Polarimetric Determination of Boron SIR: Borate ion forms complexes with a wide variety of polyols ( I ) , and whenever the poly01 is optically active, changes in specific rotation normally result. Only Rosenhrim and Legser (4, using d-tartaric acid as the reagent, h a w attempted to employ this effect for the polarimetric determination of borate. Cnfortunately. there is interference from many cations, which form complexcs nith the reagent, and even from inert salts, which affect the activity of the various ionic spwies. d more satisfactory method was developed hrrc for the determination of borate undcr conditions which prohibit the use of the conventional methods, such as acid-base titrimetry. Of special interwt n-as the determination of total boron, in the form of boric acid or borate salts, in the presmce of ions such as fluoride and phosphate which preclude the use of the usual alkalimetric method. Also of special interest was the determination of borate and other forms of oxidized boron in borohydride salts. Rerc the entire analysis must be done in strongly basic medium since the borohydride ion rapidly hydrolyzes, forming borate, in neutral or acidic solutions (3). Mannitol n as selected as the reagent for several reasons. Since this reagent is uncharged, changes in ionic strrngth should have a relatively minor effect. Although mannitol does form complexes n i t h several ions, this tendency is much less than that of tartrate, and fewer specific interferences were expected. The specific rotation of mannitol itself is small (-0.5") while that of the mannitol-borate complex is large (+22", based on the weight of mannitol) (W), permitting reasonably high sensitivity with minimum dependence of rotation on the amount of mannitol used. Finally, the formation constant of the mannitol-borate compkx is large, ensuring quantitative complexation and a linear calibration curve. EXPERIMENTAL
Apparatus and Reagents. All rotations nere measured u i t h a n 0. C. Rudolph and Sons Model 80 precision polarimeter, using a sodium source and a 2-decimeter tube. Each measured rotation was the average of at least six individual measurements taken by approaching the balance point alternately from opposite directions. The standard deviation of the average rotations was approximately ~ + = 0 . 0 1 ~ . Alannitol solutions were prepared b y direct weighing from reagent grade mannitol. Standard sodium borate
solutions were prepared by adding the stoichiometric quantity of sodium hydroxide to reagent grade boric acid, and standardized by alkalimetric titration in the presence of mannitol. Procedure A. DETERMISATION O F TOTALBoxox AS BORIC ACID OR BORATESALTS. Transfer the sample t o a 50-ml. volumetric flask, add 25 ml. of mannitol-buffer solution (mannitol. 100 grams per liter; aninionium hydroxide, 1.6M; ammonium chloride, 2 X ) , dilute t o t h e mark, niiu, and measure t h e rotation. Determine the tieight of boric oxide in the sample from a calibration curve prepared by the same procedure n i t h standardized boric acid solutions rcplacing the sample. Procedure B. DETERM~KATIO?; OF BORATE IKBOROHYDRIDE SALTS.Transfer the borohydride sample (approximately 3 grams of sodium borohydride) to a 50-ml. volumetric flask, add 25.0 nil. of mannitol-hydrouide solution (mannitol, 100 grams per liter; sodium hydroxide, 0.6M), dilute to the mark, miu, and measuie the rotation within 10 minutes after mixing. Determine the neight of boric oxide in the sample from a calibration curve prepared n i t h the same lot of mannitol-hydroxide solution, using measured volumes of standardized sodium borate solutions. RESULTS A N D DISCUSSION
The calibration curve for Procedure
A is substantially linear and its slope is independent of the concentration of mannitol, in agreement with the earlier findings of Isbcll and coworkers (W), provided the mannitol-borate ratio is 2 to 1 or greater on a molar basis. Deviations from linearity become dptectable nlien the ratio falls significantly bf,low this value. There is a small but significant deviation from linearity n hrn the sample contains very small amounts of boric oxide; the measured rotation for no boric oxide falls about 0.02 degree above the extrapolated intercept of the linear portion of the curve. The slope of the linear portion of the calibration curve is $0.0047" per nig. of boric oxide, permitting a prpcision of about 1 2 nig. in the drterniination. The value of the intercept is directly proportional to the concentration of mannitol @), but is very small (-0.04" for the concentration specified in the procedure) ; hence the concentration of mannitol need not be extremely carefully controlled. The curve is likewise unaffected by changes in p H over a range of 1 p H unit; unless the sample contains larger amounts of acids or bases
