ANALYTICAL CHEMISTRY
174 for the preparation of the synthetic samples of zirconium oxide used to test the procedure. LITERATURE CITED (1) Abelson, P. H., Phys. Rev., 56, 1 (1939). (2) Arrol, W. J., Research, 2 , 253 (1949). (3) Boldridge, W.F., and Hume, 0. N., U. S. Atomic Energy Commission, Rept. AECD-2552-E (1949). (4) Borst, L. B., Physics Today, 4 , 6 (1950). (6) Boyd, G. E., ANAL. CHEM.,21, 335 (1949). (6) Brown, H., and Goldberg, E., Science, 109, 347 (1949). (7) Cohn, W. E., ANAL.CHEM.,20,498 (1948). (8) Cook, C. S., and Langer, L. M., Phys. Rev., 73, 1149 (1948). (9) Cook, G. B., U. S.Atomic Energy Commission, Rept. A E R E C/R-424 (1949). (10) der Mateosian, E., Goldhaber, M., Muehlhause, C. O., and McKeown, M., Phys. Rev., 72, 1271 (1947). (11) Edwards. F. C.. and Voiat. A.. ANAL.CHEM.,21, 1204 (1949). (12) Goldberg, E. D‘., and Brown, H., Ibid., 22, 308 (1950). (13) Gordon, C. L., Ibid., 21, 96 (1949). (14) Grummitt, W., and Wilkinson, G., Canadian Atomic Energy Commission, Repl. MC-167 (1945). (15) Hahn, O., and Strassmann, F., Natt.rwissenschaffen, 31, 499 (1943). (16) Hens, F., Z . Anorg. Chem., 37, 31 (1903).
(17) Hevesy, G., and Levi, H., Kg1. Danske Videnskab. Selskab, Math.-Fys. Medd., 14, 5, (1936); 15, 11 (1938). (18) Kern, B. D., Zaffarano, D. J., and Mitchell, A. C. G., Phy.9. Rev., 73, 1142 (1948). (19) Livingood, J. J., and Seaborg, G. T., Ibid., 55, 414 (1939). (20) . . Maxwell. R. D.. Raymond. H. R., and Garrison. W-.M.. J. Chem. Phys., 17, 1340 (1949). (21) Miller. L. C.. and Curtiss. L. F.. Phws. Rev.. 70. 983 (1946). (22j Myers; R. J.; Metzler, D.’ E., and Swift, E.’H.,’J. Am. Chem. SOC.,72,3767 (1950). (23) Newton, A. S.,Phys. Rev., 75, 17 (1949). (24) Reid, A. F.. and Weil. A. S..U. S.Atomic Energy -. Commission, Rept. AECD-2324 (1948). (25) Seaborg, G. T., and Livingood, J. J., J . Am. Chem. SOC.,60, 1784 (1938). (26) Seaborg, G. T., Wilson, S.G.. Wilson, V. C., and Coryell, C. D., U. S. Atomic Energy Commission, Rept. MDDC-763 (1946). (27) Seren, L., Friedlander, H. N., and Turkel, S.H., Phys. Rev., 72, 888 (1947). (28) Tobias, C. A., and Dunn, R. W., Science, 109, 109 (1949). (29) Way, K., Fano, L., Scott, M. R., and Thew, K., Natl. Bur. Standards, C ~ T C499 . (1950). (30) White, J. R., and Cameron, A. E., Phys. Rev., 74, 991 (1948). (31) Willard, H. H., and Diehl, H., “Advanced Quantitative Analysis,” pp. 348-9, New York, D. Van Nostrand & Co., 1943. RECEIVED May 14, 1951.
Analysis of Mixtures of Dicobalt Octacarbonyl and Cobalt Carbonyl Anion e
HEINZ W. STERNBERG, IRVING WENDER, AND MILTON ORCHIN Bureau of Mines, Bruceton, Pa.
