and disappearing species. A balance of increments among experimental values used in Figure 2 was calculated as shown by the example in Table I11 covering the range of 40 to 105 minutes. Over the entire range of 0 to 180 minutes the totals of the increments ~ -128.9 and Z Z F = 132.8. were 2 2 = This is considered a reasonable balance for a valid profile. This method might be applicable generally for the analysis of mixtures containing labile halogen \{-here the reduction products boil below 400" C.
Ger. Patent 569,570 (Feb. 4, 1933). (2) Cheronis, I\J. D., Eiitrikin, J. B., "Semimicro Qualitative Organic Analysis," 2nd ed., p. 331, Interecience, lieu, York. (3) Funke, Albert, Engeler, C. O., Jacob, Joseph, Depierre, France. Compt. rend. 228,716 (1949). (4) E. S. patent applied for. (5) Tomita, M., Kimura, K., J . Pharm. SOC. Japan 70,44 (1950). (6) Ullman, F.,Sponagel, P., Ann. 350, 83 (1906). RECEIVEDfor review May 28, 1,960. Accepted September 21, 1960. Divlslon of Analytical Chemistry. 137th ICIeeting, ACS, Cleveland, Ohio, .ipril 1960.
ACKNOWLEDGMENT
The authors thank E. H. Rosenbrock for the preparation of several of the tolyl ethers used as known standards, H. G. Scholten and D. F. Kisniewski for development work on the utilization of bis[a - (z - tolyl) - p - tolyl] ether column liquid, and J. D. Doedens and H. E. Hennis for helpful suggestions and discussions. LITERATURE CITED
(1)Brunner, Arnold (to I. G. Farbenind.),
Gas Chromatographic Characterization of Fatty Acids Identification Constants for Mono- and Dicarboxylic Methyl Esters THOMAS K. MIWA, KENNETH 1. MIKOLAJCZAK, FONTAINE R. EARLE, and IVAN A. WOLFF Northern Utilization Research and Developmenf Division, Agricultural Research Service, U. S. Department of Agriculture, I8 I5 North University, Peoria, 111. Readily reproducible numerical constants were determined for characterizing mono- and dicarboxylic methyl esters by gas-liquid chromatography. For a specified column packing and carrier gas, these constants, called equivalent chain lengths, are independent of experimental conditions. A combination of two values, obtained by use of polar and nonpolar packing, is sufficient to characterize most fatty acids.
c
identification of components on a gas chromatogram is established by comparing retention times. volumes, or ratios with those of standards. These retention characteristics, however, vary according to experimental conditions-e.g., changes in column temperature. James (6) has proposed a graphical means of identifying homologs using curves based on retention ratios that are subject to change in slope when the ratios change. I n screening seed oils to select those having industrially interesting fatty acid compositions (S), retention values that remained constant regardless of changes in experimental conditions were needed to characterize known and unknown components. A scheme was devised whereby each coniponent on the chroniatogram was expressed by a chain length equivalent to a homolog of the saturated straight-chain monocarboxylic acids. For a specified column packing and carrier gas, in the range where the relationship between molecular u-eight of the reference saturated straight-chain monocarboxylic methyl esters and logarithm of their ONVENTIONALLY,
temperature, but the E.C.L. remained constant. This procedure IT-as applied to a variety of fatty acids from oils containing conjugated unsaturation or hydroxy. epoxy. cyclopropenoic, carbonyl, and other groups contributing to the complexity of their qtructure. Dicarboxylic acids commonly encountered in degradative studies of unsaturated fatty acids were characterized in similar manner. After submitting an abstract to present this paper, describing the E.C.L. procedure used here since January 1959, an article by Koodford and van Gent (15) came to our attention in which they independently proposed a similar scheme, and refer 'to their parameter as a "carbon number." I n the present article me provide con-
retention times was linear, interpolation between the retention times of the reference esters permitted determination of an equivalent chain length (E.C.L.) for each component. This E.C.L. value was independent of operating conditions such as column temperature, carrier gas flow rate, and column diniensions. I n practice, a reference curve was established by plotting on a semilog graph the retention times (log-scale) of two or more known, normal, saturated monocarboxylic methyl esters against their chain lengths (number of carbons in the acid). Values for components of subsequent samples put through under the same operational conditions were then read from the curve using observed retention times. The slopes of the curves varied mith changes in column
Table I.
