chromic acid used. It also varies slightly with time. Horrever, a duplicate blank determination at the beginning of each day provides a reliable blank for the entire day’s analyses. The reagent blank used t o obtain the data in Table I1 differed in amounts of chromic acid and potassium cyanide from the blanks used t o obtain the data in Table 111. In addition, the blanks for serum samples numbers 1 and 2 differed from serum samples numbers 3, 4, and 5 . In the latter case a different bottle of reagent grade potassium cyanide was used. This illustrates the importance of using the same stock solutions and the same amounts of all stock solutions in the distillation procedure. It was experimentally verified that significant variation in the chloride concentration in the synthetic bromidechloride mixtures or in the trichloroacetic acid concentration in the deproteinized blood serum samples resulted in no measurable change in the bromide concentration found after distillation by the designated method. A variation of = k l O O ~ o in the chloride concentration resulted in no significant change in the data given in Table 11. -Ilso, a variation of ?=IO% in the trichloroacetic acid concentration produced no measurable change in the data given in Table 111. The concentration range of bromide found in normal blood serum (1.7 to 4.2 1i.p.m.) by this method agrees quite well with the ranges found by other investigators (5-6). The speed (an
entire analysis of bromide in blood serum requires only 15 minutes), the accuracy and precision (in all blood samples run the precision is ~ t 0 . p.p.m. 5 and the accuracy is within 1.4 p.p.m.), and the sensitivity (using the designated procedure 0.01 p.p.m. is readily detected) should make it possible for clinical laboratories to analyze for bromide in blood serum by the described procedure. In addition, commercial equipment is available a t loiv cost. The recovery of added bromide from deproteinized blood serum samples is poorer than similar recoveries from synthetic bromide-chloride mixtures. This is quite likely due to an incomplete removal of protein in the deproteinization procedure. It was experimentally found that deproteinization procedures using smaller concentrations of trichloroacetic acid than employed here left appreciable quantities of protein in the solution to be distilled. The small amount of protein gave significantly lower results. However, unless results accurate to much better than 1 p.p.m. are desired, a more complete protein removal is unnecessary. Using a semimicro setul)--i.e., a 10-ml. beaker with about 4 ml. of solution and 10-ml. burets for addition and dilution reagents and using a 0.5-p.p.ni. bromide null reference solution-it should be possible t o be able to analyze 0.1-ml. samples of deproteinized serum as rapidly, precisely, and accurately as the serum samples determined in this paper.
LITERATURE CITED
(1) Allport, L. S . ,Keyser, J . W., “Colorimetric Analysis,” pp. 73-6, Vol. 1,
Chapman and Hall, London, 1957. (2) Boltz, 0. F., “Colorimetric Determination of Nonmetals,” pp. 181-95, Vol. 8 , Interscience, New York, 1958. (3) Bomen, H. J. M.,Bzochem. J . 73, 381 (1959). (4) Conwag, E. J., Flood, J. D., Zbzd., 30,716 (1936). ( 5 ) Hunter, G., Zbid., 60,261 (1955). (6) Kaplan, D., Schnerb, I., ANAL. CHEM.30,1703 (1958). (7) Kolthoff, I. M., Stenger, V. A , , “Volumetric Analysis,” p. 248, Yol. 11, Interscience, Sew York, 1947. (8) llalmstadt, H. V., Winefordner, J. D., ANAL.CHEM.32, 281 (1960). (9) Malmstadt, H. V., Winefordner, J. D., Anal. Chznz. Acta 20, 283 (1959). (10) Malmstadt. H. V.. Winefordner. ’ J. D.. Ibid.. 24. 91 11961). (11) llalmstadt,’ HI V.,’ Winefordner, J. D., Clzn. Chenz. 5 , 284 (1959). (12) Pllalmetadt, H. V., Winefordner, .! D., J . A m . Tt’uter Tt’orks S s s o c . 51, 133 (1959). (13) Fatelson, S., “Microtechniques of Clinical Chemistry,” p. 135, 2nd Ed., Charles C Thomas, Springfield, Ill , 1 qf31
4) Neufeld, d. H., Can. J . Research 14B, 160 (1936). 5 ) Sneed,
M. C., Maynard, , J. L., Brasted. R. C.. “Coniorehensive Inorganic ’Chemisiry,” p. ‘235, 1-01, 111, Van Sostrand, Sew York, 1953. 6) Van der Meulen, J. H., Cheir?. TT‘pekblad 28,82 (1931). 7 ) Van der Meulen. J. H.. Ibid.., 28., 238 11931). (18) Van Pinxteren, J. A . C., Analyst 77, 367 (1952). RECEIVED for review October 19, 1962. Accepted January 9, 1963. Taken in part from the M.S. Thesis of h u n g Tin, University of Florida, December 1962.