A study of the hydroformyIation (oxo) process in this laboratory required the analysis of mixtures containing both dicobalt octacarbonyl, [Co(CO),],, and cobalt carbonyl anion, [Co(CO)4]-, the latter being present either as cobalt hydrocarbonyl, HCo(CO)4, or as a cobalt salt. Because no method was available for this purpose, the authors devised the present procedure. The two-step method depends on the decomposition of all carbonyls, by treatment with iodine, to yield carbon monoxide; and the selective precipitation of the anion by nickel o-phenanthroline chloride, followed by decomposition of the precipitate with liberation of carbon monoxide. As the determination is based on a gasometric procedure, it is suited for the analysis of mixtures, such as those from the hydroformylation reaction, which may contain cobalt salts and metallic cobalt.
M
OST of the mechanisms that have been proposed (1, 9)
for the hydroformylation (oxo) reaction involve the postulation that the homogeneous catalyst is either dicobalt octacarbonyl or cobalt hydrocarbonyl, but no method for the determination of these carbonyls in the reaction products has yet been devised. The dark solution of aldehydes, ketones, and alcohols that is removed from the reaction vessel often contains cobalt not only as dicobalt octacarbonyl, cobalt hydrocarbonyl, cobalt salts of the hydrocarbonyl, and cobalt salts of organic acids, but also as finely divided metal (attracted by a magnet). I n the course of an investigation in this laboratory, it became desirable to devise a simple and rapid method for the accurate determination of dicobalt octacarbonyl, [ C O ( C O ) ~and ] ~ , cobalt hydrocarbonyl, HCo(CO)d, in the mixtures usually obtained in the hydroforniylation reaction. I n a study of these carbonyls ( 3 ,4 ) it was shown that dicobalt octacarbonyl is quantitatively decomposed by bromine in glacial acetic acid or iodine in benzene according to Equation 1 (8).
[Co(CO),]z
+ 212
--f
2C012
+ 8Cb
(1)
Although the decomposition of the free cobalt hydrocarbonyl by halogens is not described in the literature, Hieber and Teller (6) mention that excess iodine in benzene decomposes cadmium cobalt carbonyl, Cd [ C O ( C O ) ~ ]quantitatively ~, into the corresponding metal halides and carbon monoxide. An aqueous solution of nickel o-phenanthroline chloride has been used by Hieber and coworkers for the quantitative determination of cobalt hydrocarbonyl in aqueous ( 5 )and ammoniacal solution (4). It should be emphasized that this reagent is not specific for the acid, H C O ( C O ) ~ but , rather for the cobalt carbonyl anion, [Co(CO),]-. OUTLINE OF METHOD
The present method is based on the following facts: Dicobalt octacarbonyl, cobalt carbonyl anion, and cobalt hydrocarbonyl are quantitatively decomposed by excess iodine according to Equations 1 , 2 , and 3.
+ 312 +2CoI2 4-8 3 0 + 212HCo(CO)r + 312 2CoL + 8 C 0 + 2HI 2[ Co(Co)4]-
--f
(2) (3)
V O L U M E 2 4 , N O , I, J A N U A R Y 1 9 5 2
175
The amount of carbon monoxide evolved serves as a measure of the total amount of carbon:-l (dicobalt octacarbonyl and cobalt carbohyl aninn) present. Sickel o-phenanthroline cation does not react with dicobalt octacarbonyl. With cobalt carbonyl anion, however, a salt is formed according to Equation 4.
+ [Iii(o-Phthr),]
2[ C O ( C @ )-~ ]
+++
[Iii(o-Phthr)~] [Co(CO),]p (4)
This salt is insoluble in water and in hydrocarbon solvent such as hexane or toluene and may, therefore, be separated from dicobalt octacarbonyl, which is readily soluble in these solvents. The freshly precipitated nickel o-phenanthroline salt of cobalt hydrocarbonyl is quantitatively decomposed by excess iodine according to Equation 5.
