1047
V O L U M E 2 6 , NO. 6, J U N E 1 9 5 4 methylene chloride but not in tribenzylamine solvated in carbon tetrachloride or benzene and its homologs. These methods for the separation of niobium from tantalum should prove valuable t o those interested in a rapid separation of the two metals. Of the systems reported, the separation of the two metals in the hydrochloric acid system is much more desirable for tantalum does not readily extract from any hydrochloric acid concentration. Technologically, these amine systems may be applied successfully t o the industrial separation of niobium from tantalum with cheap chemicals and replace the present tedious fractionation methods. ACKNOWLEDGMENT
The authors wish to thank P. Z. TTesterdahl and L. hl. Frakes for performing many of the experiments in obtaining the data.
LITERATURE CITED
Bush, 11. T., and Densen, P. &I., ANAL.CHEY.,20, 121 (1948). Kiehl, S.J., Fox, R. L., and Hardt, H. B., J . Am. Chem. Soc., 59, 2395 (1937). Leddicotte, G. W., and Moore, F. L., Ibid., 74, 1618 (1952). hIoore, F. L., U. S. Atomic Energy Commission, “USAEC Secret Report,” ORNL 1314 (July 1952). Rosenheim, A., and Roehrick, E., 2. anorg. Chem., 204, 342 (1932). Ruff. 0.. , and ~ - Schiller. E.. Ihid..72. 329 11911). , , Russ, R., Ihid., 31,~42 (1902). Sidgmick, N. V., “The Chemical Elements and Their Compounds,” Oxford, Clarendon Press, 1950. Smith. E. L.. and Page. J. E., J . Soc. Chem. I n d . , 67, 48 (1948). Way, K., et al., Natl-Bur. Standards, CUT. 499 (1950). Weinland, R. F., and Storz, L., Ber., 39, 3056 (1906). Weinland, R. F., and Storz, L., 2. anorg. Chem., 54, 223 (1907). ~
~~
,
I
RECEIVED for review April 22, 1953. Accepted April
9, 1954.
Fluoride Determination by Electron Transfer Catalysis W.
D. ARMSTRONG
and
LEON SINGER
Department of Physiological Chemistry, M e d i c a l School, University
The marked catalytic effect of fluoride on the cerousceric electron exchange reaction suggested that this reaction could be used as a microanalytical method for fluoride determination without the necessity of removing ions which interfere in present methods. Cations in general were found to exert negligible catalytic effects. The catalytic effect of fluoride, was about 900 times that of sulfate and about 370 times that of phosphate. These circumstances limit the application of the method to those samples which do not contain these ions or in which the effect of the interfering ions can be made constant. The fluoride content of bone ash has been determined by making constant the effect of interfering ions. The results of electron transfer catalysis by various negative ions are considered to be in qualitative agreement with Libby’s description of the mechanism of electron transfer catalysis.
H
ORNIG and Libby (1) discovered a marked catalytic effect
of fluoride ions on the rate of electron transfer between cerous and ceric ions in nitric acid solution and showed that the increase in electron transfer rate was proportional t o the fluoride concentration. A proposed mechanism (1, 2 ) of electron transfer catalysis describes the placing of a small negative ion (in this case fluoride) between the positive exchanging ions, so as to form a symmetrical transition complex in which the hydration atmospheres of the positive ions are shared. This model of electron exchange catalysis suggested the possibility that the effect of fluoride on this type of electron transfer might be so specific as to permit the determination of fluoride in natural materials, without the necessity of separation of fluoride from other ions in the sample. This report presents a comparison of the catalytic effect of several ions commonly present in natural materials with that of fluoride, and indicates the limitations of fluoride analysis based upon this principle. EXPERIMENTAL
The stock solution of radioactive cerium-144 was treated with hydrogen peroxide in order t o convert the ion t o the cerous state. The excess hydrogen peroxide was removed by heating the solution on a water bath. Concentrations which would give approximately 75,000 counts per minute of the radioisotope, under conditions of radioactive counting used in this method, were employed in each reaction mixture.
of
Minnesota, Minneapolis, M i n n .
