Quantitative Gas-Liquid Chromatography of Fatty and Resin Acid Methyl Esters F. H. Max Nestler and Duane F. Zinkel U.S . Department of Agriculture, Madison, Wis. 53705
Forest Products Laboratory, Forest Sercice,
Gas chromatographic retention and thermal conductivity detector response data have been obtained for the methyl esters of the principal resin and fatty acids found in pine. Retention data are also included for the esters of a large number of hydrogenated resin acids, as well as for several unusual resin and fatty acids. Special problems with respect to methyl levopimarate are discussed. A two-column system of DEGS and SE-SO/EGiP is suggested as most appropriate for the quantitative analysis of natural resin-fatty acid mixtures containing levopimarate and the higher molecular weight fatty acids.
ALTHOUGH GAS-LIQUID CHROMATOGRAPHY (GLC) iS being used more frequently for the analysis of resin acids [for structure and numbering of resin acids, see ( I ) ] , it has seldom been used for the mixtures of fatty and resin acids commonly found in softwood extracts and tall oil. Since a quantitative separation of resin acids from fatty acids has not been described, it was necessary to develop conditions whereby such complex mixtures could be resolved into their individual components. A previous paper ( 2 ) discussed our preliminary results for the qualitative gas chromatography of resin acid methyl esters in diethylene glycol succinate (DEGS), a liquid phase that is also efficient for the separation of fatty acid esters (3). The present paper, though concerned mainly with the quantitative determination of components in fattyresin acid mixtures, includes additional retention data for the resin acid esters in various liquid phases. EXPERIMENTAL All reference to the use of the various resin and fatty acids applies t o measurements made with the methyl esters. The fatty acid shorthand designation illustrated in Burchfield and Storrs (3) is used--e.g., 18:29312cis,cis, is the code for cis-9,cis-12-octadecadienoic acid, or equally for its methyl ester. Apparatus and Procedure. I n most respects, few major changes are made in those details reported previously (2). Column and detector temperatures were measured potentiometrically by means of iron-constantan thermocouples. The column couple was installed in stainless steel tubing filled with uncoated solid support and placed a t the oven midpoint. The accuracy with which column and detector temperatures could be set was 0.1O C and 0.5’ C , respectively, with control of both t o a precision of +0.1 O C or better. The accuracy of the vaporizer temperature, checked periodically by potentiometer, was within 1-3 C of the pyrometer scale readings. Normal operating parameters, unless noted otherwise, were those used previously (2). Detector filaments were rhenium-tungsten type WX (Gow-Mac Instrument CO.). O
(1) R. McCrindle and K. H. Overton, “Advances in Organic Chemistry,” Vol. 5, Wiley, New York, 1965, pp. 50-4. (2) F. H. Max Nestler and D. F. Zinkel, ANAL.CHEM.,35, 1747 (1963). (3) H. P. Burchfield and E. E. Storrs, “Biochemical Applications of Gas Chromatography,” Chap. 7, Academic Press, New York, 1962.
1 1 18
ANALYTICAL CHEMISTRY
Carrier gas flow rate, FA‘, is the nominal value, uncorrected for water vapor. The instrument room was temperature controlled a t 25” C. Packing and Column Preparation. The support-coating procedure (2) was altered by eliminating the rotary evaporator. Instead, solvent was rapidly evaporated by drawing a high velocity air stream across the top of the heated slurry. This method resulted in a minimum of mechanical working of the coated solid support. The deactivated SE-30 packing (SE30/EGiP) was prepared in a two-step procedure in which 1 EGiP was coated onto the support followed by 9 % SE-30. Commercially available products were used in preparing all packings: SE-30, QF-1-0065, Versamid-900, DEGS (ReEngineering Laborasearch grade, B series)-Analytical tories, Inc. ; PEG-20M, phenyl diethanolamine succinateWilkins Instrument and Research, Inc. ; DC-550-Dow Corning Corp. ; EGiP, QF-1 (FS-I265)-Applied Science Laboratories. Stainless steel U-shaped columns (6 feet X li4-inch 0.d.) were packed to very nearly the same bulk density by allowing coated support to “free fall” into each leg alternately, followed by mild vibration. The uniformity of the packings was checked after installation of the column in the instrument by measuring its permeability for the helium carrier gas at room temperature. Contrary to the usual practice, columns were preconditioned by heating overnight a t less than 10” above the operating temperature and 50 ml/minute carrier gas flow rate. This was particularly important with DEGS, because of the high rate of vaporization of this polyester. Standby (overnight) conditions were 150-175” C and F A ‘ = 25 ml/minute of helium. By continuously collecting column effluent from all packings used in this work, data were obtained which allowed an estimation of the approximate liquid phase loading remaining during the life of the column. Methyl Ester Standards. Fatty acids esters were 99+% pure (Applied Science Laboratories, State College, Pa.). Methylation of resin acids, and of fatty acids when required, was carried out with diazomethane as previously described (2). Silver nitrate-alumina column chromatography ( 4 ) was used in the purification of methyl pimarate, palustrate, and sandaracopimarate. These, and the remaining resin acid ester standards (Table VI) prepared from the pure acids, were filtered through short columns of neutral alumina 111, recrystallized from methanol, and vacuum dried, except that in the case of methyl abietate and palustrate, the crystallization step was precluded as these esters are liquids. Data Treatment. All relative retention values are based on elution times from the air/solvent peak. The determination of peak areas has been critically reviewed (5). I n the present work, areas were measured with a compensating polar planimeter (Dietzgen Catalog No. 1806) and/or by the application of the triangulation method (5) A
= w0.45 *
ho
where peak width, w , was taken a t a height equal to 0.45
(4) D. F. Zinkel and 3. W. Rowe, J . Chromatog., 13, 74 (1964). (5) L. Condal-Bosch, J. Chem. Educ., 41, A235 (1964).