Conditions for Experimental Comparison of Retention Values
Col. Structure (Glass, U-
Set A
Shaped), I.D. X Length, Cm. 0 6 X 275
D
Same as 9 0 6 X 125 0 3 X 125
E
0 6 X 200
F
Same as E
B C
of Helium,
R.ll./hIin. 30 (40p.5.i.) 90 (40p.s.i.) 87 (25 p.e.i.) 30 (25 p.5.i.)
Same as A 228 Same as 4
2 1 2 2
100 (40p,s.i.)
190
2.3
Same as E
210
2.8
Flow Rate
Stationary Phase Apiezon L o n Celite 545, 100-150 mesh (20:SO) Same a s A SameasA Same as 4 ,except 60-100 mesh (15:85) Resoflex 446 on Celite 545, 100150 mesh (20:80) Same a s E
Theor. Plates (Palmitate), Thousand 30
Temp. of Column +th, C. 250
VOL. 32. NO. 13, DECEMBER 1960
5 0
Q
1739
Table II. Retention Values Under Various Operational Conditions Methyl Methyl Methyl Methyl Arachi- Methyl Retention Behenate Seta Myristate Palmitate Stearate date Characterietics 24.50 46.80 172 5 d 12.95 90.0 Time, minutes 15.80 B 4.42 30.10 8.30 57 45 18.27 37.10 C 4.55 9.08 77 0 3.02 D 0.90 11 20 1.56 5.82 14.99 51 34 27.70 E 7.25 12.54 21 71 F 1.40 A 0.39 0.74 2.70 5 18 Volume, liters 1.42 ti 0.40 5 17 0.75 2.71 3.23 C 0.40 1.59 6 70 0.79 D 0.03 0.09 0 34 0.05 0.17 5 13 1.50 2.77 E 0.72 2 17 1.25 F 1.oo Ratio, stearate = 1 .OO A 0.277 0.524 1.92 3 69 1.00 1.90 3 64 B 0.280 0.526 4 22 1.00 2.03 0,497 C 0.249 3 71 1.00 1.93 0.516 D 0.298 1 .oo 3 42 E 1.85 1.00 2.99 F 1.73 E.C.L. B 14.06 22 00 18.00 20.00 16 00 B 14.02 22 00 18.00 20.00 15.98 18.00 C 14.13 22 00 20.00 16.00 D 14.26 22 00 18.00 20.00 15.96 18.00 E 20.00 22 00 F 18,OQ 20.00 22 00 I
See Table I.
siderably more detail, especially on the utility and applicability of the method to a wide variety of lipide materials and to the consistent changes in parameters ascribable to specific functional groups. APPARATUS A N D MATERIALS
G a s Chromatograph. A Burrell Kromo-Tog K-5 equipped with thermal conductivity detector cells was used. The columns are described in Table I. Column Packings. llpiezon L grease and LAC-2-R 446 polyester (diethylene glycol-pentaerythritol adipate or,Resoflex 446) were the liquid phases coated on Johns-Manville Celite 545 solid support, previously acid-washed, neutralized, and dried. Fatty Acid Methyl Esters. Methyl esters used as standards were purchased from commercial sources. Others were prepared from fatty acids reacted with diazomethane (1) or from oils transesterified with anhydrous methanol and catalytic amounts of either acid (HC1) or base
(NaOCH3).
PROCEDURES
Determination of Retention Values. Retention times were measured from the instant t h e samples were ejected from the syringe. T h e time elapsed from insertion of t h e needle to ejection of the sample was approximately 2 seconds. Retention ratios were calculated relative to the retention time of stearate. E.C.L. was determined as described. The standard solution contained methyl esters of 14 normal saturated monocarboxylic acids ranging from Cg to 6 2 2 .