Hydrogenation of Linolenate Fractiona tio n of I s 0 me ric Esters by Countercurrent Distribution with an Argentation System C. R. SCHOLFIELD, E. P. JONES, R. 0. BUTTERFIELD, and H. J. DUTTON Northern Regional Research laboratory, U. S. Departmenf o f Agriculture, Peoria, 111.
b Methyl linolenate was hydrogenated to an iodine value of 153.5 with 0.570 nickel catalyst at 140’ C. and atmospheric pressure. The product was separated into monoenoic, dienoic, and trienoic esters b y countercurrent distribution using acetonitrile and petroleum ether. Each of these fractions was countercurrently distributed between 0.2N silver nitrate in 90% methanol and petroleum ether. cis-Monoenes were well separated from trans-monoenes. In the diene fraction, separation was less complete because there were differences in both configuration and relative position 386
ANALYTICAL CHEMISTRY
of double bonds. Analysis of the fractions b y such methods as infrared absorption, alkali and lipoxidase isomerization, and oxidative cleavage, followed b y dibasic acid analysis, gave additional information about the products formed during hydrogenation of linolenate.
D
URING
THE
HYDROGEN.4TION
Of
fatty acids and their derivatives many isomeric products are formed by shifts in position and configuration of double bonds. A study of the products formed during partial hydrogenation of
linolenate was reported in a previous paper (8),which reviewed earlier works. Nichols ( 7 ) in 1952 suggested that methyl oleate and methyl elaidate could be separated by countercurrent distribution between methanolic silver nitrate solutions and iso-octane. This separation has been performed using 0%’ 4gSO3 in 90% methanol and petroleum ether, and the same solvent system has been applied t o other isomeric monoenes and dienes ( 3 ) . The application of this solvent system to hydrogenated methyl linolenate gives a n improved fractionation of fatty acid isomers as described in this paper.
Cis
Table
I
I.
Analyses o f Diene Fractions
A B C D E' F Weieht. c% 3.8 4.4 39.3 3.1 38.1 11.3 ~truni, % i y 79.5 7l.i 60.7 5k.2 95.2 4i.1 Preformed conjugation, a231 ... 14.3 2.5 5.6 0.3 0.5 Alkali conjugatable diene, yob 37.S 54.2 42.1 16.2 0.8 2.1 Lipoxidase conjugatable diene, r0 . . . ... 27.4 ... 0.2 0.4 a Calcd. as methyl elaidate. Unpublished results a t this laboratory indicate that if methyl Yi 12-czs, trans- and trans,czs-linoleates were used as standard, values would be l . l i 2 times those listed in table. * Heated 4-3/4 hours at 180' C. ( 4 ) . ~
Transfer Number
Figure 1. Countercurrent distribution of monoenes from hydrogenated methyl linolenate between 0.2M AgNOa in 90% methanol and petroleum ether Transfer number i s plotted against scale reading of a differential recording refractometer
Various samples from three hydrogenations of methyl linolenate have been fractionated by countercurrent distribution n-ith the silver nitrate-solvent system. Monoene, diene, and triene fractions from one hydrogenated sample are described in detail, and some results from the other ti\-o hydrogenations are included for comparison. EXPERIMENTAL
The hydrogenations of methyl linolenate and initial fractionations were similar to those described earlier (8). I n Run 1, methyl linolenate (30.4 grams) was reduced to a n iodine value of 153.5 a t 140' C. with hydrogen a t atniosplieric pressure and with a comniercial catalyst (0.57, nickel on kieselguhr in hardened oil). The hydrogenated product was separated by countercurrent distribution between acetonitrile and petroleum ether into monoeiic, diene, and triene fractions. From the areah under the countercsurrent distribution curve, the h! drogeiiatcd product was calculated to rontain 0.9% stearate, 37.2% nionoene, 40.8% diene, and 21.1% triene. Monoene Fraction. 6.8-gram portion of t h e nionoene fraction from R u n 1 (52.6% trans) was distributed
