tained with formaldehyde. H o m v e r , the formate reaction is considerably slower than the reaction with formaldehyde. LITERATURE CITED
(1) Criegee, R., Lohaus, G., Ber. 84, 219
(1951).
( 2 ) Desnuelle,
I'., Saudet, ll., Bull. soc. china. Frunce 12, 871 (1945). (3) Feigl, F., "Spot Tests," T'ol. 11, p. 255, Elsevier, h e w York, 1954. (4)Johnson, G. R. A , , Scholes, G., Weiss, J.. J . Chem. SOC.1953, 3091. (5) Metzler, D. E., Snell, E. E., J . Am. Chetn. Soc. 74, 979 (195%). ( 6 ) Sinehnm, .\. K., Cheni. /?PI'S. 5 5 , 355 (1955 I .
( 7 ) Schryver, S.B., Proc. Roy. Soc. (Lond o n ) 82B, 226 (1910). (8) Tanenbsum, M., Bricker, C. E., ,%SAL. CHEJI.23, 354-7 (1951 1. (9) 1-on Pechman, H., Ber. 2 5 , 31i5 (1892). (10) Ibid., 27, 3'20 (1894).
RECEIVED for revieiv Xarch 26, 1958. .\ccepted September 23, 1958.
Gravimetric Determination of Zirconium in Titanium J. H. HILL and M. J. MILES Titanium Metals Corp. o f America, Henderson, Nev.
b Mandelic acid is a r a p i d and accurate reagent for determining zirconium in titanium alloys without interference from many metals. Zirconium tetramandelate can b e precipitated quantitatively from either hydrochloric o r perchloric acid solutions. Iron, aluminum, vanadium, tin, copper, chromium, cobalt, magnesium, manganese, molybdenum, and nickel do not interfere when present in the amounts usually found in titanium alloys. Hafnium and niobium interfere, causing high results.
I
use of zirconiuni as a constituent of titanium alloys ha? iiwessitated a fast and accurate method of determining zirconium in samples containing a large proportion of tltaniuni. Methods using cupferron, selenious acid, phosphate, and phenylarsonic acid have been used for the determination of zirconium in ores and in steel ( 2 ) . Titanium reacts with each of these reagents and interferes extensively n ith the determination of zirconium by these methods. Kumins ( 7 ) discovered that mandelic acid is aliiiost a specific reagent for t h r quantitative precipitation of zirconium and hafnium. Other 11orkers (3-10) have studied the 'determination of zirconium with mandelic acid and its derivatives. Some of this work has show ii that the zirconium precipitates formed with p-chloromandelic and p-bromoniandelic acids are superior to zirconium tetraniandelate because they h a w higher molecular n rights, do not require a mandelate nash solution, and can be weighed directly. However, mandelic acid was chosen bccausc i t is readily available and somewhat Ion er in cost than the chloro and bromo dcrimtives. The chloro and bromo derivatives are so similar to mandelic acid in the reactions involved that they SCRCASED
252
0
ANALYTICAL CHEMISTRY
can probably be substituted for mandelic acid in the folloning procedure. Kumins ( 7 ) and Hahn (3-5) indicated that titanium and such common alloying constituents as iron, aluminum, vanadium, tin, copper, chromium, cobalt, magnesium, manganese, and nickel do not interfere with the determination of zirconium by precipitation with mandelic acid. Their interference studies were not repeated here, except that vanadium n a s used in larger quantities than used by Kumins to more closely approximate the composition of the common allo! 5 . Kuiiiins found that large quantities of sulfate interfere. causing lon results. Hydrochloric and perchloric acid solutions have been recommended in the literature as the best media for precipitating zirconipm mandelate. Although titanium in a 1 to 1 ratio did not interfere ?I ith the zirconium mandelate method, it did 11hen present in larger amounts. This interference was probably caused by hydrolysis of titanium(1V) during digestion of the zirconium mandelate and by occlusion of titanium in the precipitate. This interference may be overcome by dissolving the initial precipitate in 1 4 ammonium hydroxide, filtering to remove titanium hydroxide, and reprecipitating from acid solution. Because titanium(II1) does not hydrolyze as readily as titaniuni(IV), precipitation from solutions of titanium(II1) is recommended. Also, the color of the trivalent ion serves as a convenient indicator to shorn when the precipitate is adequately TT ashed after the initial filtration. T o overcome the interference from titanium. Suss ( 9 , 10) converted the initial zirconium mandelate precipitate to zirconium hydroxide using sodium hydroxide, dissolved the zirconium hydroxide in hydrochloric acid, filtered the acid solution to remove insoluble impurities, and precipitated the zirconium mandelate. The presence of
+