Inclusion of the short-chain fatty acids facilitated the tentative identification of components which emerged at the very beginnling of the run where linear relationship between molecular weight and logarithm of retention time did not hold. erirnental Conditions on ~ e t ~ Values. n ~ ~Column o ~ dimension, column temperature, and carrier gas flow rate were varied to study their effects on retention values. Six sets of conditions maintained for Apiezon L and Resofiex 446 columns are listed in Table I. The number of theoretical plates was calculated, using the conventional 16 ( X / U ) ~ formula, based on the palmitate peak (3). Characterization of Patty Acids. T h e use of liquid phases differing in chemical nature--.e.g., t h e nonpolar Apiezon L hydrocarbon grease and the polar Resoflex 446 polyester resin-was necessary to obtain. values which showed the characteristics of the acids as they differed in structure from the normal saturated homobgs. Identification of the less common fatty acid methyl esters prepared directly from seed oils was based on pre!iminazy chemical and spectrophotometric analyses, as well as on their gas chromatographic behavior relative to that of known fatty esters. RESULTS AND DISCUSSION
Effects of Experimental Conditions on Retention Values. As expected, retention times were t h e least constant of all retention values. Methyi behenate (ez2)required almost 3 hours for elution in A , b u t only 12 minutes in D (Table 11). Retention volumes remained con-
Table Ill. Equivalent Chain Lengths of the Methyl Esters E.C.L. Parent Acid -4piezon Resoflex (Common Xame) L 446 MQPJOCARBOXYLIC METHYL ESTERS" Saturated straightchain C, acid: octadecanoic (stearic) 18.0 18.0 Unsaturated straight-chain acids Monoenoic cis-9-Hexadecenoic (palmitoleic) 15.7 16.4 cis-9-Octadecenoic (oleic) 17.7 18.4 cis-6-Octadecenoic (petroselinic) 17.7 18.4 cis-13-Docosenoic (erucic) 21.7 22.4 Dienoic: cis,&9-12-octadecadienoic (linoleic) 17.6 19.0 Trienoic: cia,cis,cis-9,12,15octadecatrienoic (linolenic) 17.6 19.8 Tetraenoic: cis,cis,cis,cis-6,8,11,14eicosatetraenoic (arachidonic) 19.2 21.6 DICARBOXYLIC METHYL ESTERS Pentanedioic (glutaric) 7.6 12.4 Hexanedioic (adipic) 8.6 13.4 Nonanedioic (azelaic) 11.7 16.4 Undecanedioic 13.7 18.4 Tridecanedioic (brassylic) 15.7 20.3 a Fatty acids used as standards.
stant when the flow rate was the only condition changed ( A and B). Retention ratios were constant to within 1% when column temperature was the same ( A , B, and D). Variation in weight percentage of the liquid phase and mesh Bize of the solid support had no effect on retention ratios. A 22" 6 . reduction in column temperature changed the ratios for the components in set C to the extent that identification by comparison of retention ratios alone was no longer feasible. E.C.L. remained constant throughout the entire experiment. Similar results were obtained for Resoflex 446 columns, sets E and F, as for Apiezon E columns. Chaxacterization of Fatty Acids. The E.C.L. of reference commercial monocarboxylic methyl esters determined for liquid phases Apiezon L and Resoflex 446 are listed in Table III. By nature of the scheme, values for