between petroleum ether and 0.2M AgXOs in 90% aqueous methanol in a procedure like that employed with a n oleate-elaidate mixture (3). A lower phase of 40 ml. in each tube and an upper phase of 10 ml. for each transfer were used. The sample was recycled through a 200-tube instrument twice, and the separation of the components was monitored with a recording differential refractometer. Individual fractions were then collected in a fraction collector and combined as shown in Figure 1 to give trans and cis components. Infrared analyses indicated 96.5% trans esters in that fraction and none in the cis fraction. Dibasic acids were prepared by permanganateperiodate oxidation (6) of each fraction. Methyl esters of the dibasic acids were analyzed by gas chromatography. Results are shown in Figures 2 and 3 with data from two similar hydrogenations carried out under the same experimental conditions (Runs 2 and 3) rrhich are included for comparison. The differences in dibasic acid composition in samples from the three hydrogenations include both those caused by differences in the hydrogenations reactions and those caused by differences in analysis. Diene Fraction. I n the distribution of t h e diene fraction between petroleum ether and 0.2M AgNO, in 90Y0 aqueous methanol, the sample could not be recycled because of the great difference between the partition coefficients of the components. Figure 4 shows a distribution of 8 grams of the
Carbon Atoms in Dibnsic Acid
diene. Individual tubes were combined as shown, and the percentage of each fraction in the recovered material is shown in Table I with analytical data on the fractions. Dibasic acid composition obtained after permanganate-periodate oxidation is shown in Figure 5, and its calculation is explained in the discussion section. Triene Fraction. A 4.4-gram sample of the triene fraction was distributed between the methanolic silver nitrate and petroleum ether solvents. Fractions were collected and combined as shown in Figure 6. Analytical d a t a on the fractions are given in Table 11. Permanganateperiodate oxidation followed by dibasic acid analyses revealed little difference in the fractions. The dibasic acids contained 70 to 80% azeleic acid and very little acids of longer chain length. RESULTS AND DISCUSSION
Partially reduced fatty acid methyl esters can be separated by countercurrent distribution between acetonitrile and petroleum ether into monoene, diene, and triene fractions. This separation followed by countercurrent dis-
Table I I .
-
Analyses of Triene Fractions
"
Weieht. 5; trans, c/c"
A 8.1
R C 19.1 7'7.8 27.2 3.9
64.9 Linolenate by alkali conj ugationb 82.0 107.2 99.0 a Calcd. as methyl elaidate. * Heated 45 min. at 180" C. ( 2 ) .
Carbon Aions In Dibaslc k i d s
Figure 2. Dibasic acids from cis-monoenes of hydrogenated methyl linolenate.
Figure 3. Dibasic acids from trans-monoenes of hydrogenated methyl linolenate
Value for each acid from Runs 1, 2, and 3 shown from left to right
Value for each acid from Runs 1, 2, and 3 shown from left to right
VOL. 35, NO. 3, MARCH 1963
387
I\ IO
Figure 6. Countercurrent distribution of trienes from hydrogenated methyl linolenate between 0.2M AgNOa in 90% methanol and petroleum ether Figure 4. Countercurrent distribution of dienes from hydrogenated methyl linolenate between 0.2M AgNOa in 90% methanol and petroleum ether
Transfer number i s plotted against scale reading of a differential recording refractometer
Transfer number Is plotted against scale reading of a differential recording refractometer
nation mechanism as suggested for fats by Allen ( I ) . In the diene fraction the separation is dependent upon both configuration and relative position of the double bonds. Also the components differ so greatly in partition coefficient that it is impossible to recycle the sample. For these reasons separation is less complete than for the monoenes. Fraction A in Figure 4 is s h o r n in Table I to make up only 3.8% of the recovered sample-less than intermediate fraction B. Fraction A is a t the position of a false peak in the differential refractometer curve which we believe is caused by a change in solvent composition as the sample started to emerge from the instrument. The interpretation of the dibasic acid data for the diene fractions is difficult because two dibasic acid molecules are obtained from each diene molecule-one from the carboxyl end of the fatty acid and one from the portion between the double bonds. We have assumed that those dibasic acids of 8 or more carbon atoms are terminal-from the carboxyl end-and those with 7 or less carbon atoms are internal-from between the double bonds. This assumption is not strictly true since the dibasic acid analyses of the monoenes show that a small