Table I. Relative Retention Values for Resin Acid Methyl Esters on Various Liquid Phases“*b Liquid phase0 SE-30 . .~ EGiP DEGS PDEAS EGiP QF-1 DC-550 Vers 900 PEG-20M Ester SE-30 PIMARATE INTERNAL STANDARD, rPim SECTION A: METHYL 1.00 1.00 1.00 1.00 1.00 Pimarate 1.00 1.00 1.00 1.00 1.25 1.32 1.35 1.39 1.25 1.19 Palustrate 1.15 1.21 1.18 1.27 1.33 1.37 1.21 1.35 Levopimarate 1.18 (1.20) 1.27 1.23 1.48 1.45 ... 1.30 1.41 Isopimarate 1.28 1.22 1.16 1.21 2.00 2.30 2.08 1.46 2.07 Dehydroabietate 1.46 1.26 1.36 1.34 2.09 2.10 2.11 2.07 Abietate 1.58 1.69 1.48 1.58 1.56 2.57 2.47 2.5 1.70 1.89 Neoabietate (2.0) 1.93 (2.6) 1.77 STEARATE INTERNAL STANDARD, rig: 0 SECTION B: METHYL 1.00 1.00 1.00 1.00 1 .oo 1 .oo 18:O 1 .oo 1.00 1 .oo 4.15 4.01 4.32 2.56 2.93 1.52 1.78 1.92 Pimarat e 1.51 5.39 5.62 5.42 4.06 1.80 Palustrate 2.39 3.04 1.83 2.04 5.51 5.50 5.51 Levopimarate 3.95 2.43 3.10 (2.14) 1.80 1.86 __ 6.02 6.40 2.34 3.31 4.12 ... 1.76 1.83 2.27 Isopimarate 9.55 8.62 8.3 2.05 2.38 2.81 3.11 6.04 Dehydrobietate 1.91 8.76 9.04 8.5 2.39 2.77 3.04 4.30 6.04 Abietate 2.25 __ 11.1 (10.0) 10.2 Neobietate 2.58 2.86 3.15 (3.7) 4.93 (7.5) SECTION C: SEPARATION FACTORS, ... ... ... ... ... ... ... ... Pimarate ... Palustrate 1.32 1.02 1.25 1.19 1.03 1.02 1.21 1.15 1.02 Levopimarate 1.02 1.33 1.02 1.02 1.35 1.02 1.05 1.02 1.00 Isopimarate 1.07 1.16 1.00 1.06 1.22 1.07 1.02 1.16 __ __ 1.10 Dehydroabietate 1.35 1.05 1.16 1.12 1.47 1.5 1.06 1.10 Abietate 1.05 1.02 1 .oo 1.45 1.08 1.16 1.17 1.16 1.18 Neoabietate 1.23 1.08 1.15 (1.2) 1.20 1.14 (1 * 2) 1.15 (1.2) SECTION D : EXPERIMENTAL PARAMETERS 225 200 200 240 258 200 200 200 T, “C 200 15 18.7 4.0 8.8 10 20 17 12 20 Load, Zd 5.6 5.9 tE’ (18:O) 13.0 9.2 11.2 1.2 8.5 8.2 4.9 1.13 1.09 0.91 1.00 1 .oo 1.08 a1,2(18:1/18:0) 0.92 1.18 1.00 SE-30 silicone (methyl) rubber gum; SE-30/EGiP, 9 % SE-30 + 1 EGiP; QF-1-0065, a trifluoro propyl methyl silicone; DC-550, a phenyl methyl silicone oil; Vers-900, Versamid-900; PEG-20M, polyethylene glycol 20,000; PDEAS, phenyl diethanolamine succinate; DEGS, diethylene glycol succinate; EGiP, ethylene glycol isophthalate. * Underlined values represent esters displaced from the elution order shown in the “Ester” column. Values in parentheses are in some doubt because of decomposition of the solute or to a long retention time, Values for neoabietate in DC-550 and PEG-20M were estimated; Ar 1 0 . 2 unit. a1.2 values are computed relative to the adjacent previous compound as the standard-Le., as component 2. Liquid phases in order of increasing selectivity. Liquid phase loadings are the approximate actual values as estimated from data on amount of solvent lost from the packing, e Adjusted retention time for 18 :Oat FA‘ = 150 ml/minute for all columns except the Versamid-900,which had FA’ = 92 ml/minute. ~
. . I
Q
-
times peak height, h,. Peak dimensions were obtained with a n accurate steel rule and a 7 X measuring magnifier. The planimeter was the type having variable length pole and trace arms, permitting adjustment of the instrument to maximum sensitivity commensurate with the size of the peak integrated. Calibration with the standard test rule supplied with the planimeter showed the trace arm t o be linear throughout its length. The method pertaining t o compensatory measurements was followed t o obtain maximum instrumental accuracy. I n general, while the graphical method was used for the majority of area measurements, the planimeter was used when peak height and actual peak width-Le., a t the baseline-were approximately equal. Advantage was taken also of the ability t o adjust the peak width by decreasing the chart speed as retention time increased, thus maintaining approximately the same relative error in peak width for early and late peaks. RESULTS AND DISCUSSION
