiskey, C. F., ;1s.i~. CHEX. 24, 480, 484 (1952). (2) Biederman, IT,, and Scha-arzenbach, G., H e b . Chim. Acta, 3 1 , 4 5 9 11948). (3) Britzinger, H., Thile, H., and JIuller, U., 2. anorg. allgem. Chenr., 251, 285 (1943). (4) Hiskey, C. F., Sewman, L., and .Itkinson, R. H., ANAL.C H m r . , 24, 1988 (1952). ( 5 ) Jones, S. S.. and Long. F. .I., J . Phvs. Chem.. 56, 25 (1952). (6) Palilla, Frank C.. 11,s.thesis, Polytechnic Institute of Brooklyn, Iiovember 195 1. (7) Schoeller, TT. R., "Analytical Chemistry of Tantalum and Xiobiurn," London, Chapman and Hali, Ltd., 1937. RECEIVED for review Septemher 9 , 1952. Accepted January 27, 1953. Experimentai d a t a taken from thesis submitted b y A. L. Batik t o the chemistry faculty of the Polytechnic Institute of Brooklyn as part of the requirements for the 31,s.degree in chemistry. June 1 9 j 2 .
Thermometer Calibration for Determination of Capillary Melting Points STANLEY C. BUXCE Rensselaer Polytechnic Institute, Troy, 4'.1.. HEmiomwERs
used in the determination of capillary melting
Tpoints may be calibrated n i t h a series of pure substances of hnonn melting points. Lists of compounds suggested for use as standards for calibration which are found in Iahoratory textbooks are, however, generally unsatisfactorv. I n most cases the compounds chosen are poor standards because they are difficult to purify or store, or because the melting points of the pure compounds are not accurately knonn. I n the work reported here, a survey a a s made to find compounds TThich are suitable standards. I n order to test their suitability as standards, a number of these compounds v ere purified by recrystallization, their purity TTas confirmed by cooling curve measurements, and they were employed in calibrating thermometers under the conditions for use in melting point determinations.
Calibrations may be extended to higher temperatures by the use of reference melting points with pure metals. PURIFICATIOY OF REFERENCE CORIPOUh-DS
Six compounds were chosen from the list in Table I, and saniplea were purified, as indicated in Table 11. I n each case, the purification procedures were readilL- accomplished; crj-stallizations were conducted, in general, so as to obtain maximum purity a t the expense of efficient recover>- of material.
Table I.
Compounds L-seful for Melting Point Standards
Compound
Melting Point,
C.
COMPOUNDS FOR B'IELTIR-G POIhT STAYDARDS
A list of compounds Those melting points are suitable standurds for thermometer calibration is presented in Table I. These vere chosen to meet, as closely as possible, t n o criteria. They may be purified without excessive labor or access to special apparatus, and the melting points of the pure compounds determined by a t least tx-o investigators check to within 0.2" C. Only compounds for which a t least one of the measurements has been made by thermocouple or resistance thermometer are included. The list could probably be extended to include more compounds in the lower melting point range, but the number of compounds n hose melting points have been accurately measured by several investigators is surprisingly limited. Burriel-3Iarti undertook to prepare a list of compounds suitable for calibration of Anschutz thermometers and found it necessary to determine melting points for each of his purified compounds (3). His n-ork, which appears to have been not sufficiently recognized, has formed the basis fora part of the present list; the original work and the summary of data on pure compounds by Timmermans ( 2 5 ) have also been useful. It would be desirable to have confirming data on the melting points of the highest melting compounds listed in Table I.
(21, 48.1
(fa
(21)
Benzoic acid Urea hlannitol Succinic acid p-Kitrobenzoic acid Anthraquinone
132.8 (IO)
The purit,y of each of the final samples was checked by determining its cooling curve (8. 2 2 ) . The apparatus used need not be refined, and the thermometer need not be calibrated in order t o determine that a sample has a constant freezing point during the first third of the solidification time. I n all cases, the temperature during t,he initial portion of the plateau as found to vary less than 0.2" C. indicating that the compounds n-ere sufficiently pure that their initial freezing points n-ere within 0.2" C. of the true freezing point. Since measurements of capillary melting points can be made reproducible only within this limit unless great care is used, this vas considered satisfactory.