mandelate ion throughout this purification process requires delicate adjustment. of the hydrochloric acid concentration to dissolve the zirconium hydroxide wit'hout simultaneously reprecipitating zirconium mandelate. Any zirconium Inandelate precipitated upon the addition of hydrochloric acid would remain with the impurities filtered off in the next st'ep, causing 1017 results. Purification n-it'h amnionium hydroxide is preferable because zirconium mandelate dissolves in ammonium hydroside iyithout forming zirconium hydroxide. The Suss met'hod is limited to tit'anium metal and alloys containing from 0.05 t'o 10% zirconiuni and no provision is niadc for overcoming t,he interferences of sulfat'e and fluoride. Titanium ores and residues generally require fusion with potassium pyrosulfate. High zirconium alloys frequently require trentinent with hydrofluoric acid. The method described here provides a means for oi.ercoming interferences from sulfate arid fluoride so that ores, residues, and high-zirconium alloys can be analyzed. METHOD
Reagents. Mandelic Acid, 16% solution. Dissolve 160 grams of mandelic acid in 1 liter of water. Mandelic Acid K a s h Solution. Dissolve 2 grams of mandelic acid in 100 ml. of 1 9 hydrochloric acid. Amnionium Hydroxide Wash Solution. Dilute conceiit'ratecl ammonium 4 11-ith n-ater. hydroxide 1 Titanium Metal, zirconium-free sponge. Zirconium Xetal, U. S. Bureau of Mines, more than 99.5% zirconium. Procedure. Dissolve a 5.000-gram sample in 300 ml. of concentrated hydrochloric acid. Samples containing 507, or more of zirconium may not dissolve completely; in t h a t case, use a mixture of perchloric and hydrofluoric acids and evaporate to fumes of perchloric acid to drive off the hydrofluoric acid. Transfer a n aliquot containing from 0.2 t,o 0.15 grain of zir-
+
+
conium to a 25O-ni1. beaker and add 50 ml. of 16% mandelic acid solution. Add 20 ml. of concentrated hydrochloric acid and dilute to 100 to 125 ml. Digest at 80" to 85" C. for a t least 20 minutes. Samples n-ith sniall amounts of zirconium should digest 1 hour or more (8). Filter through No. 42 Whatman paper. Kasli TT it11 mandelic acid wash solution 5 or 6 times, or until tlie color of titanium(II1j disappears. Puncture the cone of the filter paper and wash the bulk of the precipitate into a clean beaker with n-ater, taking care to ivash behind the fold in the paper. Rinse the original beaker and wash the paper four or five times with 1 3 ammonium hydroyide solution to remove all traces of the precipitate. Stir the ammoniacal solution to dissolve the precipitate and add 4 or 5 nil. of concentrated animonium hydroxide if necessary. Filter into the original beaker, using S o . 42 K h a t m a n paper. Rinse the paper three or four times ITith ammonium hydroxide wash solution to remove the last traces of the sample. Add 25 nil. of 16% mandelic acid solution and a few drops of methyl red indicator. Add concentrated hydrochloric acid until the solution is acid, then 10 nil. in excess. Digest at 80" to 85" C. for a t least 20 minutes. Filter through a Selas crucible of medium porosity. Wash with mandelic acid n-ash solution. D r y a t 110" C. for 15 to 20 minutes. Place the Selas crucible in an unglazed protective crucible and transfer t o a iiiuiAe furnace cooled to about 200" C. Heat to 1000" C. to ignite tlie precipitate. Weigh as zirconium dioxide.
+
DISCUSSION
I n the abrciice of rc.liable staiidard zirconium-titaaiuni alloys, synthetic alloy solutions nere used to test the method. These qolutions were prepared from spectrographically analyzed zirconiuin metal of high purity obtained from the U.S. Bureau of 1Iines and from zirconium-free titanium metal. Solutions prepared b y dissolving the zirconium metal in a mixture of perchloric and hydrofluoric acids and fuming to remove hydrofluoric acid gave consistently reproducible results. When perchloric acid was used to dissolve tlie zirconium metal, the mandelate had to be precipitated froin titanium(1Vj solutions. The interference of vanadium, manganese, and inagiiesiuni was checked and t h a t of molybdenum was investigated using aliquots of perchloric acid solutions of zirconium added t o hydrochloric acid solutions of zirconiumfree titanium(1Vj. Measured volumes of standard solutions of vanadium, manganese, magnesium, and molybdenum were added to the zirconium-titanium solutions to give the desired concentration of the interference to be studied. The results are shoivn in Table I. Xetallic standards u-ere prepared b y melting carefully weighed amounts of
Table I.