Relative Retention Data. To minimize the possible effect of solute-support interactions upon retention and response data, liquid phase loadings were generally above the 15% minimum recommended by Evered and Pollard (15). The
importance of the interrelationship of the properties of support, liquid phase, solute, and the sample size of the latter upon the retention characteristics of the solute are further illustrated and emphasized by the data of Scholz and Brandt (7). I n particular, their results call attention to the susceptibility of polar o r of polarizable solutes to interact with the support, especially when insufficient liquid phase is present, or when it is not effective in further deactivating the support. Although a number of papers have appeared using gas chromatography t o analyze resin acids (8-14), many used liquid ~
(6) S. Evered and F. H. Pollard, J. Chromatog., 4, 451 (1960). (7) R. G. Scholz and W. W. Brandt, “Gas Chromatography,” N. Brenner, J. E. Callen, and M. D. Weiss, Eds., Chap. 11, Academic Press, New York, 1962. (8) J. A. Hudy, ANAL,CHEM.,31, 1754(1959). (9) T. Norin and L. Westfelt, Acta Chem. Scand., 17, 1828 (1963). (10) E. von Rudloff and A. Sato, Can. J . Chem., 41, 2165 (1963). (11) G. Pensar and H. H. Bruun, Acta Acad. Aboensis Math. Phys. (B), 24(6), 3 (1964). (12) T. W. Brooks, G. S. Fisher, and N. M. Joyce, Jr., ANAL. CHEM.,37, 1063 (1965). (13) G. Valkanas and N. Iconomou, Pharm. Acta Helv., 41, 209 (1966). (14) C. W. J. Chang and S. W. Pelletier, ANAL.CHEM.,38, 1247 (1966). VOL. 39, NO. 10, AUGUST 1967
1119
Pim Ne0 I
SE-30
225*C.
I
2i:o
24:O
Ne0
Pim
SE-JOKGIP
I
Ne0
Pim
I 2o:o !
OF-/
2m.c.
24':0
Pim
Ne0
DC-550 240'C.
I
I
2o:o
I
2OO@C.
24':O
21:o
24:O Ne0
Pim I
VERS.940 258'C f2f4'CJ
I
2i:o
24:O
Ne0
Pim
I
1
22:o
PDEAS ZOO'C.
I
24:O
0
e
L
I
I
3
I 5 RELAWE
I
I
I
I
7
9
it
13
RETENTION,
Figure 1. Schematic diagram of the overlap between resin and fatty acid esters in various liquid phases Pim = pimarate Neo = neoabietate = region where minor resin acid esters occur Liquid phase selectivity increases from top to bottom. Resin acid ester data for Versamid was obtained at 258" C; fatty acid ester data, at 214" C
phase loadings below 5 %, a situation which would be expected to result in support interaction. We question whether satisfactory chromatography of levopimarate can be achieved under these conditions. Tables I and I1 summarize our retention data for the methyl esters of the seven principal resin acids and a number of fatty acids. These results are illustrated schematically in Figure 1. Pimarate and neoabietate elute first and last, respectively, o n all liquid phases irrespective of temperature. Solute-solvent interaction appears to be the factor influencing the various relative separations to the largest extent. The data show two characteristics as liquid phase
Table 111. Relative Retention Values (18:0 = 1.0)for Unsaturated Fatty Acid Methyl Esters a t 200' c DEGS SE-30/EGiP Ester 18:2g~11cis,trunsa 2.03 1.06~ 18 :291 lltrurzs,trunsa 2.39 1.21c 18 :3 9 1 5 , 1 z ~ i ~ , ~ i ~ , ~ i ~ * 1.67 0.82 18 :39t12j16cis,cis,cis 2.04 0.90 20:34 1 1 , 14cis,cis,cisb 2.97 1.64 0 Dr. J. J. McBride, Jr., Arizona Chemical Co., Panama City, Fla., isolated from tall oil. * E. Elomaa, Oulu Oy, Oulu, Finland, isolated from tall oil. e Cf. ref. (IS).