The reactions were carried out in 50-ml. centrifuge tubes suspended in a large, stirred bath of ethylene glycol, which was placed in a household-type deep-freeze cabinet. The temperature of the bath was regulated a t -14.2’ C. and rose no more than 0.4’ over an 8-hour working period. i2t room temperature 3 ml. of 0.003M cerous nitrate hexahydrate solution in 12M nitric acid, labeled with cerium-144, were transferred t o the reaction tubes and mixed with 3 ml. of an aqueous solution of the ion t o be tested. After a temperature equilibrium had been established, usually by standing overnight, the reaction was started by the addition of 3 ml. of nonradioactive cerium(1V) solution (0.0003M ceric ammonium nitrate in 6 M nitric acid) a t - 14.2’ C. The reaction was stopped after 160 minutes and the cerium(ITr) separated from the reaction mixture by rapid and forceful injection of 20 ml. of cold ether into the reaction mixture with a syringe. Ten minutes later, 5 ml. of the ether layer containing the cerium( IV) were removed with a cold pipet and transferred to a 25-ml. volumetric flask. The ether was removed by aeration and the flask diluted t o volume with a solution containing 2 mg. per liter of nonradioactive cerium(1V). This solution acted as a carrier and diluent to impede possible adsorption of the radioactive ceric ions on container walls. The flask was allowed t o stand a t least 175 minutes (usually overnight) in order t o permit the praseodymium-144 daughter t o grow t o a concentration in equilibrium with that of cerium-144. The contents of the flask were transferred to paraffined drinking cups and the radioactivity was measured ( 3 ) . It is presumed that under these conditions only the energetic beta radiation from praseodymium-144 decay * was measured. A41dehydesand peroxides in untreated ether would reduce ceric ion during its extraction from the reaction mixture. The ether was satisfactorily purified in volumes sufficient for 5 days’ use by washing with ceric ammonium nitrate in approximately 8M pitric acid until the color of the ceric ions remained in the washings. The ether was then washed eight times with water, or until the washings were neutral, and stored in a dark bottle in the deep-freeze cabinet. In agreement with Hornig and Libby ( I ) , the dependable and rapid separations of the ether and aqueous phases, in the extraction of cerium(1V) from nitric acid solutions, were obtained with difficulty. The conditions specified were found by trial-anderror experiments, in which the relative volumes of ether and acid solutions were varied, to be optimal. Phase separation could not be obtained a t the low temperature employed when 15 ml. of ether and 9 ml. of acid were used, or when the acid concentration was increased t o 8.7M. RESULTS AND DISCUSSION
The results given in the figures are shown as the per cent of tho exchange that has not yet occurred, plotted against time or con
ANALYTICAL CHEMISTRY
1048
ccntration of the ion whose catalytic effect was being tested. The values were obtained from mixtures which contained a large excess of fluoride (1 mg.) and which had stood several hours at room temperature allowing the exchange reaction to reach completion. The half time of the uncatalyzed exchange reaction was found to he approximately 390 minutes (Figure l ) , as compared with 220 to 224 minutes observed by Hornig and Libby ( 1 ) . Since approximately 75% of the uncatalyzed exchange reactioii has not yet occurred a t 160 minutes, this time was the period in nhich the romparative catalytic effects of fluoride and other ions wew examined.
I
I
2b d.1 Figure 2.
I .
0:2
0:3 Millimoles
Ol4
0:5
d6
Effect of Certain Ions on CerousCeric Electron Exchange
Amounts on abscissa given as millimoles per 9 ml. of reaction mixture. These values to be multiplied by 10-3 for fluoride
503
50
100 Time
I50
200
250
300
11minutes
Figure 1. Uncatalyzed Cerous-Ceric Electron Exchange 611 HYOs. 3 X 10-1.W. Ce(1II) = CeW)
In Figure 1 the y intercept is a t 100% indicating zero exchange at zero time. This circumstance can be obtained only when a correction is applied to the results to allow for the apparent presence of radioactive cerium(II1) ions in the ether phase, I n 10 experiments, in which only cerium(II1) was employed, and in which the volume changes due to mutual solubilities of ether and nitric acid solutions were observed and taken into account, 0.57 i 0.17% of the radioactivity was found in the supernatant ether phase. This quantity represents cerium(II1) in the nitric acid solution trapped in the ether, and may include also cerous ions oxidized to the ceric state. Presumably the graphical correction for zero time exchange, applied by Hornig and Libby ( I ) , is of the same type. In the results given in Figures 2 and 3, this correction has not been applied, since comparative results only arc the items of interest. The results in Table I show that a number of ions commonly present in natural materials, in amounts in excess of those likely to be present in practical samples, do not alter the mciasured rate
of electron transfer. Also 54 mg. (2.3 millimoles) of sodium, added as sodium nitrate, did not alter the catalytic effect of 1 and 2 y of fluoride. Silver was included in these experiments, owing t o the anticipated use of silver nitrate to remove protein and chloride from blood plasma. Figure 2 presenk a comparison of the effects of some ions which were found to alter the measured rate of electron transfer. The
*OI
501
Ions Exhibiting No Effect on Electron Transfer between Cerous and Ceric Ions
Table I.
