where C A =
% concentration of A by weight in filter
Similarly p B x = CB.Vl~/lOO a
(1%)
It is now possible to rewrite Equation I-c to yield
(I-h) The relationship in Equation I-h, by itself, is of limited value: Its usefulness becomes significant only when it is realized that the problem posed above is one of equivalent replacement. I n other words, using the semitheoretical method, one does not need to know what the theoretical intensity should be for a given multicomponent system; rather one wants to know how the observed intensity is modified as a given concentration of element A is replaced by an equivalent concentration of element B. Under these circumstances CA = CB = C , and the desired relationship is obtained :
Whereas element B replacing an equivalent concentration of A might decrease the observed intensity ratio by lo%, another element such as D or E might increase it in this equivalent concentration exchange. Clearly, fof a multicomponent system the final intensity ratio is a product of the individual equivalent concentration exchange ratios. Reverting back to the symbols used above for the determination of element 1 in a system composed primarily of elements 1 and 2, with the equivalent concentration exchange of element 2 by elements 3, 4, 5, etc., the intensity ratio, ZR’,is given by
which finally yields
Equation I-k demonstrates how the empirical constant, F’, of Equation 4 must contain terms such as area and total mass. Furthermore, it lays the basis for a rigorous justification of the original assertions by Tingle ( I d ) and Hasler and Kemp (4). LITERATURE CITED
(I) Andermann, G., Allen, J. D., AXAL. CHEM.33, 1695 (1961). (2) Beattie, H. J., Brissey, R. M., Ibid., 26, 980 (1954). (3) Gillam, E., Heal, H. T., Brit.J. A p p l . Phys. 3, 353 (1952). (4) Hasler, RI. F., Kemp, J. W., “Methods
for Emission S ectrochemical Analysis”, p. 81, Philadeghia, Pa., 1957. (5) Kemp, J. W., Andermann, G., Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, 1956. (6) Koh, P. K., Caugherty, B., J . A p p l .
Phys. 23, 427-33 (1952). (7) Marti, W., Spectrochim. Acta 18, 1499 (1962). ( 8 ) Mitchell, B. J., ANAL. CHmf. 30, 1894 (1958). (9) Noakes, G. E., ASTM Spec. Tech. Publ. 157. 57 11953). (10) Sherman, J., Spectrochim. Acta 7, 283 (19,55). (li,>dz;d.,-co. 6, 466 (1959). (12) Tingle, W. H., Pittsburgh Conference on Analytical Chemistry and Applied __ Spectroscopy, 1955.
RECEIVEDfor review June 25. 1962. Resubmitted October 6, 1965. Accepted November 17, 1965.
Counter Double Current Distribution with Continuous Recovery for Isolation of Methyl Linolenate R. 0. BUTTERFIELD, H. J. DUTTON, and C.
R. SCHOLFIELD
Northern Regional Research laboratory, Peoria, 111. Counter double current distribution (CDCD) was used to produce unisomerized methyl linolenate (methyl 9,12,15-octadecatrienoate) of 99.9% purity from methyl esters of linseed oil without prior concentration. The rate of production is five times that of countercurrent distribution. Production of the labile lipid illustrates the general applicability of CDCD. In addition, a system is described that continuously recovers the solvent and product. This system completes the automation of the CDCD equipment so that solvent inventory, safety hazards, and labor are reduced.
B
RoblIh.ATIoN-debromination has been the accepted, if not the only, method for the preparation of methyl linolenate (methyl 9,12,15-octadecatrienoate) (4); however, it induces up to 25% of trans isomers in the naturally occurring all-cis compound. Bside from 86
ANALYTICAL CHEMISTRY
the scaling up of chromatographic procedures (7, 8 ) , only liquid-liquid extraction methods have held the possibility of isolating an unisomerized compound. I n large laboratory-scale equipment, up to 96% pure linolenic acid has been prepared on a pound per hour basis (1), and by using countercurrent distribution (CCD) monitored by a continuous flow refractometer (3) high-purity methyl linolenate has been isolated in a batch process a t a rate of 75 grams per week (9)*
With the description of counter double current distribution (CDCD) (6),a continuous method for isolating high-purity linolenate was possible. Continuous and essentially unattended operation became a reality with the addition of a continuous solvent-product recovery system. More than 300 grams of linolenate can be separated in a week from linseed methyl esters. The linolenate has a purity of greater than 99.9% and essentially no trans isomers.