number of double bonds have shifted closer to the carboxyl than the 8 position. However, the assumption does seem to be the most satisfactory way to handle the data. It is a reasonably good approximation and it allows classification of the dibasic acids in a way which gives more information than if all dibasic acids were grouped together. We have adjusted the values in Figure 5 so that the sums of terminal and internal dibasic acids are each 100%. Also, since malonic acid is unstable and its recovery is not quantitative, we have used the value for alkali conjugation as a measure of this acid. Fractions A and C are quite complex judged by the data in Table I and the
tribution of the monoene fraction between methanolic silver nitrate and petroleum ether achieves a good separation of cis- and trans-monoene fractions. Oxidative cleavage of these fractions provides a definitive analysis of the isomeric monoene mixture. Figures 2 and 3 show the dibasic acid analyses for monoenes from the three hydrogenation runs. I n the cis fractions the major part of the double bonds remain in the original positions with fewer double bonds in the 12 than in the 9 and 15 positions, I n the trans fractions the double bonds are widely scattered nith
the largest amounts in the 10, 11, 13, and 14 positions. In the cis fractions from 68 to 80% of the double bonds are in the original 9, 12, and 15 positions while in the trans fractions from 23 t o 3570 of the double bonds are in these positions. If one considers the monoenes with double bonds in the naturally occurring 9, 12, and 15 positions, these monoenes have 68 to 7470 of their double bonds in the cis configuration; of the monoenes with double bonds in other positions, only 13 to 25% are cis. This change in position and configuration is consistent with a half hydrogenation-dehydroge-
-
-frattion
-t
-Friction -F rlnternob y T c r m i n a I 1 -Fraction -C ilnternall yTerminul-,
G
'
I
-A
-Friction
3
5
7
Figure 5.
8 1 0 1 2 ~ 3s
I 12 14 3 ! o n Atoms in Dil
12 14
Dibasic acids from diene fractions obtained in distribution shown
in Figure 4
388
t I
7
ANALYTICAL CHEMISTRY
range of dibasic acids in Figure 5. The small amounts of conjugated dienes present seem to be concentrated in fractions A and B-the first fractions to emerge from the countercurrent distribution apparatus. Based on runs with known esters (3) conjugated dienes would be found in the region of fraction A . The 9,12-cis,tr~ns-or trans,&- and 9,12-cis,cis-dienes would be found on opposite sides of the maximum in fraction C. Lipoxidase conjugation of C (6) shows that there is 27% of &,cis-methylene interrupted or skipped double bond esters in this fraction-probably 9,12 and 12,15 esters as shown by the terminal dibasic acids. It is not known why dienes giving the Cd to C, internal dibasic acids occur in fractions A and C; possibly some of these give the short chain terminal dibasic acids. On the other hand, a 4-3/4-hour alkali conjugation, which should measure all skipped double bonds yielding malonic acid (4), gives a value of only 42% in fraction C and indicates that 58y0 of the fraction contains double bonds separated by more than one methylene group. By contrast, alkali conjugations of fractions E and F show t h a t almost all of the double bonds in these fractions are separated by more than one methylene group. The large amount of 9carbon terminal and 6-carbon internal dibasic acids shows that both are largely 9,154ienes. This large amount of 9,15 esters in fractions E and F,which make
up 49% of the diene, is consistent with the smaller amount of 12-esters formed in the monoene. Fractions E and F differ mainly in the trans content. Contrary to expectations E , the first to emerge from the apparatus, is the lower. It is apparent that more information on the behavior of specific positional and geometric isomers of linoleic acid in the argentation system is needed. I n the triene fraction the same factors t h a t limit the fractionation of the diene are even more important. Our first study indicated that the triene was not changed from the original linolenate (8). However in the hydrogenation of trilinolenin (9) trans bonds were found in the triene fraction, and they have also been found in triene fractions from later hydrogenations of methyl linolenate. I n our hydrogenations a t 140’ C. with 0.5% nickel, the remaining triene seems to be altered more in those experiments of slower reaction rate. Countercurrent distribution of the triene fraction (Figure 6) produces a single large band preceded by a forerun. As shown in Table 11, esters containing trans bonds are concentrated in the forerun. I n a similar distribution of triene from another hydrogenation, a second peak occurred near transfer 500 with trans content of 89%. The high triene content of linolenate measured by alkali isomerization in fraction B was also confirmed with this other sample. In this sample an intermediate fraction had
