( 9 ) Fukushima,
LITERATURE CITED
( 1 ) Brooks. C. J. W.. Biochem. J . 92, 8P ' (1964). ' ( 2 ) Brooks. C. J. W., in "Determination of Sterols," Monograph 2, Society for Analytical Chemistry, London, 1964. ( 3 ) Brooks. C. J. W.. Hanaineh., L.., Biochem. J . 87, 151 (1963). ( 4 ) Clavton. R. B., A'ature 192.1071; 193, 524 ( i 9 6 l ) , Bzochemzstry 1 , 357 (1962). ( 5 ) Darlington, D., Knights, B. A., Thomas, G. H., J . Gas. Chromatog. 1 , 2 1 (1963). ( 6 ) Iljerassi, C., Lenk, C. T., J . -4m. Chem. Soc. 76, 1722 (1954). ( 7 ) Ettre, L. S.,ANAL. CHEM.36, 31A (1964). (8) ELIW, J. v., Reichstein, T., Helv. Chtm. ilcta 24, 879 (1941). \ - ,
~
D. K., Dobriner, S., Heffler, AI. S., Kritchevsky, T. H., Herling, F., Roberts, G., J . A m . Chem.
Soc. 77, 6585 (1955). (10) Glazier, E. R., J . Org. Chem. 27, 4397 (1962). ( 1 1 ) Hanaineh, L., Ph.D. thesis, Glasgow Uiiiversity, 1964. (12) Horning, E. C., Vanden Heuvel, W. J. A , , Creech, B. G., Methods Bzochem. Anal. 11, 69 (1963). (13) Huang-Alinlon, Wilson, E., Wendler, N. L., Tishler, A l . , J . Am. Chem. Soc. 74, 5394 (1952). (14) Kliman, B., Foster, B. W., Anal. Bzochem. 3 , 403 (1962). (15) Knights, B. h., Thomas, G. H., S a t u r e 194, 833 (1962); A N A L .CHEY. 34, 1046 (1962). (16) Nelson J. P., J . Gas Chromatog. 1, 27 (1963).
(17) Peterson, D. H., Aleister, P. I)., Reineke, L. Al., Eppstein, S. H., AItirray, H. C., Leigh Osborn, H. AI., J . Am. Chem. SOC.77, 4428 (1955). (18) Reichstein, T., Flichs, H. G., Helv. Chim. i l c t a 23, 684 (1940). (19) Rosenfeld, R. S.,Lebeau, 11. C., Jandorek, R. I)., Salumaa, T., J . Chromatog. 8, 355 (1962). (20) Shoppee, C. W., Reichstein, T., Helv. Chim. Acta 23, 729 (1940). (21) Tiirner, R. B., J . Am. Chem. Soc. 75, 3489 (1953). (22) Vanden Hetlvel, W. J. A , , Horning, E. C., Riochim. Riophys. ;lcta 64, 416 (1962). (23) Wotiz, H. H., Ibz'd., 63? 180 (1'36%). (24) Wotiz, H. H., Naiikkarinen, I., Carr, H. E., Jr., Ibid., 53, 449 (1961). RECEIVED for review December 16, 1964. Accepted February 4, 1965.
Micro Vapor-Phase Hydrogenation Accessory for Gas Chromatographic Analysis of Fatty Acid Esters of Glyceride Oils T. L. MOUNTS and
H. J.
DUTTON
Northern Regional Research Laboratory, Peoria, 111. Complex mixtures of polyunsaturated and saturated fatty acids, as they occur in glycerides of natural origin, have been simplified b y hydrogenation of double bonds before introduction of their esters into a gas chromatographic column, A micro vapor-phase hydrogenation accessory for attachment a t the injection port of gas chromatographic equipment has been developed. Its use combines the hydrogenation and chromatography steps and provides a rapid technique for simplification of complex mixtures of fatty acid esters. Application of this technique to a variety of oils is described and results are compared with analysis of the products of the same oils obtained from separate batch hydrogenations. This accessory should find significant application in a variety of fields of industrial, biological, analytical, and chemical research.