+
+
+
[Ni(o-Phthr)s][Co(C0)4]~312 = [Si(o-Phthr)3]Iz 2 Co12 8c0 (5) The amqunt of carbon monoxide evolved serves as a measure of the amount of cobalt carbonyl anion present. The difference between the amount of cobalt carbonyl anion and the total amount of carbony! (dicobalt octacarbonyl and cobalt carbonyl anion) is equivalent to the amount of dicobalt octacarbonyl present. REAGENTS AND APPARATUS
Iodine Solution. Add water to 300 grams of potassium iodide and 50 grams of iodine to make 1000 ml. of solution. Take 40 ml. of this solution for samples containing 0.0 to 0.6 gram of total carbonyl. Nickel o-Phenanthroline Chloride Solution ( 4 ) . Dissolve 2.8 grams of o-phenanthroline monohydrate and 1.2 grams of nickel chloride hexahydrate in 100 ml. of water. Use 25 ml. of this solution for samples containing 0.00 to 0.25 gram of cobalt carbonyl anion. Apparatus. The apparatus used has been described (8). The gas buret has a capacity of 450 ml. The reaction vessel is a 125-ml. Erlenmeyer flask ( 7 ) fitted with a ground-glass joint and a side arm provided with a stopcock for the addition of the iodine solution. PROCEDURE
Pipet a suitable aliquot of the solution to be analyzed into the Erlenmeyer flask, add enough toluene to bring the volume of the solution to about 40 ml., attach the flask to the apparatus, and urge the system R-ith nitrogen or (preferably) carbon monoxide. Ktir the solution 1 to 2 minutes, and record the volume of gas. Add 40 ml. of the iodine solution through the side arm, and stir the solution vigorously until no more gas is evolved (usually 10 to 15 minutes). Record the final volume. Pipet a second aliquot into a glass-stoppered Erlenmeyer flask containing 25 ml. of the nickel o-phenanthroline chloride solution and 25 ml. of benzene. Shake the flask vigorously for about 1 minute; add about 30 ml. of water and 1 ml. of 6 -V hydrochloric acid solution; shake again, and allow to stand for about 1 hour. Filter with suction through a rapid filter paper; wash with benzene and finally with water. Dissolve the precipitate in about 40 ml. of pyridine and transfer the solution to the reaction vessel. Attach the reaction vessel to the apparatus, purge the system with nitrogen (or carbon monoxide), and decompose the pyridine solution with 40 ml. of iodine solution as described above. Calculate the weights of total carbonyl or of precipitated cobalt carbonyl anion as follows:
Table I.
Analyses of Blends
% ~Co(CO)rlz Sample 1 9
3
4
6
7
Calcd. 20.2 20.2 55.4 58.4 71.3 73.7 84.9
% [Co(C0)4]Calcd. 79.8 79.8 44.6 41.6 28.7 26.3 15.1
Found 20.6 18.6 56.4 59.2 70.4 72.8 83.4
Found 79.4 81.4 43.6 40.8 29.6 27.2 16.6
in an atmosphere of ox? gen for 3 hours, a dark precipitate formed, and gas corresponding to 40% of the available amount of carbon monoxide was evolved, consisting of carbon monoxide and carbon dioxide in a ratio of 4 to 1. When a benzene solution of dicobalt octacarbonyl was stirred in an atmosphere of nitrogen, 12y0 of the available carbon monoxide was evolved in 5 hours. An aqueous solution of known cobalt carbonyl anion content prepared from dicobalt octacarbonyl and ammonia ( 4 ) was stored for several days in an atmosphere of nitrogen without change. On exposure to air, however, the solution decomposed within a few minutes lvith discoloration and evolution of gas. The solution of iodine in aqueous potassium iodide reacts smoothly and quantitatively. This reagent is easily made up in any desired concentration, presents no handling problem, possesses excellent storage properties, and thus has obvious advantages over a solution of bromine in glacial acetic acid ( 3 ) . That cobalt hydrocarbonyl reacts with iodine according to Equation 3 to give HI and not Hz according to Equation 6 2HCo(CO)d
+ 212
--f
H,
+ 2 CoIp + 8 c 0
(6)
was shown by the following experiment. A freshly prepared sample ( 2 ) of approximately 0.2 gram of cobalt hydrocarbonyl dissolved in 20 ml. of toluene was decomposed with excess iodine. Mass spectrometric analysis of the evolved gas showed that carbon monoxide and hydrogen were present in a ratio of 300 to 1. If the decomposition had followed Equation 6, the ratio of carbon monoxide to hydrogen should have been 8 to 1. The traces of hydrogen probably originated from some decomposition of the cobalt hydrocarbonyl (Equation 7 ) prior to the reaction with iodine.