a n apparent value for linolenate of 124%. Triene adsorption a t 267 m,u was higher than diene adsorption a t 233 mp, Dibasic acid analyses provide little information about the differences between fractions; 70 to 80% of the dibasic acids are azelaic, and there is little dibasic acid of longer chain length. Evidently better separation and better analytical methods are needed to investigate the structure of the esters in the triene fraction. LITERATURE CITED
(1) Allen, R. R., Kiess, A. A., J . Am. Oil Chemists’ SOC.33,355-9 (1956). (2) Brice, B. A,, Swain, M. L., Herb, S. F,., Nichols, P. L:, Jr., Riemenschneider, R. W., Ibzd., 29, 279-87 (1952). (3) Dutton, H. J., Scholfield, C. R., Jones, E. P., Chem. Ind. (London) 1961 1874-6. (4)Jackson, J. E., Paschke, R. F., Tolberg, W., Boyd, H. M., Wheeler, D. H., J. Am. Oil Chemists’ SOC. 29, 229-34 (1952). (5) Jones, E. P., Stolp, J. A., Ibid., 35, 71-6 11958). (6) MacGee,’J., ANAL. CHEM.31, 298302 (1959). (7) Nichols, P. L., Jr., J . Am. Chem. SOC. 74,1091-2 (1952). ( 8 ) Scholfield, C. R., Jones, E. P., Nowakowska, J., Selke, E., Sreenivasan, B.. Dutton. H. J.. J . Am. Oil Chemists’ Sic. 37,579-82 (1960). (9) Scholfield, C. R., Jones, E.P., Dutton, H. J., ANAL. CHEM. 33, 1745-8 (1961). RECEIVED for review October 17, 1962. Accepted December 26,1962.
Paper Chromatography of Several Classes of Compounds: Correlated Rf Values in a Variety of Solvent Systems KAY FINK, RICHARD E. CLINE, and ROBERT M. FINK Departmenfs of Biophysics and Physiological Chemistry, School o f Medicine, University o f California, 10s Angeles, and the VA Hospifal, long Beach, Calif.
b Filter paper chromatographic data in 10 solvent systems are reported for several classes of compounds, including amino acids, purines, pyrimidines, sugars, organic acids, and compounds reacting with acidic p-dimethylaminobenzaldehyde. The data are especially useful as an aid in identification of radioactive metabolites, which may be detected on chromatograms at concentrations considerably lower than that required for classification by color reactions. The data are also of value in selecting solvent systems for the separation of two or more substances, and in providing information on other compounds which might overlap or interfere.
F
chromatographic data for a wide variety of compounds and for numerous solvent systems have been published, but there is a relative scarcity of correlated R f values for various classes of compounds in the same set of solvents. The latter type of data, as tabulated beloIv, can be particularly useful in experiments involving the metabolism of radioactive substrates. Such experiments may yield diverse and unexpected types of labeled metabolites, and these may often be separated and detected as radioactive spots on paper chromatograms even though the concentrations involved may be far too low for convenient use of classification procedures such as color tests. ILTER PAPER
As an illustration, use of isotopically labeled thymine as a substrate for rat liver slices (4) yielded an array of radioactive products that was eventually shown to include several previously identified metabolites with nidely differing chemical properties (dihydrothymine, P-ureidoisobutyric acid, p-aminoisobutyric acid, and urea), an alphaamino acid and a sugar that were well known but not anticipated as major radioactive products (alanine and glucose), plus two oxidation products and a nucleoside that were not previously known to occur biologically (uracil-5carboxylic acid, 5-hydroxymethyluracil, and 5-methyluridine). It was not expected that such a complex mixture of metabolites could be adequately VOL 35, NO. 3, MARCH 1963
389