the normal saturated acids were integers equal to their carbon chain Table IV. E.C.L. of Monocarboxylic Methyl Esters of Fatty Acids occurring in lengths. Values for monoenes mere Plant Seed Oils 0.3 unit less than the corresponding Percentsaturated homologs in Apiezon L and 0.4 unit greater in Resoflex 446. The positional isomers, oleic and petroselinic, Parent Acid (Common Name) Source could not be separated when chromatoSaturated straight-chain acids graphed together. Decanoic (capric) Cuphea llavea (14) 83 10.0 10.0 Dicarboxylic acids most frequently Tetracosanoic (lignoceric) Amburana species (4) 4 24.0 24 0 encountered during fatty acid degradaUnsaturated straight-chain tion studies are also listed in Table 111. acids Monoenoic Conversion of the terminal methyl to a cis-5-Eicosenoic 20.4 19.7 Liinnanthes douglasii ( I O ) 65 carbomethoxy group increased E.C.L. 22.4 21.7 cis-5-Docosenoic 7 Limnunthes douglasii (10) 2.6 to 2.7 units in Apiezon L and 7.3 22.4 21.7 cis-13-Docosenoic (erucic) Limnanthes douglasii (10) 13 to 7.4 units in Resoflex 446. All Eicosenoic 20.4 19.7 8 Rapeseed (4) 22.4 21.7 43 Docosenoic Rapeseed (4) E.C.L. values listed in the tables were 23.7 24.4 Tetracosenoic 1 Rapeseed (4) reproducible to =t0.04 unit. Dienoic E.C.L. of selected fatty acids oc21.0 1 19.G Eicosadienoic Mustard (4) curring in plant seed oils are listed in 23.0 21.6 Docosadienoic 5 Mustard (4) 20.6 Eicosadienoic 0.3 19.4 Table IV. Their source and percentage Limnanthes douglasii 22.6 Docosadienoic 10 21.4 Liinnanthes douglasii in the extracted oil, determined by 20.6 Conjugated octadecadienoic Chilopsis linearis 18.7 11 gas-liquid chromatography, are also Trienoic given. 22.5 19.4 cis,trans,trans-g,11,1379 Octadecatrienoic (aCapric acid (Go),which comprises eleostearic) 83% of Cuphea oil, and lignoceric 22.8 19.7 trans,trans,trans-9,11,13- Tung isomerized acid (CM), found in crucifers, had the Octadecatrienoic (pexpected integral values in both Apieeleostearic) 19.5 zon L and Resoflex 446. All the straight10 17.4 cis,cis,cis-6,9,12-0ctadeca- Oenothera biennh (4) trienoic chain monoenes (cis-isomers) had E.C.L. Hydroxy acids identical or homologous to the standard 24.7 19.4 12-Hydroxy-cis-9-octadec83 Castor (4) nionoenes. The structural proof of the enoic (ricinoleic) 22.7 Trace 17.4 monoenoic isomers in Limnanthes oil Hydroxyhexadecenoic Lesquerella lasiocarpa 24.7 3 19.4 Hydroxyoctadecenoic Lesquerellalasiocarpa mas established by Smith et al. (10). 26.7 21.4 45 Hydroxyeicosenoic Les uerella lasiocarpa The Czo and C2t dienes occurring in 29.8 21.3 Oxiiation of cis-g-octa9,lO-Dihvdroxvoctadecanoic mustard seed oil had values exactly decenoic 9-Hydroxy-trans,trans-10,12- Dimorphotheca aurantiaca 65 20.3 27.2 2 and 4 units greater than those for the octadecadienoic (dimor(12) common Cle linoleic acid. This patphecolic ) tern suggests a homologous relationEpoxy acids ship among the three acids. The CZO 74 19.2 23.4 12,13-Epoxy-cis-octadecenoic Vernonia anthelmintica and C2zdienes occurring in Limnanthes Epoxidation (11.) of cis-9-octa19.6 23 0 9,lO-Epoxyoctadecanoic oil differed from each other by exactly decenoic [natural occur2 units, but they had a lower E.C.L. rence reported (1s)] than the linoleic homolog, indicating Oiticica ( 4 ) 32 21.0 24.6 Keto acid: 4keto-9,11,13structural differences which have been octadecatrienoic (licanic) StercuZia foetida (4,9) 35 18.6 20.0 Cyclopropenoic acid: 8-(2-nfound to be true. The chemical octylcyclopro 1-eny1)ocstructure of the (3%diene in Limnanthes tanoic (stercuKc) has been established and will be reAcetylenic acids . ported. Dehydrobromination of 17.9 19.8 9-Octadecynoic (stearolic) 9,lO-dibromostearic The conjugated CI8 diene, believed Picramnia pentandra (4) 85 17.9 19.8 6-Octadecynoic (tariric) to be the trans, trans-isomer (8),had values 1.1 (Apiezon L) and 1.6 (Resoflex 446) units greater than the nonconjugated cis,&-linoleic acid. The conjugated all-trans-Clg triene had valr are the differences between ricinoleic (Apiezon L) and 0.3 (Resoflex 446) unit. ues 2.1 (Apiezon L) and 3.0 (Resoflex and oleic. The deviations of 9,10The hydroxymonoenes occurring in 446) units greater than the noncondihydroxystearic from stearic are $3.3 castor and Lesquerella oils all deviated jugated all-cis-linolenic acid. Isomerand +11.8 units, approximately double in a similar manner from their corization of a methylene interrupted the values calculated for the 12-hyresponding nonhydroxylated saturated cis,cis-linkage to a conjugated trans, droxy compound. The deviations of homologs. When the major fatty acid trans-linkage is therefore approxdimorphecolic from etearic are in good component in Lesquerella oil was found imately equivalent to an increase in agreement with the sum of the deviato have E.C.L. exactly 2 units greater chain length of 1.0 unit in Apiezon L tions for a single hydroxy and a conthan those of ricinoleic in castor oil, and 1.5 units in Resoflex 446. Isomerijugated trans,trans-diene; Apiezon L, it was predicted that this component zation of cis,trans,trans- t o trans,trans, 2.3 vs. 2.4, Resoflex 446, 9.2 us. 8.9. was a GZo homolog of ricinoleic acid. trans-conjugation increased the E.C.L. Epoxyoleic and epoxystearic difThe results obtained thus far by chemof eleostearic acid by 0.3 unit in both fered in E.C.L. in the same way as ical and spectrophotometric characterApiezon L and Resoflex 446. A shift oleic did from stearic. The contribuization are in complete agreement with in cis-double bonds toward the cartion of the 12,13- and 9,lO-epoxy conthis prediction. bonyl end as the 9,12,15- to the 6,9,12stituents is calculated to be 1.5 and 1.6 The contribution of the 12-hydroxy units in Apiezon L and 5.0 and 5.0 group is calculated to be 1.7 (Apiezon octadecatrienoic acid was equivalent units in Resoflex 446, respectively. L) and +6.3 (Resoflex 446) units, which to a decrease in chain length of 0.2
+
VQL. 32, NO. 13 DECEMBER 1960
B
1741
Comparison of licanic with 0-eleostearic shows differences of 1.3 (Apieaon L) and 1.8 (Resoflex 446) units, ascribable t o the 4-keto-constituent, The $10cyclopropenoic structure (9) in sterculic (G9) acid is apparently equivalent to the cis,cis-9,lZ-dienoic structure in linoleic acid; the difference is exactly 1 unit in both Apiezon L and Resoflex 446. This similarity accounts for our inability to separate the homolog malvalic acid from linoleic acid when Apieaon L and Resoflex 446 are the liquid phases. The acetylenic position isomers, stearolic and tariric acids, have identical E.C.L. In the chromatogram of methyl esters of isano oil (Figure I), the conventional retention time on the abscissa has been replaced by E.C.L. Values are based on the retention times of reference palmitate, stearate, and arachidate peaks. No previous report on the gas chromatographic analysis of isano oil is known, but components a (29%) and b (44%) are believed to be isanolic acid (Apiezon L = 17.1, Resoflex 446 = 24.6), and isanic acid (Apiezon L = 19.8, Resoflex 446 = 25.4), respectively ( 7 ) . ACKNOWLEDGMENT
The authors are indebted to 31. 6. Burnett, Glenda C, Geisinger, Clara E. McGrew, and Bonita R. Heaton for the chemical analyses and to R. L. Lohmar, C. R. Smith, Jr., T. IL. Wilson, and H. J. Dutton for the preparation
Use of CI 1.
P.