1120
ANALYTICAL CHEMISTRY
Relative Retention Values (Pimarate = 1.00)for Methyl Esters of Minor Resin Acids DEGS SE-30/EGiP Source" 0.79 0.99 (i) As-Dihydroisopimarate 0.85 1.10 (a) Tetrahydropimarate (trans-anti-trans)(16, 17) 0.86 1.04 (a, b) A8-Dihydropimarate (18) 0.87 0.87 (b, g) A8f1s-Isopimarate(18) 0.90 1.11 (b, c) AS( 4)-Dihydropimarate(18) Tetrahydroabietate (88,9a,13a-H) (19, 20) 0.92 1.18 (e, f) Acid m.p. 200-1"; Me ester m.p. 74-6" 0.92 1.20 (a) Tetrahydropimarate (trans-an/i-cis)(16, 17) Tetrahydroisopimarate (trar7s-a/7?i-/ra~is) (16) 0.94 1.19 (a) 0.96 1.13 (b, c) A*(14)-Dihydroisopimarate (16) 0.97 0.95 (C) AS815-Pimarate(18) 1 .oo 1 .oo (1) Pimarate Tetrahydroabietate (88,90,138-H) (19, 20) 1.01 1.30 (e) Acid m.p. 181.5-182"; Me ester m.p. 45-46" AB-Dihydroabietate (13P-H) (19) 1.05 1.21 (e) Acid m.p. 172-4" [O]D +6" 1.06 1.19 (e) Acid m.p. 146-7" AI3-Dihydroabietate (88,9~-H)(19) A8(14)-Dihydroabietate(Sa, 138-H) (29, 20) 1.11 1.30 (e) Acid m.p. 199-200.5", [Q]D -27" (e) Acid m.p. 150°, [ a ] D +a"; Me ester m.p. 84-5.5" 1.28 A8(14)-Dihydroabietate(Sa, 13a-H) (19) 1.12 Tetrahydroabietate (8a,9a, 13a-H) (19, 20) 1.12 1.36 (m) Me ester m.p. 96-8" Sandaracopimarate 1.13 1.05 (c, g) As-Dihydroabietate (13a-H) (19, 21) 1.13 1.29 (e, f) Acid m.p. 185' [Q]D 120" Communate (traris)d 1.28 1.05 (4 A7-Dihydroisopimarate (18) 1.28 1.33 (b, c) A7-Dihydroabietate (9a,l3a-H) (19) 1.32 1.33 (e, k) Me ester m.p. 85-5.5' [ a ] D -16.3" 1.47 1.46 (e) Me ester m.p. 39.5-40°, [a]D -25.6' A7-Dihydroabietate (9a, 138-H) (19) 1.50 1.49 (e) Acid m.p. 220', [ a ] D -15"; Me ester m.p. 130" A I 3 (15)-Dihydroabietate(88,ga-H) (19) A 7 , 9 ( l')-Abietate (13a-H) (21) 1.60 1.36 (f) A*,12-Abietate(22) 1.79 1.49 (e) AB. 1 3 ( 15)-Abietate(19) 1.92 1.61 (e) Me ester m.p. 104.5-106" (+)-Daniellate (23) 2.61 1.30 (3 A6-Dehydro-dehydroabietate ( 2 4 2.78 1.36 (8) 4.1 2.28 (h) Pinifolate (25) 18 : O 0.241 0.663 a See Table I for relative retention values for . orincioal . resin acid esters. Numbering system as per ( I ) . Source of compound: (a) Dr. J. W. ApSimoil, Carleton University, Ottawa; (b) L. J. Gough, Borough Polytechnic, London; (c) Dr. 0. E. Edwards, National Research Council. Ottawa: (d) R. V. Lawrence. Naval Stores Laboratory, Olustee, Fla.; (e) Dr. A. W. Burgstahler, University of Kansas; (f) Dr. Werner Herz, Florida State University; (g) isolated from maleic anhydride-treated gum rosin; (h) Dr. 0. Theander, Swedish Wood Research Center, Stockholm; (i) prepared by diimine reduction of As, 15-isopimarate; (j)Dr. W. Dauben, University of California, Berkeley; (k) isolated by preparative GLC of a mixture obtained from Dr. R. Lombard, Universite de Strasbourg; (I) Dr. C. Bordenca, Glidden Co., Jacksonville, Fla. ; (m) T. F. Sanderson, Hercules, Inc., Wilmington, Del. d Identical with elliotinoate (26, 27). Table IV. Estera)b
+