hydrogenation of olefins on fised-bed columns of catalyst has been described ( 7 ) , but not the applicability of vapor-phase hydrogenation to the analysis of unsaturated fatty acid esters. On the contrary, Beroza and Sarmiento ( 3 ) showed that organic esters split to generically related hydrocarbons. They exploited this discovery for the micro-scale structural determination of organic compounds by analyzing the condensed hydrocarbons with a APOR-PHASE
gas chromatograph. Microreactors used in conjunction with gas chromatographic equipment have been described by Hall, MacIver, and Weber (6) for evaluation of catalyst activity, by Kokes, Tobin, and Emmett (la).. for study of catalytic reactions, by Lindeman (14 ) for determination of olefins in the analysis of cracked gasolines, and by US (4) for investigation of the kinetics and selectivity of hydrogenation. In the analysis of glyceride oils, James and Wheatley (10) early showed that mixtures of saturated and unsat'urated fatty acid esters could be simplified by bromination of the unsaturation before analysis on a gas chromatograph and that identifications could be facilitated by comparison of the chromatograms before and after halogenation. Farquhar et al. (5) suggested microhydrogenation of unsaturated fatty acid esters before chromatographic analysis as a separate batch-type reaction. They stated that, in addition to determining which compounds of the original mixture were unsaturated, the chain length of unsaturated acids can be shown conclusively by the chroniatograph of the hydrogenation products. Recently Korn (13) applied this method of analysis in studying the fatty acids of Acanthamoeba species, and Klenk and Eberhagen (11) used it in determining the composit'ion of fish oils. In all the work cited, hydrogenation of the fatty acid esters of the glyceride oil has been conducted as a separate
batch reaction from which samples were taken for analysis by gas chromatography. I n contrast, we applied a micro vapor-phase hydrogenation accessory to gas chromatographic equipment that unifies the hydrogenation and chromatography steps and facilitates the use of this method in the analysis of glyceride oils. The analpis of a complex mixture, such as herring oil, as well as of the simpler soybean oil, is presented as applications of our analytical technique. EXPERIMENTAL
Apparatus and Materials. The microreactor has been described ( 4 ) . Catalyst. Three catalysts were used: nickel, platinum, and palladium, each prepared to contain 1 to 3y0 by weight of the metal. Each was packed in the microreactor and hydrogen gas was passed through the tube at 200' C. for 1 hour. h chloroform slurry of a reduced nickel cat'alyst (Girdler, G-52 nickel on kieselguhr) was evaporated to dryness in contact with 60- to 80-mesh Chromosorb P on a Rinco rotating evaporator and dried a t 100" C. The platinum and palladium catalysts were prepared as described by Ikroza ( 3 ) . Aisolution of palladium chloride in 5% aqueous acetic acid (or platinic chloride in water) was similarly evaporated in contact with the catalyst support and dried. Several catalyst supports were investigated before selection of ChronioVOL. 37, NO. 6, M A Y 1965
* 641
Table I.
Fatty Acid Composition of Herring Oil
Chain length 14
Double bonds 0
16
0 1 ? 0 1
18
Literature values (21)
6 13 6 2 1 22
2 3 4 0
20
1 ?
5 0
22
(9)
7 12 10 0.5 8.0
1
13.0
2
25.5
15
25.5
6
25.5
(70) Present values 8.9 20.1 8.2 (1.1)5 1.3 20.3 2.6
(12.2) (1.9) (10.8)
1 (7.9) 3 5 6 6 24.0 (4,9) a Figures in parentheses have been associated with particular fatty acids based on a semilog plot of relative retention times carbon chain length (16) and comparison of relative retention times to reported retention data for similar columns and temperature (1).
Table II.