4E
b
19.8
Oleic
0.15
A
-4
j i v a i e n t Chain l e n g t h
Figure 1. Gas chromatogram of isano oil methyl esters (Apiezon 1)
of the less common fatty acid methyl esters. LITERATURE CITED
( 1 ) Arndt, F., Blatt, A. H., eds., “Organic Synt,heses,” Coll. Vol. 2, p. 165, Wiley, New York 1943. (2) Desty, b. H., ed., “Vapour Phase
Chromatography,” p. xii, Academic Press, New York, 1957. (3) Earle, F. R., Melvin, E. H. Mason, L. H., Van Etten, C. H., Wofff, I. A, J . Am. Oil Chemists’ SOC. 36, 304 (1959). (4) Eckey, E. W., “Vegetable Fats and Oils,” Reinhold, New York, 1954. (5) Greenfield, J., J . Am. Oil Chemists’ SOC.36,565 (1959). 16) James. A. T.. J . Chromatoa. 2. 552 (1959).
.,
I
.
(7) Kneeland, J. A., Kyriacou, D., Purdv. R. H.. J. Am. Oil Chemists’
SOC.35,361 (1958). (8) Morri~,L. J., et al., Hormel Institute and University of Minnesota, unpublished data, December 1959.
(9) Nunn, J. R., J . Chem. Soc. 1952, 313. (10) Smith, C. R., Jr., Bagby, M. O., Miwa, T. K., Lohmar, R. L., Wolff, I. A., J . Org. Chem. 25,218 (1960). (11) Smith, C. R., Jr., Ko!h, K. ,F., Wolff. I. A.. J . Am. 0 2 1 Chemzsts’ Soc. 36, 219 (1959). (12) Smith, 6.R., Jr., Wilson, T. L., Melvin, E. H.. Wolff, I. -4.. J . Am. Chem. Soc. 82; 1417 (1960): (13) Tulloch. A. P.. Can. J . Chem. 38,204 ’ (i960). (14) Wilson, T. L. Miwa, T. K., Smith, C . R., Jr., J . Am. Oil Chemists’ Soc., .
in prees.
I
(15) Woodford, F. P., van Gent, C. &I., J . Lzpid Research 1 , 188 (1960).
RECEIVED for review July 8, 1960. Accepted October 4, 1960. Preeented at the Gas Chromatography Symposium, Analytical Division, 137th Meeting, ACS, Cleveland, Ohio, April 1960. The mention of firm names or trade products does not imply that they are endorsed or recommended by the Department of Agriculture over other firms or similar products not mentioned.
C
LINDEMAN and J. L. ANNIS
California Research Cop., Richmond, Calif. Conventional magnetic field mass spectrometers can readily be used as auxiliary gas chromatography detectors in a directly coupled arrangement. Such Q system has advantages in the identification of multicomponent chromatographic peaks and of components appearing as shoulders or in the background over trapping procedures. The procedure also gives Q good quantitative breakdown of multicomponent peaks. This is illustrated with known mixtu,es and a previously analyzed reformer charge stock. HE direct connection of a time-offlight mass spectrometer to a gas chromatograph for the analysis of complex chemical mixtures was discussed
1742
a
ANALYTICAL CHEMISTRY
recently b y Gohlke (2). A small portion of the gas entering the thermal conductivity detector was direrted directly into the mass spectrometer. I n a sense, two parallel detectors are being used simultaneously to record separations that occur in the chromatograph column. The advantage of the mass spectrometer is that it identifies the material present in a chromatograph peak and gives a relative measure of how much is present. The direct combination of the mass spectrometer and gas chromatograph is a very powerful tool for the analysis of volatile organic mixtures. The extent to which a conventional magnetic field mass spectrometer could be used for this purpose is, therefore, of interest. It was soon found that the spectra ob-
tained on such an instrument not only were adequate for identifying unicomponent chromatographic peaks but also several components could be detected simultaneously in chromatographic peaks which contained more than one component. The mass epectra obtained could be used to determine the relative concentrations of these components. INSTRUMENTATION
The method used to connect the mass spectrometer and gas chromatograph is shown schematically in Figure 1. A portion of the gases exhausted from the thermal conductivity cell is sampled through a capillary tubing which is connected to the mass spectrometer just