Fe+'+
Compound Used SaNOg KNOa -\Ig!(SO3)2 F e ( S 0 i ) a.QH?O
Ag
.hgN03
Ion Na a
K+ 1\Ig+'
I-
-
Maximum Amount
of Ion Tested, hlg.
54.0 0.06 2.0a 6.0
6.4
Sa1 0.5 Temp. -14.2' C ; time 1 6 0 minutes; Hh-03 concn. 6M; total roluine R nil. a A definite acceleration of the electron exchange was produced.by 10 m g . of magnesium.
Figure 3. Interference between Phosphate and Fluoride in Cerous-Ceric Electron Transfer A.
Fluoride alone
B. Calculated effect of fluoride in presence of 1 . 7 mg.
phosphorus (as phosphate) obtained by assuming that catalysis by each increment bf fluoride is additive to that of Dhosuhate. and Droduces same effect on electron transfer 'as is'given in A C. Observed effect of fluoride in presence of 1.7 mg phosphorus (as phosphate)
V O L U M E 26, NO. 6, J U N E 1 9 5 4
1049
high order of catalytic effect of fluoride on this example of electron transfer, discovered by Hornig and Libby (I), was confirmed. The same workers stated that chloridc had a slight catalytic effect in the same direction as fluoride. However, in this work, chloride added as sodium chloride, in three trials over Tide ranges of concentration, produced an apparent slight negative catalytic effect. Also, an inhibition in measured catalytic effect of fluoride was obNerved in the presence of chloride. A detectable inhibition \yap noted a t 0.5mX chloride; at 1.Om.U chloride, the effect of 1 -1 of fluoride was reduced by i . 8 % , Calcium ion, whose slight cat,alytic effect decreases in increment with increasing amount, is (except for magnesium) the only cation for which definite evidencc of catalysis of the cerous-ceric electron transfer was obtained in this study. The effect of calcium and chloride on electron transfer could be controlled in practical fluoride analysis, by adjusting to a concentration above which their effects became nearly constant. The catalytic effect of 0.1 millimole of sulfate is of the order of 1/900th that of the same amount of fluoride. On this account, some materials would require removal of sulfate prior to fluoride determination b y electron transfer catalysis. The catalyt,ic effect of O.lmW phosphate is approximately 1/370th that of an equal amount of fluoride (Figure 2), and the relative efficiency of phosphate increases with increasing amounts of the ion. The measured catalytic effect of phosphate on electron transfer cannot be attributed t o fluoride contaminating the dibasic sodium phosphate uscd to supply the phosphate ion, bccause the observed effect wap not altered by repeated treatments of the salt with concentrated nitric acid, followed by removal of the acid by evaporation at water bath temperature. Evidence was obtained to indicate that the catalytic effects of phosphate and fluoride, when both are present, are not additive. The results given in Figure 3 show that the effects of 1 and 2 y of fluoride are augmented in the presence of phosphate ion equivalent to 1.7 mg. of phosphorus. The high catalytic effect of phosphate introduces a serious interference in the application of electron transfer catalysis to fluoride analysis, since many materials, particularly those of biological origin, may contain upward of several hundred times as much phosphate as fluoride. While fluoride and phosphate may be separated by chemical manipulations and distillation ( d ) , it. would not be advantageous, on account of cost, to use electron transfer catalysis for fluoride determinations in those cases in which phosphate must be eliminated from the sample. This procedure w-ould have found a justifiable, routine use only in the
hoped for, but unrealized, circumstance that phosphate and other ions do not alter appreciably the rate of electron transfer. Fluoride determinations by electron transfer catalysis can be made without removing the interfering ions, if the quantities and, henre, the effects of these ions can be made constant. This can easily be attained with calcified tissues, since the composition of the mineral part of a given type of ralrified tissue-for example, bone-is virtually fixed by nature. Table I1 s h o w the results of fluoride determinations in bovine bone ash and gives the results of recovery of added fluoride. In this xyork, 5-nig, samples of bone ash were used. The fluoride contents of these samples were obtained by reference to standard curves of electron transfer, with known amounts of fluoride in the presence of the quantities of phosphorus and calcium found in 5 mg. of bone ash. These quantities of calcium and phosphorus were actual1)- obtained hy U R P of a solution of the same bone ash from which the fluoride had been removed by repeated treatment with nitric acid in a platinum dish, followed by removal of thtx excess acid by heat.