Homologs C1a: 0
C,e: 0
c18:o
c*o:o czz :0
Herring Oil Reduced by Micro Vapor-Phase Hydrogenation
Summed isologs 8.9 29.4 24.2 24.7 12.8
'
Hydrogenation, av. f u MVP Batch 6.9 f 0.5 6.7 f 0.4 27.9 f 0 . 8 27.8 f 0 . 6 25.2 f 1 . 1 25.9 f 1 . 1 22.7 f 1 . 1 22.7 f 1 . 9 17.3 f 1 . 6 16.8 f 1 . 1
Timc, min. Figure 1. Separation of methyl esters of fatty acids of soybean oil
sorb P. Celite and kieselguhr were unsatisfactory, since they adsorbed the sample, causing tailing of peaks. Micro-glass beads and Chromosorb W did not cause tailing of peaks and are suitable supports for the catalyst. Gas Chromatography. 4 6-foot, 1/4-inch 0.d. aluminum column containing 15Q/, EGSS-X on 100-120 Chromosorb P (Applied Science Laboratories, State College, Pa.) was maintained a t 175' C . in a n Aerograph instrument equipped with a dual-filament thermal-conductivity detector. Hydrogen serves as the carrier gas and is introduced through the catalyst tube into the column. The signal from the thermal-conductivity detector is fed into a Ridgefield electronic integrat'or (Esterline Angus Instrument Corp., Indianapolis, Ind.) which records the area under each peak in the curve. Sample Introduction. Microsyringes, No. 701-K (10 pl.) and No. 7001 (1 ~ 1 . ) with ) 4-inch needles (Hamilton Co., Whittier, Calif.), were used to introduce the methyl esters into the microreactor. Various effective catalyst bed depths, up to 43 mm., are obtained by slipping calibrated sleeves over the needle before puncture of the septum. Procedure. Methyl esters of fatty acids of glyceride oils were prepared with sodium methoside catalyst and methanol ( 1 5 ) . The catalyst bed is maintained at 200" C. and the hydrogen flow rate a t 60 cc. per minute. Normal injection, designed to yield saturated 642
ANALYTICAL CHEMISTRY
products, utilized the full 43-mm. depth of the catalyst bed. By use of the 4inch needle the sample may be injected at 0-mm. depth of catalyst bed and no hydrogenation takes place. ,Methyl esters were injected at 43-mm. and 0mm. depth of catalyst bed. To avoid premature vaporization, the sample is drawn out of the needle and into the syringe. The amount of esters used was 0.1 to 2 pl., depending on the complexity of the oil to be analyzed. A standard National Institutes of Health (NIH) mixture of CIO-C~L) saturated fatty acid ester was injected periodically (at 43- and 0-mm. depths) to verify specific retention times and to identify the hydrogenated products. The retention time relative to C14ester increased slightly, less than 1% for the C20 ester, because of the increased length through which the sample must pass. The area of each peak was multiplied by the square root of the molecular weight of the component before calculating the per cent contribution of each compound in the mixture. This calibration correction for a thermal conductivity detector was suggested by Horning et al. (8). To ascertain that the esters are completely hydrogenate3 in the microreactor, samples of the oils were hydrogenated in a manometric hydrogenation apparatus at 30-33' C. with 1% palladium-on-alumina catalyst a t atmospheric pressure and iodine values were determined to be zero. Samples of these preparations were injected at 0- and 43-mm. depth of catalyst bed.
a.
b.
Injection a t 0-mm. depth of catalyst bed 1. Palmitate 2. Stearate 3. Oleate 4. Linoleate 5. Linolenate Injection a t 43 mm. 1. Palmitate 2. Stearate
RESULTS
Application of the procedure to methyl esters of soybean oil is illustrated in Figure 1. When injection is made at 0-mm. depth of catalyst bed, the gas chromatographic separation of the esters (curve a) indicates the expected pattern of unsaturated and saturated C16andCISfatty acid esters. When the same ester mixture is injected at 43 mm., curve b is obtained and shows only a mixture of C16 and C18 fatty acids. Quantitative analysis of curve b indicates that peak 2, associated with Cls saturated acid, contains 88% of the sample. The sum of the percentages obtained for peaks 2 to 5 in curve a, identified by retention times as the saturated and unsaturated isologs of C18 fatty acid, is approximately 88%. Herring oil presents a much more complex problem analytically. The chromatographic analysis of the methyl esters was compared to values reported in the literature (Table I) to indicate that our data were comparable
Xll
x:
1
13 4 5
x4
It
8 9 10
b ?
I2
IS
20
10
30 Til.,
Figure 2.
I4
I
I
10
50
6
111.
Separation of methyl esters of fatty acids of herring oil a.
b.
to those obtained previously. Differences are caused in part by the natural variance between different fish oil samples. Chromatograms obtained when the ester mixture is injected a t 0 and 43 mm. are illustrated in Figure 2. Curve 6 demonstrates that herring oil is a mixture of fatty acids having chain lengths of 14, 16, 18, 20, and 22 carbon atoms. The sum of the percentages for the isologs of each chain length'before hydogenation should approach the percentage obtained for a particular saturated homolog after hydrogenation. The summed values are compared to the values obtained by each of the two hydrogenation methods in Table 11. Micro vapor-phase hydrogenation of menhaden oil was also accomplished; the analyses obtained were as good as those reported for herring oil. When the fully saturated products obtained by the manometric hydrogenation technique were injected a t 0-mm. depth of catalyst bed, chromatograms were identical to those when the methyl ester mixture was injected a t 43 mm. Therefore complete hydrogenation is achieved in the microreactor. Although the nickel catalyst was used in the microreactor for the hydrogenations presented in Figures 1 and 2, the experiments were repeated with the platinum and palladium catalysts. Similar results were obtained.