+
~ H C O ( C O ) ~ S H ~ [CO(CO),],
(7)
Although cobalt carbonyl anion may be determined gravimetrically ( 4 ) by weighing the precipitated nickel o-phenanthroline salt of cobalt hydrocarbonyl (Equation 4),the authors have found that this salt can be quantitatively decomposed according to Equation 5. This reaction makes possible the rapid and accurate determination of cobalt carbonyl anion by a gasometric method. It is important to carry out the decomposition with the freshly precipitated salt. Analyses of precipitates that had been allowed to stand overnight in a vacuum desiccator gave only 80 to 90y0 of the theoretical amount of carbon monoxide. The nickel o-phenanthroline salt of cobalt hydrocarbonyl often separates as an oil that is easily emulsified. Acidification with hydrochloric acid, however, causes the formation of a crystalline precipitate that can be readily flltered. ANALYTICAL RESULTS
In order to test the present method, solutions of known amounts ml. of gas evolved ( N T P ) X 0.171 Grams of C O ( C @ )=~ 89.6 DISCUSSION OF METHOD
The importance of protecting dicobalt octacarbonyl and cobalt carbonyl anion from prolonged exposure to oxygen is illustrated by the following experiments. When a benzene solution of dicobalt octacarbonyl was stirred
of dicobalt octacarbonyl and cobalt carbonyl anion ( 4 ) were prepared. The blends had a total carbonyl content of about 0.6 gram per 10 ml. and varied in cobalt carbonyl anion content from 79.8 to 13.9y0. The results of the analyses of these blends are listed in Table I. The method is well suited for the determination of small amounts of carbonyl and may readily be adapted to micro procedures because 89.6 m]. of carbon monoxide are evolved by 1 millimole (0.171 gram) of carbonyl.
ANALYTICAL CHEMISTRY
176 I n general, the strict precautions observed in working with carbon monoxide should be practiced in working with the cobalt carbonyls. Dicobalt octacarbonyl has a low vapor pressure and there is little danger of inhaling its vapor. Cobalt hydrocarbonyl has a high vapor pressure and may be very toxic, but, in the absence of an atmosphere of carbon monoxide, it starts to decompose well below room temperature into hydrogen and dicobalt octacarbonyl. The apparatus used in the analytical determinations was placed in a large well ventilated room with an exit tube leading to the hood for venting carbon monoxide. The hydrocarbonyl was prepared in a hood. ACKNOWLEDGMENT
The authors wish to thank Jack Sharkey and R. A. Friedel for the gas analyses and Sol Metlin for assistance in preparation of dicobalt octacarbonyl.
LITERATURE CITED
Adkins, H., and Krsek, G., J. Am. Chem. Soc., 70, 383 (1948). Blanchard, A.A., and Gilmont, P., Ibid., 62, 1192 (1940). Hieber, W.,Muhlbauer, F., and Ehmann, E. A., Ber., 65, 1090 (1934).
Hieber, W., and Schulten, H., 2.anorg. u. allgem. Chem., 232, 17 (1937). Ibid., 243,145 (1939). Hieber, W., and Teller, U., Ibid., 249, 43 (1942). Koller, C.R.,and Barusch, M.R., TND. ENG.CHEM.,ANAL.ED., 14,907 (1942). Orchin, M., and Wender, I.,A N ~ LCHEY., . 21, 875 (1949). This apparatus is now available from H. Tarmy Co., 1035 Watson St., Pittsburgh 19, Pa. Wender, I., and Orchin, hl., Bur. Mines, Rept. Invest. 4270 (1948). RECEIVED June 5 , 1951.