selectivity increases: First, the resin acids as a group undergo greater separation between pimarate and neoabietate and generally have larger separation factors; second, the resin acids are retained longer with respect to methyl stearate. The magnitude of the relative retention ratios for Neo/Pim and Pimll8:O (sections A and B, Table I) parallels quite closely the values of al,?for 18: 1/18:0. The lack of essentially any solute-solvent interaction for (15) R. K. Beerthuis, G. Dijkstra, J. G. Keppler, and J. H. Recourt, Ann. N . Y. Acad. Sci., 72, 616 ( I 959). (16) J. W. ApSimon, P. V. Demarco, and J. Lemke, Can. J. Chem., 43, 2793 (1965). (17) W. Herz and R. N. Mirrington, J . Org. Cliem., 30, 3198 (1965). (18) 0. E. Edwards and R. Howe, Can. J. Cliem., 37, 760 (1961). (19) A. W. Burgstahler, J. N. Marx, and D. F. Zinkel, unpublished data. (20) J. W. Huffman, T. Kamiya, L. H. Wright, J. J. Schmid, and W. Herz, J . Org. Chem., 31,4128 (1966). (21) W. Herz and H. J. Wahlborg, J. Org. Cliem., 30,1881 (1965). (22) A. B. Burgstahler and L. R. Worden, J. Am. Cliem. SOC.,86, 96 (1964). (23) Identical with lambertianate. W. G. Dauben and V. F. German, Tetral7edroti, 22, 679 (1966). (24) G. Dupont, R. Dulou, G. Ourisson, and C. Thibault, Bull. SOC.Chim. France, 1955, p. 708. (25) G. Enzell and 0. Theander, Acta Chem. Scatid., 16, 607 (1962). (26) N. M. Joye, Jr., E. M. Roberts, R. V. Lawrence, L. J. Gough, M. D. Soffer, and 0. Korman, J. Org. Cliem., 30,429 (1965). (27) T. Norin, Acta Chem. Scarid., 19, 1020 (1965).
the resin acid esters on SE-30 was shown in measurements of relative retention a t several different column temperatures: a l , ?(dehydroabietate-abietate) = 1.19 a t 189" C and 1.17 a t 240" C. In comparison, cy1,? varied from 1.11 (189' C ) to 1.03 (240" C) for DEGS. Thus, separation of this pair with D E G S would be impossible at high temperatures if column efficiency were also low. This could account for the lack of dehydroabietate-abietate separation obtained by Valkanas and Iconomou (13) on DEGS. Ackman (28) has presented extensive data on separations of fatty acids with D E G S as influenced by column temperature. None of the liquid phases tested show much promise for effectively separating palustrate and levopimarate with an ordinary analytical column. PEG-2OM gave al,? = 1.03, but definitely not the separation or the elution order implied by Rudloff and Sato (IO); simple calculation (29) shows that approximately 18,000 theoretical plates would be required to separate these two isomers with a resolution of unity (4~7 separation of peak apexes). An a1,2of 1.05 was obtained with QF-1, but the value is in doubt because of considerable instability of levopimarate in this liquid phase. Extensive decomposition of methyl levopimarate was also observed in
(28) R. G. Ackman, J. Gas Chromatog., 1(6), 11 (1963). (29) H. Purnell, "Gas Chromatography," Chaps. 7 and 12, Wiley, New York. 1962. VOL. 39, NO. 10, AUGUST 1967
1121
.?OI
Table V. Relative Retention Values (Pim = 1.00) of Artifact Peaks Observed in the Chromatograms of Methyl Levopimarate at 200" C Peak DEGS SE-30/EGiP N0.a h i m mim Peak identity 1 0.64 ... Unknown 2 0.71 ... Unknown 0.78 0.76 Unknown 0.85 0.82 Unknown 0.97 0.97 Unknown 1.33 1.23 Levopimarate 2.08 1.55 Abietate 2.29 1.36 Dehydroabietate 2.43 1.83 Neoabietate .Other minor peaks have sometimes been seen on DEGS chromatograms, m i r n = 0.49, 0.55, and 1.11.