Injection a t 0 mm. 1. Myristate 2. Palmitate 3. Cl6:l 4. C16:?,5 5. Stearate 6. O l e a t e 7. linoleate 8 to 14. Clo and C22 esterr Injection a t 43 mm. 1. Myristate 2. Palmitate 5. Stearate 8. Arachidate 12. Behenate
DISCUSSION
Micro vapor-phase hydrogenation has been shown to convert unsaturated fatty acid esters in mixtures with saturated fatty acid esters to the corresponding saturated homologs. The values for the long-chain esters are slightly higher after hydrogenation. This may indicate that the highly unsaturated long-chain compounds are partially polymerized on the separatory column and therefore the value obtained after MVP hydrogenation is a more accurate figure for the amount of long-chain ester compound in the sample. The efficacy of the MVP hydrogenation method is apparent when one compares the analyses of the saturated methyl esters with those from the separate batch hydrogenation. For a statistically significant difference a t a 95% level, the means for batch and micro vapor-phase hydrogenation procedures would have to differ by more than 1.5 times the average standard deviation. In no instance does the difference in means exceed 1.5 times the average standard deviation; therefore, it is concluded that within experimental error, micro vapor-phase hydrogenation yields the same results as batch hydrogenation. A significant saving of time and ease of operations results from using this technique over a sepa,rate batch hydro-
genation for the analysis of complex mixtures of methyl esters. As proposed by Beroza ( d ) , the high-speed hydrogenation is no doubt caused by the small amount of injected compound compared to the relatively large amount of catalyst and to the vapor state of the compound, in which it is immediately available for adsorption and reaction on the catalyst surface. There is an apparent conflict in the findings of Beroza and ours. He reported fission of even the shortest ester molecule, while we observed vapor-phase hydrogenation. I n examining the respective conditions of the experiments, however, the anomaly is resolved. Our catalyst temperatures were mild in comparison, being maintained a t 200" C., rather than 280" C. ; while the flow rate, about 60 ml. per minute, was three times faster. These factors appear to lead to the observation of hydrogenation of unsaturation rather than fission of the molecule. Our conditions favor hydrogenation, whereas Beroza's favor fission. LITERATURE CITED
(1) Ackman, R.G.,J. Gas Chromalog. 16, 11 (1963). (2) Beroza. M..ANAL. CHEM.34. 1801 (1962). (3) Beroza, M., Sarmiento, R., Ibid., 35, 1353 (1963). I
,
'
VOL. 37, NO. 6, M A Y 1 9 6 5
643
(4) Dutton, H. J., Mounts, T. L., J . Catalysis 3,KO.4, 363 (1964). ( , 5 ) Faryuhar, J . W., Insull, W., Rosen, P., Stoffel, TI'., Ahrens, E. H., S u t r . A b a t ~Rev. . 17, No.8, Part 11, 1 (1959). (6) Hall, W. K., NacIver, 1). S., Weber, H . P., l a d . Eng. Chem. 5 2 , 421 (1960). ( 7 ) Hoelscher, H. E., Poyner, W. G., Weger, E., Chem. Revs. 54, 575 (1953). ( 8 ) Homing, E. C., Ahrens, E. H., Jr., Lipsky, S. R., Rlattson, F. H., Mead, J . F., Turner, D. A., Goldwater, U'. H., J . Lipid Research 5, 20 (1964). ( 9 ) Humko Products, Sterick Bldg., Memphis, Tenn., Humko Tech. Bull., 1963.
(10) James, A . T., Wheatley, 5'. R., Biochem. J . 63, 269 (1956). (11) Klenk, E., Eberhagen, D.. HovveSeylers Z . Physiol. Cheh. 328, KO. 3-6, 180 (1962). 112) Kokes. R . J.. Tobin. H.. Jr.. Emmett. P. H.,J . A m . Chem. S O C . '5860 ~~, (1955). (13) Korn, E. D., J . Biol. Chem. 238, 3584 (1963). (14) Lindeman, L. P., Chem. Eng. .Vetus 40, N o . 38, 61 (1962). (15) Luddy, F. E., Barford, 11. A., Riemanschneider, R . W.,J . .4m. Oil Chemists' SOC.37, 447 (1960).
(16) Woodford, F. P., \.anGent, C. R l , , J . Lipid Res. 1, 188 (1960). RECEIVEDfor review September 21, 1964. Accepted Rlarch 4, 1965. Pittsburgh Conference on Analytical Chemistry and
Applied Spectroscopy, Pittsburgh, Pa., March 2 to 6, 1964. Article not copyrighted. The Northern Laboratory is part of the Korthern Utilization Research and Development Division, Agricultural Research Service, U. S. Department of Agriculture. hlention of firm names or trade products does not imply that they are endorsed or recommended by the Department over comparable products of other manufacturers.