&an can be accommodated by the buffer, the sample need not be neutralized prior to analysis. Steady rotations are achieved within a few minutes after mixing; when the solutions are allowed to stand for long periods rotations increase slonly, presumably because of eutraction of boron from the flask. Changes in temperature in the interval between 25" and 40' C. produced no detectable changes in rotation. Indifferent salts such as sodium chloride produced no detectable effects n-hm present in concentrations up to 1111, the highest concentration studied, in the final solution. Phosphate ion has only a very slight effect on rotation; n hen this ion was present a t a concentration of 0.2M in the final solution, the rotation of a solution containing &5mg. of B203nas oiily 0.02" larger than TT hen this ion was absent, a difference barely nithin the detectable limits of the method. Fluoride ion produced no detectable effect in concentrations as high as 0.LU in the final solution. Honever, the fluoborate ion is formed in appreciable amounts in acidic solutions containing fluoride and boric acid ( 7 ) . Since this ion is hydrolyzed only very slomly in faintly alkaline media (5, ?'), it would be necessary to boil such samples in strongly basic media to effect complete hydrolysis prior to determining the total boron content. The addition of sodium hydroxide to mannitol solutions produces a pronounced negative rotation. This rotation is probably due to the formation of small concentrations of the conjugate base of mannitol. Although mannitol is a very weak acid, with a dissociation constant of only 4.2 X ( 6 ) , significant conversion to the base form R ould be expected under the conditions of Procedure 13. In fact, both the slope and the intercept of this calibration curve are dependent upon the concentration of mannitol and of sodium h>droxide. Both of thcse vuriables must be controlled nithin close limits to achieve accurate analyses. For the same reason, the sample itself must not contain significant quantities of strong basps or acids of any kind; fortunately, borohydride samples seldom do. The prcsence of interferences from this source can be detected by making duplicate analyses with different weights of samples; a dependence of the determined percentage of boric oxide on the sample weight indicates the presence of such interference. The calibration curve does not deviate from VOL. 33, NO. 3, MARCH 1961
471
linearity at low concentrations of boric oxide with Procedure B. For the conditions specified, the intercept and slope of the calibration curve are about -0.38' and f0.0063' per mg. of boric oxide, respectively. For best results with Procedure B, the concentrations of sodium hydroxide and mannitol should be equal, on a molar basis. Then steady rotations are achieved within a few minutes after mixing, the calibration curve is linear, and relatively high sensitivity is realized. If the concentration of sodium hydroxide greatly exceeds that of mannitol, rotations change significantly with time of standing, the rate of change increasing as the hydroxide concentration increases. The observed changes in rotation are much larger than can be accounted for by extraction of boron from the container and must result from slow attainment of equilibrium. Even under the strongly alkaline conditions existing in Procedure B, borohydride hydrolyzes a t a significant rate. For the conditions specified, rotations increase a t the rate of about 0.001 to 0.002' per minute, and hence, should be measured promptly after mixing. The following results obtained in the analysis of sodium borohydride samples illustrate the applicability of the method. The analysis of duplicate samples of one lot of impure sodium borohydride weighing 3.144 and 3.071 grams gave values of 4.22 and 4.35% boric oxide, respectively. Two addi-
tional analyses of the same lot using samples weighing 1.524 and 1.936 grams, but with 90 and 45 mg. of added boric oxide, respectively, showed a total boric oxide content of 157 and 131 mg., respectively. After the weight of added boric oxide is subtracted, these values correspond to 4.39 and 4.44% boric oxide in the original sample. The average of the four determinations is 4.35% boric oxide with a standard deviation of *0.09%',. A sample of the purest sodium borohydride available to us was found to contain 0.23 mg.-atom of borate boron per gram by polarimetric analysis. Very careful determinations of total boron and of hydride hydrogen in this sample gave values of 26.16 and 103.3 mg.-atoms per gram, respectively. The value obtained for hydride hydrogen indicates that the sample contained 25.83 mg.-atoms of borohydride per gram, leaving 0.33 mg.-atom of boron presumably in the form of borate, which agrees well with the polarimetric measurement. A similar analysis on a less pure sample gave values of 23.69, 91.07, and 0.84 mg.atom per gram for total boron, hydride hydrogen, and borate, respectively. The concentration of borate calculated by difference from the total boron and hydride hydrogen determinations is 0.92 mg.-atom per gram, in good agreement with the value of 0.84 obtained by polarimetric determination. Although the methods described above are only about one fifth as sen-