1
I
1
I
I
1
I
--
IO
5-
-
3
p g2
+I.O
32-
$ '1 q
FATTY ACIDS
-
$ 1 0.5-
-
1
0.2-
0.3
0.1
a highly loaded (20%) QF-1 o n Anakrom ABS column, even
I
'
O,'
I
I
"
I
I
1
I
I
" 26 28 32 CARBON NUMBER8 Nln-ALKANES, n - C # I
I
I
1
34
36
38
after treatment of the column in situ with hexamethyldisilazane. I n a recent publication, Chang and Pelletier (14) reported a Figure 2. Relationship between relative retention of n-alkanes separation factor of 1.09 for these two esters using 4 % Q F - ~ (solid line) and methyl esters of fatty and resin acids in DEGS at 200" c a t a temperature of 193" C. Our attempts to measure r18:o for levopimarate with three different low-loaded QF-1 and the quantitative interpretation of gas chromatograms columns were completely unsuccessful because of excessive has been reviewed (32). In this work, the response Of the loss of the ester by isomerization. Two columns were filled thermal conductivity detector relative to the internal standard with of % QF-l on Anakrom ABS, 70,80 mesh, is defined as a "correction factor," different samples of liquid phase being used for each; the third contained a packing of 1 % QF-1 on 30-mesh Haloport F, a polytetrafluoroethylene support. The chromatography of levopimarate is discussed later in more detail. where is the weight ratio of compound to internal stanThe use of two liquid phases having extremes in selectivity dard and ra(s)is the ratio of compound peak area to internal -e.g., SE-30/EGiP and DEGS, permits an analytically usestandard peak area. The relative molar response factor (33), ful separation of a mixture containing both solute systems, reciprocally related to f&), is the form generally used in the as shown in Figure 1. On SE-30, the resin acids are comliterature for reporting the response of this type of detector, pressed into that region of the r18:o scale, equivalent roughly but is less useful for actual analysis. to the fatty acids 19:O through 21 : O permitting those above Detector response data are presented in Tables VI and VI1 21 : O to be detected. At the opposite extreme, D E G S for the methyl esters of resin and fatty acids o n DEGS. The opens the retention scale for fatty acids below 23 : 0. D E G S response factors were found to be linear with sample size provides a n optimum separation of the important unsaturated except for levopimarate and 24:O. With the exception of fatty acids (3) as well as of the resin acids. Our own results levopimarate, the isomeric resin acid esters listed give very for several unsaturated acids, including some found in tall nearly equal responses as expected theoretically (33, 34). oil (30,31), are given in Table 111. The correction factor is the average value ftc(l8:o) = 1.25 Limitations t o the above scheme arise principally from the 0.03 (i2.6%). Hudy (8) noted that correction factors presence of hydrogenated resin acids, a number of other less ranged from 1.3 to 1.6 (16:O = 1.00). More recently (12), commonly occurring resin acids, and the degradation products several of the principal resin acids have been found to have from levopimarate. These three groups of compounds the same specific response with a thermal conductivity de(Tables IV and V) overlap the relative retention values of the tector. I n the latter paper, however, the response of levopifatty acids and the principal resin acids as indicated on Figure marate was 0.70 that of the other six esters [the equivalent of a 1 by the crosshatched rectangles. Actually, the situation correction factor of fw(18:o) = 1.81 because of isomerization should not be as restrictive as it appears, because not all these on the column. We also experienced difficulty with levopimaterials will occur in a mixture a t one time. marate, as discussed later. Figure 2 shows in detail the relationship between the two Correction factors for the saturated fatty acid esters, when solute systems studied, superimposed on a log plot for the plotted (Figure 3) as a function of acid carbon number (in n-alkanes, for DEGS as the liquid phase a t 200" C. The effect, molecular weight), showed an essentially linear beequation for the hydrocarbon line, obtained by a least squares havior which agrees with previous observations (34-36), 0.1150N, treatment of the data, is log rig = -3.0678 where N is the number of carbon atoms in the molecule. (32) L. V. Andreev, M. I. Afanas'ev, 0.G. Chabrova, and M. S. Quantitative Correction Factors. The response of the Vigdergauz, Rirssiun Clzem. Rec.., 34, 378 (1965). thermal conductivity detector (katharometer) has been the (33) A. E. Messner, D. M. Rosie, and P. A. Argabright, ANAL. subject of many experimental and theoretical investigations, CHEM.,31, 230 (1959). (34) L. A. Horrocks. D. G. Cornwell. and J. B. Brown, J. Lipid Res., 2,92 (1961). (30) T. Lehtinen, E. Elomaa, and J. Alhojarvi, Suomerz Kem(35) J. V. Killheffer, Jr., and E. Jungermann, J. Am. Oil Chemists' istilehti, 37, 27 (1964) and references cited therein. Soc., 37,456 (1960). (36) W. A. Pons, Jr., and V. L. Frampton, Zbid.,42,786 (1965). (31) J. J. McBride, Jr., private communication, 1965.