Nonorthophosphate Contaminant of NeutronIrradiated Rock Phosphates Procedures f o r Its Removal A . J. MAcKENZIE AND J. W. BORLAND Bureau of Plant I n d u s t r y , Soils, a n d Agricultural Engineering, U . S . D e p a r t m e n t of Agriculture, Beltsville, Md. Orthophosphate compounds, upon irradiation with slow (thermal) neutrons, yield a radioactive product, the P32content of which is only partly in the orthophosphate condition. The compounds are therefore not suited for tracer experiments without treatment to convert all the P32to the orthophosphate form. Calcium phosphates tested showed a considerable amount of nonorthophosphate P32,depending partially upon conditions of irradiation. A method is presented for determining nonorthophosphate P32in
S
LOW (thermal) neutron irradiation of orthophosphate com-
pounds, no matter how pure they may be, yields a radioactive product, the P3*content of which is only partly in the orthophosphate condition. The nonorthophosphate, presumably produced as a consequence of the recoil of the P31 nucleus from gamma-ray emission as it passes to the ground state, has been identified as phosphite in irradiated alkali phosphates ( 4 , 6 , 8 )and as hypophosphate in irradiated calcium phosphates ( 3 ) . The possible presence of P32 in more than one form in irradiated samples necessitates a careful examination of any irradiated phosphate that is to be used as a source of radioactive phosphorus in tracer experiments. The basic requirement for a proper label is that all the P32in the labeling compound be in, or readily convertible to, the same form as the phosphorus compound to be labeled. Recent work in this laboratory ( 9 ) points to pyrophosphate as the dominant form of nonorthophosphate in the irradiated potassium dihydrogen phosphate distributed by the Oak Ridge S a tional Laboratory-the Pa2 source used extensively in late years by this bureau for labeling phosphate fertilizers for greenhouse and field experiments (6). Moreover, it was found that the labeling processes provide conditions favorable to the reconversion of the nonorthophosphate, so that the labeled fertilizers carry an insignificant amount of it. The same end could probably be attained in any labeling process involving synthesis of a compound. On the other hand, the introduction of Pa2 into a natural material, such a8 phosphate rock, which cannot be synthesized in the laboratory, must be accomplished by direct irradiation of the
calcium phosphates. A treatment is shown that will result in an irradiated rock phosphate suitable for plant investigations. This work should impress users of neutron-irradiated compounds with the frequency of occurrence of this phenomenon. The presence of more than one radioactive isotope, if not noted and eliminated by conversion or compensation, invalidates tracer experiments. A n important fertilization material, rock phosphate, can now be used in tracer experiments.
material. As the irradiated rock must be utilized directly as the tracer source, a nonorthophosphate content is a serious drawback (4).Phosphate rock is used extensively (more than 800,000 short tons in 1949) for direct application to the soil, and thus considerable interest attaches to the measurement of its effectiveness as a source of phosphorus in the growing of vegetation with the aid of the radioactive tracer technique. Looking to this end, a study was made of the amount and distribution of nonorthophosphate in irradiated phosphate rock, and of treatments that might be used to destroy the nonorthophosphate without seriously altering the properties of the rock. Results of this study are reported here. PHOSPHORUS SPECIFIC ACTlVlTY AND NONORTHOPHOSPHATE CONTENT OF SOME IRRADlATED CALCIUM PHOSPHATES
In the spring of 1950 eight calcium phosphates and a sample of potassium acid phosphate m-ere irradiated in the neutron pile a t Oak Ridge under various conditions of neutron flux and temperature. Samples of each of the nine materials Tere exposed to a high neutron flux with an intermediate (water-cooled) temperature, a low neutron flux with a low temperature, or a high neutron flux with a high temperature. These samples, after irradiation, were returned to the laboratory and phosphorus specific activity and nonorthophosphate content were determined (Table I). It can be seen from Table I that a great part of the phosphorua activity in neutron-irradiated calcium phosphates may exist as a nonorthophosphate contaminant, The nonorthophosphate content in each case was higher in the calcium phosphates than in