sitive as the Rosenheim and Leyser method, they are more generally applicable and less subject to interferences than the latter. LITERATURE CITED
(1) Boeseken, J., "Advances in Carbohydrate Chemistry,,' Vol. 4, pp. 189-
210, W.W. Pigman and &I. L. Wolfram, eds., Academic Press, ?Jew York, 1949. (2) Isbell, H. S., Brewster, J. F., Holt, N. B., Frush, H. L., J . Research Natl. Bur. Standards 40, 129 (1948). (3) Pecsok, R. L., J . Bm. Chern. Soc. 75, 2862 (1953). 14) Rosenheim. -4.. Levser. F.. Z. anora. allgem. Chem: 119, ("1921). ' (5) Ryss, I. G., Slutskaya, h1. M., Zhur. Fiz. Khim. 21, 549 (1947). (6) Terechov, P., Collection Czechoslov. Chem. Communs. 1, 551 (1929). ( 7 ) Wamser, C. .4.. J . Am. Chem. Soc. 70, 1209 (1948) \
-I
DOSALDD. DEFORD
ARTHUR S. BLONDER' ROBERTS.BRAN AN^
Department of Chemistry Northwestern University Evanston, Ill. RECEIVED for revieiv September 26, 1960. Accepted Decemher 21, 1960. Research supported by Callery Chemical Go. under a contract with the Bureau of k r o nautics, Department of the Xavy. 1 Present address, Department of Biochemistry, Northwestern University Medical School, Chicago, Ill. 2 Present address, Arniour Research Foundation, Chicago, 111.
Determination of Carbon, Oxygen, and Silicon in Solids by Activation Analysis with 15-M.e.v. Deuterons SIR: We have carried out experiments to test the possibility of determining small amounts of carbon in pure silicon dioxide by means of radioactivation with cyclotron deuterons of 15 m.e.v. The results show that it is feasible to determine as little as a few tenths per cent of C in Si02 by counting induced positron radioactivity in the bombarded sample without chemical processing. From the data the ratio of Si to 0 may also be estimated. The principal limit on the sensitivity for detecting C is the large amount of 0 present; for samples lower in 0 the sensitivity should be much better. I n our experiments we used the stationary external target of the M.I.T. Cyclotron, a water-cooled A1 block upon which solid target materials approximately 10 cm. long and 1 em. aide can be placed. Five specimens Tere bombarded simultaneously in small A1 capsules machined from A1 rod to 0.33 mm. (89 mg./sq. cm.) wall thickness 472
ANALYTICAL CHEMISTRY
and held in place by covering with 0.0017-inch (11.6 mg./sq. em.) Dural foil. Specimens were 15 mg. of powdered material, and within each specimen the deuteron energy ranged from about 8 m.e.v. (reduced from an initial 15 m.e.v. by absorption in the covering foil and capsule wall) to 0 m.e.v. According to nuclear data summarized by Koch ( I ) , in this energy range the important nuclear reactions are C'*(dJn)NL3,016(d,n)F17J017(d,n)F18,0'8 (d,2n)F18, Si*g(d,n)P3*,and Si30(d,p)Si31. The ratio of C t o 0 is determined by analysis of a decay curve of annihilation positron activity induced by a short deuteron bombardment of the sample and comparison standards. The components 1.1-minute F1', 2.5-minute P30, 10.0-minute "3, and 112-minute F18, all pure positron emitters ( 2 ) , may be resolved, and the relative counting rates of 10.0-minute N13 and 112minute Fl* are a measure of the ratio of C to 0 in the sample. The ratio of
0 to Si is determined by counting the specimen simultaneously with y-scintillation and p-proportional detectors, mounted on opposite sides of the specimen position, for the relative activities of 112-minute F1s and 157-minute Si3I, a negatron emitter with virtually no y in its decay (g). Interference of the 0.65-m.e.v. p+ of FIB in the measurement of the 1.48-m.e.v. 8- of Si31 is eliminated by means of an A1 absorber of 250 mg./sq. em. Figure 1 is a plot of the counting rate ratio of N13 to PI8 for 5 specimens of SiO, nith varying content of C (as Sic). Bombardment was carried out for 1 minute, and the specimens were immediately transferred from the bombardment capsules to planchets for counting. A single channel scintillation spectrometer was centered on the 0.51-m.e.v. annihilation r-photopeak, and the measured activities were extrapolated to the time a t which bombardment ended. The resulting decay