*
+
'
1 122
ANALYTICAL CHEMISTRY
,
1
1
I
I
I
06! I2 0
I
n
I /4
I
0
16 0
i8
0
CARBON
l
1
I
I
20 0
22 0
24 0
Table VI.
Estera Dihydropimarate Pimarate Tetrahydroabietate A*(14)-Dihydroabietate AS@’-Dihydroabietate Sandaracopimarate Levopimarate Palustrate Isopimarate Abietate Dehydroabietate Neoabietate Average Standard deviation, u
260
NUMBER
Figure 3. Quantitative correction factors for saturated fatty acid methyl esters (For the solid and dashed lines, the values of the ordinate are for r and ( r 3, respectively rll = mol wt (X:O)/mol wt (18:O) and is consistent with the general behavior of homologous compounds (33). The correction factor is very nearly equal t o the ratio of the appropriate molecular weights (the solid line, for which r.,f is the ordinate). The dashed line represents (r.vf)2’3 the behavior reported by Horrocks et al. ( 3 4 ) . The data in Table VI1 were fitted to a n optimum straight line equation by a least squares treatment (the value for 24 : 0 was omitted). This gave the expression: .fL’
=
0.1099
(37) J. M. Miller and A. E. Lawson, Jr., ANAL.CHEM., 37, 1348 (1965).
fW(lS :o)
1.28 1.26 1.24 1.27 1.30 1.28 1.4b
1.23 1.23 1.26 1.21 1.26 1.25. f .03
Binary mixtures of ester +18:0 were used. See text. c Excludes levopimarate.
0
b
Table VII. Quantitative Thermal Conductivity Detector Correction Factors for Methyl Esters of Fatty Acids on DEGS Experimental Corrected fw(18 :a) Ua fU(18 :o+ rMC Ester
20:o 22:o 24:O
A. SATURATED ACIDS 0.80 h0.03 0.80 0.94 10.03 0.90 0.96 10.02 0.95 1.00 1 .OO 1.12 10.01 1.10 1.23 fO,03 1.20 1.41 10.09 ...
18:l 18:2 18:3
B. UNSATURATED ACIDS~ 1.04 f0.02 1.02 ... 1.09 It0.02
14:O
16:O 17:O 18 :o
+ 0.509N
from which f L c l = 1,026 for 18 :0. If all fw’ values are divided by this figure, they become essentially equal to the ratio of molecular weights, as shown in Table VII. The value of this correlation lies in the fact that detector response for this family of solutes may be established with a minimum number of experimental measurements. Furthermore, any single member is equally suitable as an internal standard within the standard deviations shown. Purnell (29) and, more recently, Miller and Lawson (37) have discussed some conditions that lead to anomalous response of the katharometer detector. F o r the detector conditions used in this work, we observed anomalous effects with both n-alkanes and saturated fatty acid esters of molecular size, N 2 24. To our knowledge, this has not been previously reported. Thus, 22:O has a normal response factor as shown in Table VII, while 24:O is definitely anomalous. Although operation of the detector at 22.5” C and 210 mA results in normal response for 24:0, stable operating conditions are difficult to maintain. However, detector temperature or filament current below our normal operating conditions was observed to result in anomalous behavior for 22 : 0. The resin acid esters d o not exhibit abnormal behavior in that they show essentially no dependence of &$.(s) upon variation in detector temperature from 205-25’ and filament current from 190-210 mA. The correction factors for the resin acid esters & ( I ~ . ~ ) = 1.31 excluding levopimarate] were slightly higher o n a n SE30/EGiP column than on DEGS. This result is believed t o be due t o a slight preferential transesterification of the fatty acid esters in DEGS. In a study of the efficiency of the diazomethane methylation
Quantitative Thermal Conductivity Detector Correction Factors for Methyl Esters of Resin Acids on DEGS
a
Standard deviation
bfW(18:o.(S)
corrected
=
= 6.