Determination of Sulfur-Compound Distributions in Petroleum Samples by Gas Chromatography with a Coulometric Detector RONALD L. MARTIN and JOHN A. GRANT Research and Development Department, American Oil Co., Whiting, Ind.
b The distributions of sulfur compounds in a variety of petroleum' samples were determined b y a combination of gas chromatography and microcoulometric sulfur detection. The detection system responds to sulfur compounds but not to hydrocarbons. Under optimum operating conditions, it quantitatively determines individual sulfur compounds at sulfur levels down to about 5 p.p.m. Gas chromatographic separations are made on a siliconerubber column, from which sulfur compounds are eluted nearly in order of boiling point. Sulfur-compound distributions b y boiling point are shown for gasoline, coke-still naphtha, light catalytic cycle oil, virgin naphtha, kerosene, and gas-oil, Only in the lower-boiling samples can individual sulfur compounds be determined. For a crude oil, gas chromatography-in conjunction with other analytical techniques-has been used to determine the distribution of the five principal sulfur-compound types as a function of carbon number to C20. These data illustrate a typical sulfur distribution for virgin petroleums.
I
of the types and distributions of sulfur com1)ounds is needed in many phases of ~ ~ e t r o l e uprocessing. m As a ineans for acquiring such knowledge, gas chroniatogt,aIihj. with selective niicrocoulometric sulfur detection was investigated. 'I'he coulometric detector reslmnds to comllounds containing sulfur but not to hj.di.oc~arbonh,and thei~forecan determinp tracc amounts of sulfur com1)ounds. even if hydrocarbons are eluted with t hcni. KCRISASEI) KNOWLEDGE
644
ANALYTICAL CHEMISTRY
hficrocoulometric sulfur detectors for gas chromatography were developed independently by Coulson and Cavanagh ( 7 ) , and Klaas (IS). The detector of Coulson and Cavanagh, which followed the design of an earlier halide detector (6),has been used extensively in the analysis of pesticides ( 4 , 5 , 7 )and was suggested for petroleum analysis ( 7 ) . Klaas ( I S ) developed a similar detector and successfully applied it to petroleum samples; selective gas chromatography columns were used in clever fashion to determine sulfur-compound types- in .naphthas. Fredericks and Harlow (9) also used a coulometric detector for sulfur compounds; they modified the "halide" detector (6) to accurately and selectively determine thiols in natural gas. This type of detector was not used in our study because it does not respond to other types of sulfur compounds. In this work, microcoulometric sulfur detection was tested for quantitative performance, interferences, and applicability to all types of petroleum samples. Sulfur distributions by boiling point for seven types of petroleum samples were determined by the combination of gas chromatography and coulometric detection. The gas chromatographic separations were obtained with conventional nonselective columns. Analyses for sulfur-compound types were obtained by other techniques to supplement the gas chromatographic data. For a Middle-East crude, results from gas chromatography, mass spectrometry, and liquid-solid chromatography were combined to obtain distributions of the five principal sulfur types as a function of carbon number to Cz0.
MICROCOULOMETRIC SULFUR DETECTOR
The microcoulometric detection system is manufactured by Dohrmann Instrument Co. (Model C-100) according to the design of Coulson and Cavanagh (6, 7 ) . It has two main components-combustion tube and titration cell. The combustion tube is located between the column and titration cell, and serves to osidize the column effluent; sulfur compounds form sulfur dioxide, which subsequently is titrated automatically with coulometrically generated iodine. Hydrocarbons are converted to carbon dioxide and water and generally do not affect the titration. The current used for iodine generation is recorded to give an ordinary differential chromatogram for sulfur compounds. The combustion tube, which is maint,ained a t 750" C., is 10 inches by 5/16 inch and packed with quartz chips. Xitrogen sweep and oxygen flows of 150 ml. per minute each are brought in at the head of the tube along with the column effluent. The detector is sensitive as well as selective; the minimum detectable amount is about gram of sulfur (between t'hat for thermal-conductivity and flame-ionization detectors), and samples with sulfur contents in the partsper-million range can be analyzed. The detector time constant is larger than that of conventional detectors because of the time needed for combustion and titration; this limits the effectiveness for closely-spaced 1)eaks eluted in the first several minutes, but otherwise causes no problems. The detector is easy to operate and maintain. Quantitative Performance of D e tector. F o r quantitative analyses,