Table 11.
Analysis of Bovine Bone by Electron Transfer Catalysis Bone S s h 10 2 20 t 0.17"
S o . of samples analyzed Fluoride found, y Fluoride recovered, y Fluoride. 7~ o . 0 ~ 0.003~ a Standard deviation of mean
*
+
Bone Ash 0.50 y Fluoride 8 2.7" 0.10" 0.52
ACKNOWLEDGMEKT
The authors thank I. 11.Kolthoff, who called their attention to the work of Hornig and Libby and who supplied helpful discussions LITER'ATURE CITED
Libby, W.F., J . Phys. C'hcm.. 56, 869 (1952). (2) Libby, W.F., Ibid.,56, 872 (1952). (3) Singer. L.. and Armstrong. W.D., Sucleonics. 11, 55 (1953) (4) Willard, H. H., and Winter, 0. B., IKD. EBG.CHEM.,. ~ - . A L , ,ED., 5 , 7 (1933). (1) Hornig, H. C., and
RECEIVED f o r review Xovember 22, 1953. Accepted lIarci1 3 , 1954, Investigation supported b y Medical Research and Development Board, Office of the Surgeon General, Department of the Army, under Contract KO, DA-49-007-AID-390.
Polarographic Determination of Alpha-Ketoglutaric Acid JAMES K. PALMER' and CLIFFORD 0. JENSEN D e p a r t m e n t o f Agricultural and Biological Chemistry, The Pennsylvania State University, State College, Pa.
I
S THE course of studies on the metabolism of plant tissues, a rapid, sensitive, and specific method for determining a-ketoglutaric acid was desired. The most commonly used method for determining a-ketoglutaric acid was developed by Friedemann and Haugen ( 4 ) . It is based on the formation of a red-brown color u hen the 2,4-dinitrophenylhydrazone of a-ketoglutai ic acid is treated with sodium hydroxide. This method is sensitive, but it is specific for a-ketoglutaric acid only vhen carried out under carefully defined condition?. I t also suffers from the reintive lack of stability of the colored conipound and the rangc' is small. Several other procedures have been clmplored for determining keto acids. Clift and Cook ( 3 ) descrihcd a titration method 1 Piesent address f l a i e n Conn
Connecticut 4griculturai Experiment Station
Yew
based on thc formation of a bisulfite addition product. Keto acids also have been determined by measuring the carbon dioxide produced on oxidation with ceric sulfate or permanganate ( 7 , 11). These methods are not specific for a-ketoglutaric acid. Chi omatographic separations have been described employing silica gel ( 5 ) )ion exchange resins (I), and filter paper Celite (8), ( 2 , f0). Procedures using silica gel columns and paper chromatograms have given good recoveries of a-ketoglutaric acid, but these methods itre not convenient for the determination of a single acid. Previously, certain acids of the tricarboxylic acid cycle have been successfully determined polarographically ( 9 , 1 2 ) . The polarographic behavior of a-ketoglutaric acid is such that waves suitable for quantitative analysis are obtained in solutions buffered a t pH 1.8 and 8.2. The current increment observed under