Effect of Common Alloying Elements
Zr Added, yo Zr Found, yo Recovery, Synthetic 1 T i 92% 7 98 7 95 09 G T 99 100 1 8 00 100 2 A\* 7 98 100 0
9 91 c) 99 9 99 I v . 9 97
o9 100 100 99
Analysis of Synthetic Metallic Standards
Zr Found, Recovery,
yo
Synthetic 2 . Ti 7 0 7 , V 5 5 , 110 See, X n 10c,'c, l f g 0 scC 9 98
Table II.
6 1 1 9
Compocition, c> 1.00 Zr, 1 0 0 110, 98 Ti
102 10% 99 101
3.00 Zr, 1.00 310, 96 Ti
3.07 3.10 3 14 -1v. 3 . 10
102 103 105 103
2.00 Zr:98Ti
2 05 2 03 2 03 A\-. 2.04
10% 101 101 102
1111 10yc, 1lg 0.5C& 4.04 4 06 4 03 AV. 4 04
101 2 ioi 7 101 0 101 3
Synthetic 4. Ti 7 8 7 , T' 5%, 110 5 9 , 1 995
32n lo%,, M g 0.5y0 1 985 99 5 2 015 101 0 2 020 101 3 hv. 2 , 0 0 5 100,5
SA-nthetic 5 . Ti is",, 1- 5c,;, 110 5Yc, hln lorc, hIg 0.5% 1 000
101 0 101 0 100 5 100.8
1010 1 010 1 005 Av. 1.008
titanium. molybdcniun, and zircoiiium into buttons, n-liich w r e turned over and remelted t n ice to achieve uniformity. Shavings from tlie buttons were dissolved in hydrochloric acid and analyzed by this procedure. Results of these analyses, slio~in in Table 11, indicate the reliability of the method. dlloys high in zirconium ivere dissolved in hydrochloric acid and analyzed Table Ill.
/C
02 02 99 01
Synthetic 3. T i 76C;$. V 5%, 110 5 7 , 3.99
5
0
1 1 0 -%v. 1
by this procedure. Proximate analyses were 100 =k 0.5% in alloys cont'aining over 50% zirconium. It is oft'en necessary t'o fuse titanium ores and residues n-ith potassium pyrosulfate before they can be dissolved for analysis. A zirconium solution 11-3s prepared b y fusing a sample of tlie zirconium metal in the minimum amount of potassium pyrosulfate and dissolving the fusion in hydrochloric acid. Aliquots of t'lie zirconium solution were addcd t'o a hydrochloric acid solution of zirconium-free titanium and analyzed b y this procedure. The results are s h o ~ v nunder saniples 1, 2 , and 3 in Tablc 111. Fusions with larger amounts of pyrosulfnte produced solutions which g a ~ rlow and erratic rcsults causcd by interference of the excess sulfate. Therefore, a n attempt' )vas made to separate the zirconiuni from the sulfatc prior to analysis. Aliquots of a solution containing approximately 10 grams of potassium pyrosulfate per gram of zirconiuni were added to hydro-
Zirconium in Presence of Titanium
Ti
Sample
Piesent, GI ams
Taken
Zr, Gram
1
1 00
0 0894
2
1 00
0 0447
3
2 00
0.0224
4
1.00
0.0954
Rerovered
Av
Av.
Av.
Av.
0 0897 0 0894 0 0899 0 0899 0 0897 0 0446 0 0446 0 0448 0 0443 0 0446 0.0221 0.0214 0.0224 0.0212 0.0218 0.0948 0.0950 0,0951 0.0946 0.0949
Recovery, yo 3 0 6 6 3 99 8 QD 8 100 2 oo 1 99 8 98.7 95.5 100,o 04.G 97.3 99.4 99.6 99.7 99.2 99.5
100 100 700 100 100
VOL. 31, NO. 2, FEBRUARY 1959
253
chloric acid solutions of zirconium-free titanium. The titanium and zirconium were separated from the sulfate with an ammonium hydroxide precipitation (2). The combined hydroxides were collected on filter paper, redissolved with hydrochloric acid, and analyzed. The results are shown under sample 4 of Table 111. The precision of the mandelic acid method is almost equal to the precision of the analytical balance used to weigh the zirconium dioxide residue. Based on the U. S. Bureau of Mines analysis of the zirconium used in preparing the synthetic alloys, the accuracy of the method is comparable to the precision. No interference was encountered from such common alloying constituents as iron, aluminum, vanadium, tin, chromium, manganese, and molybdenum. The reports of Kumins and Hahn indicate that copper, cobalt, magnesium, and nickel do not interfere in the amounts usually found in titanium alloys. Hafnium precipitates quantitatively with the zirconium, but these elements are so similar in their properties that no effort was made to dis-
tinguish between them. According to Hahn (I),both the hafnium and the zirconium may be determined with p-bromomandelic acid by solving simultaneous equations. Because the reactions of p-bromomandelic acid with zirconium are so similar to those of mandelic acid, p-bromomandelic acid could undoubtedly be used in this procedure. Niobium interferes, causing high results. Excess sulfate, fluoride, citrate, oxalate, tartrate, and other complcxing agents interfere, causing low and erratic results but with the exception of sulfate and fluoride, are unimportant because they would not be encountered in ordinary analyses. Fluoride may be removed by evaporating to fumes with perchloric acid. Interference from sulfate can be eliminated by precipitating the Zirconium as zirconium hydroxide with ammonium hydroxide and washing free of sulfate prior to precipitation as the mandelate.