0.1109
0.812 0.906 0.953
1 .oo
1.094 1.188 1.282
+ 0 . 0 5 0 9 (see ~ text). 1.026
procedure, it was found that the fatty and resin acids were quantitatively methylated under our conditions (2). When a large excess of freshly prepared reagent was used, no interfering side products were formed nor was there a loss of such resin acids as the highly reactive diene, levopimaric acid. Methyl Levopimarate. The stability of this ester is particularly important t o the quantitative analysis of fatty and resin acid ester mixtures. Subsequent work has corroborated the earlier observation ( 2 ) that it was not stable in solution, but also has shown that its gas chromatography is more complex than was presumed. The data now demonstrate that, depending upon the reactivity of the packing, on-column reactions can also occur. These can take place rapidly enough t o result in the presence of discrete, extraneous peaks in the chromatograms as well as of a zone (recorded as a low-level plateau) of mixed degradation products. One reaction appears t o be the same as that occurring with this ester in solution, and can be observed even on a new, supposedly inactive D E G S packing (15-20z loading). Second, on a n “aged” VOL. 39, NO. 10, AUGUST 1967
1123
A. NEW DEGS COLUMN-FRESH LEVOPIMARAX
Q 0. NEW DEGS COLUMN-OLD LEVOPIMARATE
c. OLD
OEGS COLUMN-FRESH LEVOPIMARATE
even o n a fresh column with fresh levopimarate; it was always accompanied by dehydroabietate.) Even for a chromatogram such as that of Figure 4A, some decomposition occurred as was evident in a higher than average (about 10%) correction factor and a sample size effect, wherein the factor decreased with an increase in the size of the ester peak. This is consistent with the explanation that saturation of active sites on the support would result in a smaller relative loss of the labile ester as the sample size increased. Thus, duplicate 1-p1 injections of ester solution each gave f v ( s ) = 1.39, but a value of 1.34, resulted for a 5-pl sample. Therefore, attainment of the theoretical correction factor for levopimarate is probably impossible even on DEGS, the packing least reactive of those studied. Chromatography of levopimarate on nonpolar liquid phases provides direct evidence that incomplete deactivation of the solid support was the most important cause of the observed degradation of this ester by double bond isomerization. Loss of levopimarate on SE-30 was significantly reduced when the support was precoated with 1 EGiP (Figure 4 0 ) . However, this 1 of EGiP apparently does not deactivate the support as efficiently as 1 5 2 0 % D E G S itself. Freshly prepared methyl palustrate appears to chromatograph without noticeable decomposition and, indeed, its correction factor is as expected. However, on standing under nitrogen, a solution instability leading to dehydroabietate has been observed; it is apparently similar t o that with levopimarate but proceeds a t a much slower rate. Internal Standard. The choice of a quantitative standard for a complex mixture of both resin and fatty acid ester solute systems is considerably limited by the lack of “open” places of sufficient width on the retention scale in which t o place a completely resolved reference peak. Attempts to use one of the higher n-alkanes were frustrated by failure t o obtain quantitative response with the thermal conductivity detector. From the data presented herein and the infrequent occurrence in nature, methyl margarate (17 : 0) is the best compromise as the internal standard for mixtures of fatty and resin acids with our dual column system of DEGS and SE-30/EGiP packings. Although the data reported are with respect to methyl stearate and methyl pimarate, it can be readily converted to the 17:O basis by applying the data from Tables I1 and VII. F o r analysis of resin acids alone, the use of methyl arachidate (20:O) as an internal standard would be preferable.
z
0. SE-3O/EGiP -FRESH L EVOPIMARATE
I
I
IO
30
I 20 RETEIITION TIME ~ I M I h ’ U T f S I
IO
0
Figure 4. Chromatograms of methyl levopimarate (18:O added) Peak numbers identified in Table V (well-used) DEGS column, it is evident that double bond isomerization (38) of methyl levopimarate also occurs. Table V lists all peaks which have been observed upon chromatographing this ester with a number of different D E G S packings. Ester solutions ranged in age from freshly prepared (< 1 hour old) to those which had been refrigerated under nitrogen for almost ll/s years. The numbering of artifact peaks has been changed in the present report so that those previously ( 2 ) designated l ’ , 1, 2 , 3, correspond, respectively, to 2 , 3,4, 5. The relationship of column packing activity to the levopimarate chromatogram is shown in the following examples. Figure 4, A and C, contrasts a fresh with a well used D E G S column using a freshly prepared methyl levopimarate solution. Approximately one third of the original weight of liquid phase had been lost in 4C. Figure 4, A and B, contrasts a fresh and a 11/2-year-old levopimarate solution o n a fresh D E G S column. (Peak 1 was observed only when the levopimarate peak was collected and reinjected and was seen (38) R. V. Lawrence, Tappi,45,654 (1962).
1 124
ANALYTICAL CHEMISTRY
z
ACKNO WLEDGhIENT
The Forest Products Laboratory is maintained in cooperation with the University of Wisconsin. This work was supported in part by Pulp Chemicals Association, New York, N. Y. Mention of trade or company names is for purposes of identification only and does not imply endorsement by the U. S. Department of Agriculture.
RECEIVED for review February 13, 1967. Accepted June 9, 1967. Presented in part, Division of Cellulose, Wood, and Fiber Chemistry, 153rd meeting, ACS, Miami Beach, Fla., April 1967.