EXPERIMENTAL
Counter Double Current Distribution. Distributions were performed
with a robot-operated 25-tube C D C D apparatus ( H . 0. Post Scientific Instrument Co.) with hexane and acetonitrile (9), the immiscible solvent pair. Linseed methyl esters and a methyl linolenate concentrate (81%) were used as feed materials and were diluted 1 to 2 with hexane to reduce viscosity and to improve stabihty to oxidation in the pump reservoir. The ratio of the immiscible solvents, the position of the feed tube, and the rate of feed were varied from run to run. Preparation of Feed Materials. Alkali-refined linseed oil was transesterified with sodium methoxide. Product composition was 58.5% linolenate, 15.7% linoleate, 17.4% oleate, and 8.4% saturates. A methyl linolenate concentrate was prepared by urea crystallization of the linseed methyl esters ( 5 ) . Its composition was 8l.lyO linolenate, 16.470 linoleate, and 2.5% oleate.
Feed
f
2.0 $0
1.6
w
a 3
;1.2 E w
s . I -
.g 0.8 L
0.4
I
0
0 Tube Number Figure 1. run 3
I 2
I I I I 3 4 5 6 Total Weight in 70 ml., g.
I
I
7
8
Figure 2. Variation of linolenate partition coefficient in a hexane-acetonitrile system with total concentration
Distribution of esters at equilibrium for
+
RESULTS AND DISCUSSION
The following convention is used throughout : The upper layer (raffinate) is considered to move from left to right and the lower layer (extract) from right to left. Tubes are numbered from left t o right, so that the lower layer is removed from the machine at tube 1 on the left and the upper layer, at tube 25 on the right. The solvent ratio is the Yolume of upper solvent phase divided by that of the lower solvent phase. I n the first series of experiments, methyl linolenate concentrate was used as feed material. Table I lists experimental conditions and results for three runs. I n run 1 inore than 5 7 7 , (46 X 99.5;Sl.l) of the methyl linolenate in the original sample was recovered in the extract fraction. To obtain purity higher than 99.57,, the position of the feed tube or the solvent ratio, or both, was changed. Run 2 illustrates the effect of increasing the solvent ratio and decreasing the tube number position of the feed. Under these conditions, linolenate purity was better than 99.9%, but recovery of linolenate in the extract fraction dropped to 2175. I n run 3, the solvent ratio was again increased but was compensated for, in part, by changing the position of the feed tube t o maintain the best recovery rate. Figure 1 is a plot of the concentrations present in the instrument tubes for run 3 at 250 transfers when steady-state condition> had been achieved, since the sum of esters in the raffinate and extract equaled the feed. Since concentration of methyl linolenate from linseed oil by urea crystallization is a laborious step and since the CDCD operation is essentially automatic, the direct use of linseed oil methyl esters as feed material, though less efficient as a process, was investigated. Table I1 lists the conditions and results for three more runs. The low purity of
I
1
Feed
2: 0
Tube Number Figure 3. Distribution of linseed methyl esters at equilibrium for run 6
Table I.
Counter Double Current Distribution Preparation of Methyl Linolenate Starting with a Concentrate"
Run
Solvent ratio
Lower layer, vol., ml.
1 2 3
0.40 0.47 0.50
25 25 50
a
Effluent solvent analysis Lower Wt. 7 , Upper, of feed Ln, 7cb Ln, 7, ~~
Feed tube
Feed weight, mg.
14 12 10
120 160 350
46 17 16
99.5 100 100
48 66 72
Original oil composition. 81.17, linolenate, 16.47, linoleate, 2.5% oleate. Linolenate purity.
Table II.
Counter Double Current Distribution Preparation of Methyl Linolenate Starting with Unfractionated Linseed Methyl Esters"
Run
Solvent ratio
Lower layer vol., ml.
4 5 6
0.40 0.40 0.40
25 50 50
Original oil composition.
8.47, saturates.
Feed
Feed weight,
14 13 9
240 272 350
tube
mg.