appreciation to the U. S. Bureau of Mines for providing the high purity zirconium metal used to prepare the standards, and R. L. Powell, Titanium Xetals Corp. of America, who supervised this research. REFERENCES
( 1 ) Fritz, J. S., Johnson, Marlene, ANAL. CHEM.27, 1653 ( 1 9 5 9 . (2) Furman, N. H., Scott’s Standard hlethods of Chemical Analvsia.” 5th ed., Vol. 1, pp. 1095-104, Van Xostrand, New York, 1925. (3) Hahn, R. C., AXAL. CHEW 21, 1579 (1949). (4) Ibid., 23, 1259 (1951). (5) Hahn, R. C., Webber, Leon, J . Am. Chem. SOC.77, 4777 (1956). ( 6 ) Klingenberg, J. J., Papucei, R. A . ANAL.CHEM.24, 1861 (1952). (7) Kumina, C. A., Ibid., 19,376 (1949). ’
(8) Mills, E. C , Herman, S. E., Analyst
78.256 (1953). (9) &E, ‘ Henry, private communication, August 1958. (10) Suss, Henry, Sam Tour Rept. KO 9141, Sam Tour & Co., 41 Trinity PI., Yew York, N. Y., November 1951.
ACKNOWLEDGMENT
The authors wish to express their
RECEIVED for review February 21, 1957 Accepted September 22, 1958.
Fire Assay for Platinum and Palladium in Ores and Concentrates M. E. V. PLUMMER, C.
L. LEWIS,’
and
F.
E. BEAMISH
Division o f Analytical Research, University of Toronto, Toronto 5, Canada
b This investigation was undertaken partly to provide a t least one alternative procedure to the one generally applicable method for the determination of platinum metals in ores and concentrates. The new fire assay involves a collection of platinum and palladium by fusions with sodium carbonate a t 1450’ C. in a carbon crucible to produce an alloy with ironnickel-copper. The method of parting the iron button and subsequent isolation of platinum and palladium by cation exchange are described.
M
of the platinum metals available to commerce originate from ores which contain about 0.001% of the precious metals. Of the six platinum metals, platinum and palladium practically ah-ays predominate. With most ores osmium values are not known and usually ruthenium and iridium values are known only approximately. For OST
1 Present address, Falconbridge Xickel Mines Ltd., Richvale, Ontario.
254
e
ANALYTICAL CHEMISTRY
these more insoluble metals the values are generally calculated on the basis of the commercial recovery. Analytical determinations in ores and concentrates involve a fire assay in which the platinum metals are concentrated b y solvent extraction by lead from a molten solution of the ores with fluxes. Rarely may the assager resort to an entirely wet treatment of the ores or concentrates to convert the platinum metals to dissolved constituents of familiar compositions. The principle of this procedure is good, but no data have been recorded to indicate its efficiency relative to recovery by fire assay. Furthermore, the procedure is time-consuming. A week is usually required and accuracy of recovery of microgram amounts of platinum metals is not encouraged by the multiplicity of extractions, relatively large volumes of liquids entailed, and difficult techniques involved. Fire assay with lead as the collector remains the time-honored method. It is surprising that few data have been recorded to prove its efficiency. This is
partly due to the absence of alternative procedures; the wet procedure may have proved too difficult. However, an examination of the chemistry of the lead extraction from the fused ore introduces some doubt concerning the possibility of complete extraction of some of the platinum metals (1, 2, 6, 9, l a ) . Among the difficulties is the insolubility of some of the platinum metals in fused lead. Iridium resists dissolution to the degree that a good quantitative method for this metal is based upon this property (10). Furthermore. the platinum metals in ores may exist in forms which resist reduction by carbon and subsequent extraction by lead. This question, used too frequently by those who would promote valueless oIes, plagues the conscience of the platinum metal analyst. The fact that it cannot be disregarded is shoJm by the adverse influence of tellurium in the fire assay for gold. The result of these and other problems has been H continued search for a new approach to the assay for plat-