Effluent solvent analysis Lower Wt. 70 Upper, of feed Ln, % Ln, 46.6 26.4 33.1
94 100 99.9
58.57, linolenate, 15.7% linoleate, 17.47,
7.0 41 36
oleate, and
VOL. 38, NO. 1, JANUARY 1966
e
87
lbrsr Points
n
CDCD I
phase divided by concentration in the lower phase) for methyl linolenate does not vary over a wide range of concentrations -i.e. , exceeding that attained in actual runs indicated by arrow. However, other studies showed that a t concentrations found in the CDCD ~Vlgrrnrr apparatus the partition coefficient for methyl linoleate decreases in the presence of increasing amounts of linolenate. The impairment in resolution in run 4 (Table 11) is caused by the decreased partition coefficient of linoleate. Comparing runs 5 and 6 in Table I1 shows the effect of moving the feed tube toward lower numbers to increase the *Strm Elit recovery of linolenate. From Figure 3, which shows the concentrations present in the instrument tubes for run 6, it is seen that the feed at tube 9 is in approximately the optimal position for the proCollrctior Flask duction of methyl linolenate, since moving the feed tube to the left would increase contamination and moving it t o the right would reduce yield. By using Figure 4. Diagram of solvent and product this position of feed, 2.25 grams of recovery stills and connections to counter methyl linolenate per hour wab prodouble current distribution apparatus duced from linseed methyl esters. One drawback in the use of CDCD is the large volume of solvent needed (6). the product in run 4 was apparently due For the work reported here, in which t o an overloading of the instrument. the instrument was operated a t about Although run 4 was a t nearly the same 22 transfers per hour, 75 ml. of solvents feed rate as runs 5 and 6, the solvent was used a t each transfer; consevolume per tube was about half that of quently, more than 1.5 liters of solvent runs 5 and 6. was consumed each hour. Distillation As shown in Figure 2, the partition and recovery of solvents and returning coefficient (concentration in the upper them t o the 3-liter reservoirs of the
I )
88
instrument required the continuous attention of one person. To eliminate this requirement and complete the automation of the apparatus, two stills (Figure 4) were designed to provide continuous solvent and product recovery. Though similar t o one described by Bush (Z), these stills accept batch-type samples from CDCD rather than a continuous flow. They also store the stripped samples under nitrogen and lift the solvent vapors so that after condensation the solvents can flow by gravity back into the CDCD apparatus. Another advantage of this continuous solvent-recovery system is the reduction in solvent inventory required and the improvement in laboratory safety. A problem inherent in the procedure is the inevitable concentration of pigmented polar impurities with the linolenate. Because their partition coefficients are much lower than the linolenate, they can be separated readily from the linolenate by a two-stage separatory funnel extraction with acetonitrile and hexane. The recovered product is then distilled a t less than 0.1 mm. of mercury to yield linolenate of high purity.
ANALYTICAL CHEMISTRY
-
ACKNOWLEDGMENT
We are grateful for the help received from F. J. Castle in designing and making the stills and from L. R. Coulson and S. J. Littlejohn in running the many analyses. LITERATURE CITED
(1) Beal, R. E., Sohns, V. E., Eisenhauer, R. A., Griffin, E. L., Jr., J . Am. Oil Chemists' SOC.38, 524 (1961). ( 2 ) Bush, &I. T., Anal. Biochem. 1. 274 (1960): (3) Butterfield, R. O., Dutton, H. J., AXAL.CHEM.36, 903 (1964). (4) hIcCutcheon, J. W., Org. Syn. 22, 82 (1942). (5) Parker, W. E., Swern, D., J . Am. Oil Chemists' Soc. 34, 43 (1957). . (6) Post, o., Craig, L. c., ~ A L CHEM. 35, 641 (1963). (7) Privett, 0 . S., Nickell, E. C., J . Am. Oil Chemists' SOC.40, 189 (1963). (8) Riemenschneider, R. W., Herb, S. F., Kichols, P. L., Jr., Ibid., 26, 371 (1949). (9) Scholfield, C. R., Kowakowska, Janina, Dutton, H. J., Ibid., 37, 27 (1960).
RECEIVEDfor review July 26, 1965. Accepted November 12, 1965. 16th Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Pittsburgh, Pa., March 1965. The Northern Laboratory is headquarters for the Northern Utilization Research and Development Division, Agricultural Research Service, U. S. Department of Agriculture. Mention of trade or company names is for identification only and does not imply endorsement by the department.